Aiaa Mdo Tc 1998 White Paper on Mdo

White Paper on Industrial Experience with MDO


The new white Paper on industrial experience with MDO consists of several invited papers and a summary report from the 1998 Symposium on Multidisciplinary Analysis and Optimization.

Invited Papers:

The Role of MDO within Aerospace Design and Progress towards an MDO Capability, Peter Bartholomew (Defence Evaluation and Research Agency, UK).

Issues in Industrial Multidisciplinary Optimization, J. Bennett, P. Fenyes, W. Haering, M. Neal (GM).

MDO Technology Needs in Aeroelastic Structural Design, H.G. Hönlinger (German Aerospace Center) and J. Krammer and M. Stettner (Daimler-Benz Aerospace).

Multidiscipline Design as Applied to Space, Charles F. Lillie, Michael J. Wehner and Tom Fitzgerald (TRW).

http://endo.sandia.gov/AIAA_MDOTC/sponsored/mao98_whitepaper.html (1 of 3)12/29/2006 12:28:17 PM


Multidisciplinary Design Practices from the F-16 Agile Falcon, Michael H. Love (Lockheed Martin).

The F-22 Structural/Aeroelastic Design Process with MDO Examples, Nick Radovcich and David Layton (Lockheed Martin).

A Collaborative Optimization Environment for Turbine Engine Development, Peter J. Röhl, Beichang He, Peter M. Finnigan (GE CR&D).

Boeing Rotorcraft Experience with Rotor Design and Optimization, Frank Tarzanin, Darrell K. Young (Boeing).

The Challenge and Promise of Blended-Wing-Body Optimization, Sean Wakayama (Boeing) and Ilan Kroo (Stanford).



A Description of the F/A-18E/F Design and Design Process, James A. Young, Ronald D. Anderson, and Rudolph N. Yurkovich (Boeing, St. Louis).

Summary paper:

A Summary of Industry MDO Applications and Needs, Joseph P. Giesing (Boeing) and Jean-Francois M. Barthelemy (NASA Langley).

● Presentation.

● Paper.

Back to MDO TC Home Page

Last Updated: August 10, 1999

Michael Eldred, [email protected]


http://endo.sandia.gov/9234/Personnel/mseldre

mailto:[email protected]

AIAA-98-4705

1

American Institute of Aeronautics and Astronautics

THE ROLE OF MDO WITHIN AEROSPACE DESIGN ANDPROGRESS TOWARDS AN MDO CAPABILITY

Peter Bartholomew †

Defence Evaluation and Research Agency,Farnborough, Hampshire GU14 0LX

United Kingdom

ABSTRACT

This paper reviews recent progress made in MDOwithin the European aerospace industry through theactivities of a sequence of international collaborativepartnerships of increasing complexity. Firstly, adefinition of MDO is provided and its function as a keytool in the context of concurrent engineering isdiscussed. Issues addressed include the limited supportgiven by many MDO tools to detail stressing,validation of aeroelastic optimisation, the role ofproduct models, the definition and execution of MDOprocess under user control and trade-off studies forrequirement capture. The need for the adoption ofstandards in the definition of the product model and thelikely impact of the CALS philosophy of ‘create dataonce and use many times’ are highlighted.

INTRODUCTION

Multidisciplinary design optimisation enables theefficiency of designs to be optimised and supportstrade-off studies between the design objectives ofdiverse disciplines. The MDO process is intended foruse within the context of modern engineering designenvironment, which is characterised by the commercialimperative to reduce time cycles and costs. Thesecommercial pressures, together with the immensevolume of design, manufacturing and maintenance datainherent to complex modern equipment, demand aheavily computerised environment.

Current practice, as exemplified by Concurrent Engi-neering (CE), is to move the design of complexequipment away from a process involving a sequenceof specialist departments and to emphasise its multi-disciplinary nature through the use of integrated prod-uct teams. Both the structural integrity of engineeringproducts and demonstration of the performance ofproposed designs are increasingly reliant on the use of

computer models created during the design process.Although the software tools existing within individualdisciplines may be reasonably mature, the challenge isnow to provide the tools necessary to support such anintegrated approach.

The scope of multidisciplinary design optimisation(MDO) is limited to the design of products based onthe simulation of physical objects in their environment.The use of multiple simulations is a key concept ofMDO. This may involve diverse tools such as: fluidflow solvers (to determine local and overall externalforces); structural analysis and detail stressing (todetermine structural deformations and internalstresses); electromagnetic analysis (to determine radarsignatures from local and overall returns from incidentbeams); cost modelling and tools for design for reli-ability. The physics modelling may be mathematical orexperimental but the simulation of ‘human interaction’effects, for example through the use of flightsimulators, is excluded.

At a general level, when considering the overall mis-sion performance of an aircraft, tools exist to aid theconceptual design of both military and civil aircraft andare used during the early stages of the project.Although these adopt a fully multidisciplinaryapproach, only the simplest, Level 1, empirical modelsare employed to approximate the physics whichinfluences the overall design. Currently most MDOapplications, for use in the preliminary design phases ofa project, are based on major simplifications inmathematical modelling at level 2, such as beamstructural models or panel methods for aerodynamics.

The objective is now to achieve the same degree ofintegration with level 3, state-of-the-art analyses. Thelimiting factor in the use of such best, proven models isthe capacity of current computation technology.Analyses using computational fluid dynamics, compu-tational electro-mechanics, or detailed finite elementmodels are separately capable of pressing computerresources to the limit, and this is compoundedby the introduction of sensitivity calculations and

† DERA Fellow, Aero/Structures Dept, AIAA Member© British Crown Copyright (1999), DERA. Published withpermission of the Controller of Her Majecsty’s Stationary Office

2


optimisation. It is evident from conferences devoted toMDO1-3 that the move to higher fidelity analysis tools,which have formerly been the preserve of specialistdepartments, is general.

The software framework one may require to controlsuch a process, user interface issues and the form ofproduct data used to support design, manufacture andoperation are discussed in this paper in the context of aseries of MDO collaborative activities within theEuropean Aerospace industry. While the conceptualdesign tools referenced above tend to be close-coupled,it is of interest that the tools used in the variouscollaborations have all been loosely coupled.

STRUCTURAL OPTIMISATIONGARTEUR SM(AG13)

Detail design

One of the problems in introducing MDO is thecomplexity of the design process itself. Even within thesingle discipline of structures, finite element programswill be supplemented by a range of data sheets, detailstressing programs and manual methods, all used toestablish structural integrity. It is essential for thecredibility of an MDO process that it should be able toaccommodate the detailed design processes normallyused within the company.

The GARTEUR Structures and Materials panel hassupported collaborative research activities on StructuralOptimisation from 1990 onwards. In particular theGARTEUR Action Group SM(AG13) addressed theuse of panel design codes within the overall strengthand stiffness design process for aircraft wings. Here,even within the context of a single discipline, theMDO-related issue of multilevel design arises, since theFE-based codes, commonly used to improve overallwing efficiency, may be supplemented by codes fordetailed panel stability design and assessment, appliedon a panel by panel basis.

Codes for the buckling design of composite panels

were available from DASA Airbus, NLR and U.Cardiffand others were purpose-written as required. Structuraloptimisation codes were available from BAe, DASA,SAAB, Dornier, Aerospatiale, NLR and DERA. Themajor codes were presented by their originators andcompared, and multilevel methods for the integration ofpanel and overall structural optimisation wereinvestigated.

Rear spar

Front spar

Middle sparBottom skin

(Top skin omitted)

'Lumped' stringers

Fig. 2 : Simple wing model

The methods developed were evaluated using civil andmilitary aircraft wings of differing complexity asbenchmark problems4, the simplest being that shown infigure 2. A larger problem of a commuter-aircraftwing, from DASA Airbus, is regarded as an industrial-scale problem and the development of strategies forexploiting composite materials in compression structurewere regarded as important.

Overall it was found to be possible to include thedetailed design of composite stiffened wing panelsagainst buckling within the overall strength and stiff-ness design process for the wing using relatively simplestrategies, although it is acknowledged that interactioneffects between adjacent panels are not addressed bythese methods.

MULTIDISCIPLINARY DESIGNOPTIMISATION OF AIRCRAFT WINGS

GARTEUR SM(AG21)

Aeroelastic Optimisation

GARTEUR Action Group SM(AG21) on multi-disciplinary wing design concentrates upon theintegration of strength and aeroelastic aspects of thedesign of high aspect ratio wings typical of modernregional transport aircraft, as illustrated in figure 3.The DERA contribution is based on the use of the in-house structural optimisation code, STARS5 which, likeseveral others, embodies aeroelastics as a tightly-coupled functionality. Both the aeroelastic predictionsand design strategies to come out of the optimisation

l b

strg-plyt

skin-plyt

h

Fig. 1 : Dimensions of compression panel

3


will be compared with those of the other partnerswithin the group.

Fig. 3 GARTEUR SM(AG21) model

While several European companies have long had thecapability of combining aeroelastic design with basicstrength requirements within the context of what areprincipally structural design codes, the progress of afollow-on GARTEUR Action Group is discussed inproviding a forum for the validation and comparison ofthe capabilities of various companies. Such com-parison is felt to be important since, from othercollaborative projects, it has been found that signifi-cantly different ‘solutions’ can found by differentgroups

MDO OF AEROSPACE VEHICLESEU IMT PROJECT BE95-2056

Project outline

The MDO project represented a first step into multi-disciplinary analysis and design optimisation for manyof the partners. The application selected to demon-strate new capabilities developed during the projectwas based on the A3xx concept currently under devel-opment by the Airbus partners. A whole aircraft modelwas provided for aeroelastic and controls studies, butthe design activity was focused upon the wing.

The project was subdivided into a series of tasks shownin figure 5. All partners participated in the definitiontasks 1-3 and from then on separate groups wereresponsible for the investigations conducted by tasks 4-7. The project was supported by the softwareinfrastructure group working in task 8 in whichparticipating partners were drawn from each of thesetask groups. The final stage of the activity was to drawtogether the lessons learnt from the project asrecommendations in task 9.

Aerodynamic and Structural design

The objective of the work was to develop and demon-strate a capability for the aerodynamic and structuraldesign of a wing which would minimise the directoperating cost (DOC) of the A3xx concept aircraft.The form chosen for the DOC was simply a linearcombination of mass and drag relative to that of thereference aircraft viz

( )∆ ∆ ∆DOC = . + .10 10W Decon

where W is the mass in tonnes andD the drag in counts. The majority ofthe optimisation work performed wasbased on the use of a few gross wingdesign parameters, namely: area,aspect ratio, rear spar location, sweep,crank thickness and tip twist.

The initial work conducted by thepartners in Task 5 was simply tooptimise the wing with respect to thetwo surface shape parameters, crankthickness and tip twist, and tocompare results for aerodynamic dragand structural mass corresponding tothis baseline case8.

The optimisation results in figure 7

Task 1: Project Management

Task 2: Simplified MDO process

Task 3: Primary Sensitivity Study

Task 4:PlanformOptimisation

Task 5:Surface-shapeOptimisation

Task 6:StructuralOptimisation

Task 7:ControlOptimisation

Task 8: Prototype MDO framework

Task 9: Recommendations

Fig. 5 : Task structure for EU MDO project

4


show a considerable variation between partners. Inparticular, significant differences are found dependingupon the treatment of fuselage effects and the design ofa wing in isolation also changed significantly when thewing was treated as part of a trimmed aircraft model.

Such differences are not a simple matter of right andwrong, but rather depend upon an understanding of theimportant characteristics of the flow and of thelimitations of the various numerical approaches. Atthis stage in the development of MDO there is little orno interest in close-coupled black-box methods. Astrong need was perceived to use familiar legacy codeswithin a loose-coupled modular framework that enabledthe output from every process to be evaluated beforeproceeding.

While differences in the aerodynamics provide themain contribution to the variation of results, similardifficulties are also encountered resolving differencesof design arising from the structural optimisation,despite this being regarded as a relatively mature tech-nology. A reasonable consensus was achieved for thefinite element results, but the optimisation, particularlythat of the commercial codes, tended to be over-sensitive to details of the method selected and parame-ter settings and did not necessarily converge tooptimum solutions.

The DERA-specific work introduced multiple flightconditions into the optimisation. Aerodynamic analysis

Fig. 9 : Spanwise distribution of lift for heavycruise

sis of the wing is performed at light, economic andheavy cruise and the drag calculated is combined withthe mass given by structural optimisation, to give anestimate of direct operating cost in the form

( )∆ ∆ ∆ ∆ ∆DOC = . + + +10 0 4 0 2 0 4W D D Dlight econ heavy. . .

Some of the trends were similar in the single andmultiple condition optimisation. In particular it wasnoticeable that the lift moves slightly inboard as infigure 9. This changes the trim of the aircraft, reduc-ing the downforce required on the tailplane, and hencedecreases the total lift of the wing. This results in areduction in the lift-induced drag for all flightconditions.

At both economic and heavycruise conditions there is aweakening of the shockwaves which also tend tomove inboard. The reducedcontribution to the total dragfrom the shock wave drag isparticularly important forheavy cruise. Optimising thewing for the economic cruisecondition, in the hope that thedesign will also provesatisfactory at light and heavycruise, gives poor results inheavy cruise condition. Byoptimising the wing formultiple cruise conditions,the drag at heavy cruise isimproved without losing theimprovement at the otherflight conditions.

-5

-4

-3

-2

-1

0

1

2

0.06 0.07 0.08 0.09 0.1 0.11

Crank thickness

Tip

twis

t

Reference

Wing+fuse

Wing only

Wing+c/s

Wing / no trim

Wg+c/s / no trim

Fig. 7 : Optimum designs calculated for baseline problem

5


This task illustrates the need for flexibility within anMDO process, to allow the user to configure theoptimisation process to accommodate multiple assess-ment tools, specific to each problem.

Product models and TDMB

The complexity of the data flow which links thedisciplines of aerodynamics and structures, is illustratedin figure 10. This starts with a requirements system,which is assumed to be external to the MDO system, inwhich some freedom is assumed to exist to fine-tunethe relative importance of various aspects ofperformance. An outline concept is then developed asa parameterised product model. This is followed byvarious assessments, here shown as aerodynamics andstructures, with the possibility of making detailed shapeand thickness changes for a given configuration.

Referring to figure 10, it is clear that large amounts ofdata, which may well be stored in separate databases,must be communicated between the component parts ofthe MDO system. The key issue for data transfer is thesetting of common standards for the interpretation ofinformation across disciplines. For MDO, thestandards must cover all aspects of product geometrydefinition and design requirements, together withspecific discipline-based data that reflects theconstraints upon the design.

During the early meetings of the MDO project, a seriesof key activities were decided upon which defined thenature of the project. One was to adopt the BAeprogram TDMB7 (Technical Data Modeller &Browser) as the repository for the product model.TDMB provides a text editor user interface whichsupports an definition of data objects and then expandsto store instance data capable of representing severalvariants of the product together with performance dataderived from aerodynamic and structural analysis.

A fully parameterised representation of the aircraftconfiguration was developed, with tools to generateaerodynamic data, finite element models andaeroelastic models used for performance assessment.This data-representation serves the project by providingpartners with a common product model upon whichdesign studies were based. The data models defined inTDMB will be exportable to the STEP/EXPRESS datadefinition language to enable future migration to othersystems which conform to evolving standards forproduct models. The wider use of data which conformswith the STEP standards6 is an important element ofachieving the CALS objective of ‘creating data onceand using many times’ through the product life cycle.

MDO process

A major factor which will influence the overall successof any MDO implementation is the approach adopted tothe co-ordination and scheduling of the diverse range ofactivities necessary to complete a full design cycle.This aspect of MDO must be adequately defined in theearly stages of the development process in order todraw together the different disciplines and allowconcepts to be explored.

A framework specification document was written by theTask 8 partners and various software tools wereprovided. These include tools for: software versionmanagement, data definition, database technology,process definition, process execution on distributednetworks, data visualisation and optimisation.

Several alternate frameworks were employed andevaluated against the user and system requirementspreviously developed. The frameworks assessedincluded commercial MDO frameworks and toolsets, aprocess-driven Workflow Management tool andNetwork middleware.

The frameworks tended to operate with a pre-definedsequence of operations and failed to provide the userwith sufficient flexibility to reconfigure the processduring the early exploratory phases of a design study.The interactive definition of a complex process is aprime requirement of any optimisation framework.

The strength of work flow management tool is thetraceability and control it offers, whereby onlyapproved users may initiate processes and that onlyprovided the input data has not been invalidated bychanges by an upstream process. Network middlewaresystems enabled the computer resources of the networkof machines to be utilised with the facility that one mayexpect of a single machine, but tended to require user-intervention and were weak at running chainedprocesses.

As may be expected the purpose-written MDO frame-works provided the most flexible integration supportbut did not necessarily distinguish the process supportaspects (including the registration of tools, the defini-tion of process chains and their execution) from datamanagement (product models and requirements) orfrom embedded tools (for the visualisation of variouscategories of data or optimisation functionality).Further development is needed if the frameworks are tooperate in a standards driven environment accessingdata from corporate data bases.

6


UserRequirements

System

00 Requirements data

07 Pareto Frontier

06 New parameters

07b Structural mass

06b New thicknesses

02 Product description

03bFinite Element data

06 New parameters

04b Structural response

05b Struct performance

00 New requirements

1.Caseset-up

2. Computer-

aided design

3b.Finite elementpre-processor

4a.Flow solverEuler or N-S

3a.CFD gridgenerator

7.Capture design

decisions

5.Multi-disciplineassesssment

6b.Structural

Optimisation

5b.Structural

assessment

4b.Finite element

analysis

6a.Airfoil sectionoptimisation

5a.Flow

assessment

6.Multi-discipline

optimisation

03a Multiblock FD grid

06a New detail shape

04a Aero pressures

05b Lift & drag

01 Case data

06a New detail shape

04b Structural response

06b New thicknesses

Problem set up

Aerodynamics Structures

Revision ofrequirements

01 Case data

Multi-disciplinetrade-offs

05 Mass, lift & drag

Fig. 10 : Data flow showing multidisciplinary tools

7


The role of the optimiser

The role of the optimiser has also been the subject ofslight variation within the various partner frameworks.At the simplest, the optimiser calls for functionevaluations, possibly including gradients, at a sequenceof design points and, in effect, controls the process. Asthe function evaluations call for increasingly time-consuming analyses with complex data interactionsand, possibly, requiring user-intervention, this becomesa less attractive option.

An alternative approach is still to start the design cyclewith the optimiser initiating a design change, but toreturn control to the framework for the performanceassessment phase. The optimiser must then be capableof being restarted once the performance assessment iscomplete. In software terms, the optimiser may thenappear as just another MDO process, to be called asrequired, but its controlling role within the process ofdesign should still be recognised.

FRONTIER / ESPRIT PROJECT 20082

Project outline

Finally the contribution of the EU project Frontier9

towards the capture of requirements is described. It isalmost inevitable that any MDO problem, as initiallyformulated, will not automatically lead to the requiredproduct, since impact of constraints and the balance ofconflicting requirements will not be fully understood atthe outset. In this project, a Pareto-frontier approach isused together with multi-criterion decision making(MCDM) software to capture customer preferences.Clearly, if cost were a criterion, this leads to acost/performance assessment which is a key input toany requirement capture process.

Although Frontier is a relatively small project, it is ofthe widest scope in that it considers design againstmultiple objectives. The project partners consist ofuniversities who are, in the main, acting as suppliers ofnew technology and industrial partners who areproviding user trials relevant to their industry sector.

Fig. 12: DERA ‘user trial’ model

8


Fig. 13 : PARETO boundary

Requirement capture for military aircraft

The user trial to be conducted by DERA in partnershipwith BAe is based on the design of a military wing andseeks to achieve an acceptable compromise betweenaircraft range and turn performance. Figure 12 showsthe pressure distribution calculated from CFD on theleft with the finite element mesh and loading derivedfrom it on the right. In this instance the aerodynamicmodel is taken as the master model, but in the longerterm it would be expected that both the aerodynamicsand structures models would be derived from acommon product model.

The approach taken is a multilevel Pareto-optimisationin which the wing thicknesses (wing-box depth) atvarious stations are used as top-level variables linkingthe structures and aerodynamic disciplines. Thestructural optimisation simply the sizes the compositecovers and sub-structure for each geometry, while theaerodynamic optimisation modifies the airfoil shape tomaximise a weighted sum of lift to drag ratioscorresponding to a supersonic turn condition andtransonic cruise.

The supersonic turn rate and transonic range shown infigure 13 are then calculated from the drag, mass andfuel volumes. Each curve corresponds to a given

spanwise thickness distribution but with the aero-dynamic shape optimised to give differing levels oftransonic to supersonic performance. In general thethicker wings give greater range due to their increasedfuel capacity, but ultimately (case 9) higher drag willreduce the range.

The Pareto frontier itself, indicated in grey in figure 13,bounds the region in which it is possible to designproducts to meet the conflicting requirements. The bestproducts have performance characteristics which lieclose to the ‘top-right’ part of the boundary. From hereit is only possible to improve one characteristic at theexpense of the other.

The use of genetic algorithms is to be assessed as amethod of achieving convergence to the boundary ofthe region. Typically such direct search methodsrequire many function evaluations, each one of whichwill call on a full structural optimisation for mass aswell as an aerodynamic minimisation of drag for twoflight conditions.

The fact that these tasks are computationally intensivemakes the activity appropriate for high-performancecomputing in the longer term, but to reduce thecomputing costs during this project, response surfaceshave been calculated for the wing mass and drag. ThePareto frontier may then be calculated on the basis of

9


the cheaper response surface information rather thanfrom further calls to the underlying design software.This will enable sufficient computing resources to bedevoted to the assessment of genetic algorithms withinthe Pareto frontier approach and to evaluating theMCDM software tool for deducing the weightingsattached to the various design objectives from customerpreferences. This aspect of the Frontier project is ofparticular interest as it extends the scope of MDO sothat it assists with identification of the designrequirements that the product should meet.

CONCLUSIONS

A number of developments relevant to the practical useof MDO have been identified By reference to asequence of collaborative research activities withinEuropean aerospace industry. A definition of MDO asincorporating state-of-the-art analysis tools is providedand its function as a key tool in the context of concur-rent engineering is discussed.

It is believed essential for the credibility of an MDOprocess that it should be able to accommodate thedetailed design processes normally used by engineerswithin the company to assess and validate their prod-ucts. Scepticism as to the results from each step of anMDO process is vital and the comparative studies con-ducted by partnership have often produced widelyvarying results. The validation of methods such aswithin the GARTEUR activity on aeroelastic design isseen as an essential activity.

The central the role of the product model is highlightedand the desirability of using STEP to standardise theform in which product data is shared and exchangedamongst processes is to be emphasised.

A good framework for MDO which provides a flexibleuser interface for the definition, execution andmonitoring of MDO processes is essential and furtherdevelopment of clear architectures for such software isstill required. While conceptual design tools are oftenclose-coupled, loosely coupled systems appear to bemore appropriate to MDO in that they assist the verifi-cation of results by specialists. Some loss in processefficiency or even the generation of sub-optimaldesigns is acceptable provided the design process isunderstood and credible. The use of trade-off studiesand Pareto optimisation methods to assist in the captureof requirements also offers worthwhile benefits.

MDO is seen as providing the means to avoid thefragmentation inherent in established methods whichextends the time required for the design cycle andlimits the efficiency of final designs. MDO permits the

constraints of a diverse range of disciplines to bereflected from the earliest stages of the design process.This approach will facilitate the design of higher per-formance products with improved cost, structuralintegrity and maintainability. The methods will alsooffer the opportunity to maximise the exploitation ofnew materials technology within designs while mini-mising risk, and will have significant impact on projectdesign times and cost.

ACKNOWLEDGMENTS

Work referenced in this paper was funded undercontracts BE95-2056 and ESPRIT 20082 of theEuropean Community, by the UK DTI CARADprogramme and by the UK MOD corporate researchprogramme.

REFERENCES

1 NASA Workshop on MDO; Hampton VA, March13-16, 1995

2 6th AIAA/NASA/ISSMO Symposium on MDOBellevue WA, September 4-6, 1996

3 20th Congress of the International Council of theAeronautical Sciences; Sorrento, Naples,September 8-13, 1996

4 Final Report of the Garteur Action Group onStructural Optimisation SM(AG13),GARTEur TP078 & TP079,DERA/AS/ASD/TR97015 (Feb 1997).

5 Bartholomew P, Vinson S; STARS: MathematicalFoundations; In Software Systems for StructuralOptimisation, Birkhauser Verlag, Basel 1993

6 ISO 10301, STEP,http://www.ukcic.org/step/step.htm

7 Allwright S, Technical Data Management forcollaborative Multidiscipline Optimisation, AIAA-96-4160, 6th Symposium on MultidisciplinaryAnalysis and Optimization, Seattle WA, Sept 1996

8 A.Gould, “Surface shape optimisation”,MDO/TR/BAE/AG980105, 1998

9 The FRONTIER Project, ESPRIT Project 20082.http://frontier.ii.uib.no/

American Institute of Aeronautics and Astronautics1

98-4727Issues in Industrial Multidisciplinary Optimization

J. Bennett, P. Fenyes, W. Haering, M. NealBody Engineering and Integration Department

General Motors Research and Development Center30500 Mound Road Warren, MI 48090-9055

Abstract

Several mathematically based multidisciplinarydesign strategies are illustrated with an exploratorymultidisciplinary analysis and optimization package ona simple example problem. These examples are used tomotivate a discussion of required data handling andprocessing modules. These requirements envision asituation where some disciplines may havecomputationally expensive analysis capabilities andwhere not all disciplines have easily availableapproximations for all required quantities

Introduction

There are generally two areas of development inmultidisciplinary optimization and design systems. Thefirst is the formal mathematical approach that isgenerally characterized by the work presented at theMultidisciplinary Design and OptimizationConferences. The second is a more ad hoc approachwhich is evolving from the traditional design andanalysis communities and is typified more by aMultidisciplinary Analysis capability that is evolving inthe commercial CAD and CAE environments. Theseenvironments envision a common parametricdescription of the artifact and an ability to generateinput information for several disciplines from thisformat. Then analyses will be performed using complexcommercial or proprietary codes and decisions made onhow to modify the initial design. This process is usuallycharacterized by significant human interaction todevelop the artifact model, generate the analysismodels, execute the analysis models and finally toexamine the output and make decisions. The formalmathematical approach tends to use much simpler andeasily modified local analysis methods (that execute onthe order of minutes or seconds) and to concentrate onmultidisciplinary design algorithms which interact withthe analysis methods in an almost automatic fashion.

Copyright c 1998 by the American Institute ofAeronautics and Astronautics, Inc. All rights reserved.

This paper will discuss some of the issues associatedwith developing an industrial, rapid, multidisciplinarydesign system that makes use of some aspects ofmodern multidisciplinary optimization research whilebeing constrained by analysis software packages that donot all have consistent local optimization capabilitiesand are not easily modified.

The work is conducted in the experimentalIntegrated Vehicle Design Analysis (IVDA) system thathas been developed at GM R&D Center over the pastfew years. This system was described in some detail in[1]. Only that detail which is critical to the presentdiscussion will be included here. A flow chart of thecomplete system is shown in Figure 1. This systemenvisions a parametric description format that for aspecific instantiation of the parameters will generate acommon vehicle description that in turn is used togenerate input for an extensive set of disciplinaryanalysis capabilities. Note that this vehicle databasecontains more than just a geometric description of thevehicle in that it includes materials and their propertiesas well as mass and inertia characteristics of variouscomponents. The geometric design parameters includeboth global vehicle dimension and component structuraldimensions as shown in Figure 2. The various analysistools represent a range of capabilities. There are bothcommercial and proprietary codes and some disciplineshave design (optimization) capabilities and others donot. The analysis capabilities in each discipline wereselected to represent a preliminary analysis capability.In most cases they represent neither the simplest nor themost complex analysis capability in each discipline.They do characterize the current state of computerbased engineering analysis. An initial goal of the systemwas to be able to complete one full analysis cycle in 24hrs. For many of the disciplines the development of theinput data is considered to be the major time constraint.For this reason, highly automated model generationmethods based on templates were developed in eachdiscipline.


Figure 1. IVDA System Modules and Flow

Figure 2. Overall Body Parameters and Typical Structural Parameters

Elastic StructuresODYSSEY, NASTRAN

External AerodynamicsGM Program

Solar LoadGM Program

Occupant DynamicsCAL3D

Suspension Loads ADAMS

Fuel EconomyGM Program

CrashworthinessLPM, DYNA3D

Other Analyses*

ResultsDatabase

VehicleDatabase

Par

amet

ric

Mod

eler

Dis

cipl

ine

M

odel

ers

LibraryData

Templates

User Input

Mul

tidi

scip

linar

y

Opt

imiz

atio

n

Dis

cipl

ine

Ana

lysi

s &

S

ensi

tivi

ty C

alcu

lati

ons

WB

L

FO RO

GC

FH RH

LA

LF

LB

LR

LD

LH

h

b


As will be discussed later, an experimentalmultidisciplinary optimization capability has beenadded to the IVDA system. One of the goals of thissystem was to be able to examine several alternativeways of implementing the design strategy. For thatreason there are currently few enforced sequences inthe system and the control of the execution of thevarious modules are under the control of the variousdiscipline and coordination human operators. Toillustrate various aspects of this design process and itsimplementation in IVDA, we will begin by showingseveral examples. Each example is based on thesimple problem outlined below but each uses adifferent design strategy. The final part of the reportwill discuss the technology used to implement thesecapabilities and some of its implications.

Examples

The example problem considered is to find a rearoverhang (RO) that maximizes fuel economy. Theavailable design variables are the total vehicle length(which in this parametric model expresses rearoverhang since all quantities forward of the rear ofthe vehicle such as rear wheel location are notfunctions of the total length) and the traditional beamcross section sizing dimensions. The shape, other thanlengthening, of the rear of the vehicle is notconsidered.

The underlying mechanics of the problem arethat as the vehicle is lengthened, the drag will godown, which would tend to increase fuel economy.However, the mass of the vehicle increases whichdecreases the fuel economy. In addition, increasingthe length of the vehicle decreases the fundamentalbending and torsion frequencies. If these frequenciesare below their target values, additional mass may berequired for structural stiffening to bring thefrequency back to its requirement.

In IVDA the structural analysis and design ishandled by using a beam spring model and using theODYSSEY/NASTRAN [2] programs to calculateresponse and gradients (sensitivities) and optimizemass for given constraints. An analysis takesapproximately 10 minutes on a workstation and anoptimization 1-3 hours. The aerodynamic drag iscalculated by a neural net fit to test data so it isessentially instantaneous. Similarly, the fueleconomy calculation is a rapid spread sheetcalculation. Because both the aerodynamic and fuel

economy calculations could have been replaced withmore accurate and time consuming calculations, wewill treat the process as if all three of the calculationsrequired significant amounts of time. The flow of datais shown in Figure 3.

Figure 3. Information Flow in Example Problems

The traditional way to work this problem wouldbe that every time information is required, a full cyclethrough the analysis is performed. That is fullaerodynamic and structure calculations need to bemade to calculate drag (Cd ) and the structural massmust be adjusted to reflect any frequencyrequirements, then a fuel economy calculation can bemade. Some design or optimization process woulddrive these calculations. Because this approach tendsto require many calls to the analysis process, thistends to be a rather inefficient way to approach thesolution.

Given the capabilities in a system such as IVDA,the above problem might be implemented in severalmore efficient ways, three of which we will illustrate.Each of these mimics a non-computer-based designstrategy. In Example 1 the disciplines are only askedto provide local response information and after somepoint in the process are told to execute a design usingonly local design variables and local constraints. InExample 3, a roll down of requirements is initiatedfrom the beginning, and much local design work isexecuted, but some iteration at the global level isrequired. Example 2 is an intermediate approach thatmakes use of local directional (sensitivity)information to guide the global design. This is often

StructuralAnalysis

Fuel EconomyAnalysis

Aero Analysis

Section Sizes

Frequencies

Mass

Drag (Cd)

Front Overhang


proposed as an initial step in flowing downrequirements.

Table 1. No Initial Local Design Information

Step Length Fuel Cd Mass FrequencyEconomy (>25)

mm Mpg Kg Hz0 4627 35.96 .292 1133.3 25.481 4677 36.04 .290 1134.8 25.442 4922* 36.17 .282 1142.3 24.653 4838* 36.15 .284 1139.7 25.04

Table 2. Local Design Information


mm mpg Kg Hz0 4627 35.96 .292 1133.3 25.481 4677 36.04 .290 1134.8 25.442 4831* 36.18 .284 1138.0 24.953 4848* 36.16 .284 1139.2 24.97

Table 3. Roll Down of Design Available


mm mpg Kg Hz0 4627 35.96 .292 1133.3 25.481 4677 36.04 .290 1134.8 25.442 4970* 36.15 .281 1144.8 25.003 4972*

* Multidisciplinary Optimization Result

Example 1: No initial local designThe concept is that design variables naturally split

into those that are of a global nature and those that arelimited to a specific discipline. For this first strategy,only local response information is requested until thefinal step. This example is shown in Table 1. Step 0represents a baseline design in which the structuralcross section dimensions have been optimized for thegiven rear overhang (total length = 4627). The fueleconomy, Cd, mass, and critical frequency constraint arealso reported. Since only response information isreturned, a step is required to generate directional(sensitivity) information. An arbitrary perturbation of50mm is taken in Step 1. There is now sufficientinformation to generate a linear approximation withrespect to length for all needed quantities; such as mass,Cd ,frequency and fuel economy. An optimizationalgorithm is then applied to identify an optimum length

of 4922 which is labeled Step 2. This new length is thentransmitted to the analysis codes and the values of Cd,mass, frequency, and fuel economy are calculated. Notethat the frequency constraint of 25Hz is violated,primarily because the linear approximation was notsufficiently accurate. However, there now existssufficient information to construct a quadraticapproximation to all quantities. Using theseapproximations an optimization is again conducted,identifying an optimum length of 4838. When thislength is returned to the analysis codes the remainingvalues in Step 3 are calculated. It would be possible tostop at this point, or the local analyses could conduct alocal design in those variables that do not affect any ofthe other disciplines. In this problem this would be thecross-section design variables of the entire structure(there are 116 of these). This was done, requiring 3additional structural analyses, and reduced the mass by


.2 kg which had no measurable effect on the fueleconomy. This process initially used 4 analyses in eachdiscipline plus 2 analyses and 2 sensitivity calculationsin the final structural optimization.

Example 2: Local information availableThis implementation uses more than just response

information from the disciplines. The results for thisapproach are shown in Table 2. Step 0 is identical toStep 0 in Example 1. In the example problem, thestructures disciplines can in fact generate sensitivities ofthe mass and frequency with respect to the cross-sectiondesign variables. The length variable is not directlyknown to the structures module so a sensitivity withrespect to length cannot be generated. Therefore a Step1 which is identical to the Step 1 in the first examplemust be made. Approximations with respect to thelength (based on a linear response surface) and withrespect to the section design variables (based onsensitivities for 116 variables) can now be made. Themultidisciplinary optimization problem can then besolved using all 117 design variables. Only the length isshown in the table (4831). Upon re-analysis thefrequency constraint was slightly violated (24.95). Atthis point a quadratic approximation based on lengthand updated sensitivity values for the section variablescan be generated. The structural approximations basedon length are not precisely correct because they containnow an evaluation in which the section variables as wellas the length were changed. It is impractical to generatea response surface for all 117 variables since it wouldrequire a minimum of 118 analyses. Step 3 shows theresults of the approximate multidisciplinaryoptimization (length 4848) and the subsequent fullevaluations (frequency = 24.97, Fuel Economy = 36.16.This process used 4 analyses in each discipline and 2structural sensitivity calculations. Note that no finallocal structural optimization was performed so thereremains the possibility that this final design is notprecisely optimum.

Example 3: Roll down of designIn this implementation advantage is taken of local

design capabilities. The structural discipline capabilityhas the ability to perform optimizations (designs) whichminimize the mass subject to constraints. The results areshown in Table 3. Again Step 0 is the baseline and Step1 is the 50 mm move, however the results shown in Step1 are for a structurally optimized design in terms of thecross-section dimensions. Also the approximations usedfor the multidisciplinary optimization in Step 2 arebased on optimized structural designs. The Step 2multidisciplinary optimization produced a length of4970. When the structural optimization was againperformed the frequency constraint was initially

infeasible (24.32), but the local optimization was able toresolve this and the final design from Step 2 showed afrequency value of 25Hz. Now quadraticapproximations can be built based on these optimizedresults. The multidisciplinary optimization in Step 3then produces a value essentially identical to the lengthfor Step 2 so we can conclude that the design hasconverged. This process used 4 analyses in eachdiscipline with 6 additional analyses and sensitivitycalculations for structural optimization.

Discussion of ExamplesBecause of the nature of the problem solved it is

not possible to draw firm conclusions about eithervehicle design trends or the nature of which designstrategy is best. What has been shown is a computerimplementation that will allow these multidisciplinaryproblems to be handled mathematically and will allowdifferent design strategies to be applied. The followingsection of the paper will discuss these issues. However,first, some observations based on the examples can bemade.

The numerical differences among the quantities arein many cases extremely small. However, throughoutthe many exercises of these examples there has beensufficient consistency in the results to suggest that theyare not being driven by numerical noise. All of thedesigns consistently allowed the length to increase,which means that the gain in fuel economy fromdecreased drag offsets the decrease in fuel economy dueto the increase in mass. However, once the frequencyconstraint was encountered, the additional massrequired to meet the constraint at the longer lengthseventually overrode the fuel economy gains due to theincreased length. The range of final lengths (4838-4970) and Fuel Economy (36.15 - 36.16) suggest arather flat optimum over a relative wide range oflengths. To reliably select a true optima from thesedesigns is probably impossible with the available levelof accuracy in the analyses.

Similarly it is impossible to identify a best processfrom this simple example problem. All of the processeswork relatively well in terms of efficiency and quality ofanswer since the two sets of design variables (lengthand section dimensions) are fairly well independent forthis problem. In addition all examples were started froma structural design that was quite good (optimal) for theinitial total length so the effects of large changes in thestructural cross section design variables was eliminated.However, it is possible to see some of the relativestrengths and weaknesses. The no local design approachappears to have the most difficulty in following thefrequency constraint, but initially requires the least


effort from the local disciplines. It is easy to believe thatin a more highly coupled problem, this could lead toinefficiencies. The second approach, which brings inlocal sensitivity information if available, could beconsidered the most efficient from the standpoint ofstructural analyses required, but because of theinformation that is to be shared, is the most complex toimplement either informally or mathematically.Ultimately this process will produce a large, butpotentially simple, design problem at the global level.The last process which used local design required themost local design effort (finite element analyses). Thiswill occur when there is coupling between thedisciplines that are neglected in the roll down process.

As indicated previously, in order to implement andautomate such a system, several new capabilities areneeded. The remainder of this report will discuss thesein light of the system that was used for these examples.

A Multidisciplinary Design System. Parametric/design variable modeling

One of the fundamental issues for successfullyimplementing a computer-based multidisciplinarydesign strategy is that each discipline must be able tocommunicate with other disciplines and the decisionmaking process with the same set of design variables orparameters. If different disciplines use differentdescriptions of the same quantity or geometric entitythey have no way to communicate, particularlymathematically. This says that some commonparametric description, or a mapping among differentdescriptions, must exist. This was a fundamentalconcept of IVDA and resulted in a significant amount ofthe development effort.

Since the early 90’s the CAD vendors have beenevolving such a parametric capability for the geometricrepresentations that they create. Similarly, some CAEvendors, notably the finite element structural analysisvendors, have been evolving optimization capabilitiesbased on parameterized design models. It is logical thata bi-directional coupling could be established. Ingeneral this may not be easy since many disciplineshave evolved their own geometric preprocessors whichparameterize the discipline model in ways that aredifferent from the CAD models. We will assume theexistence of such a common parametric system in whatfollows since we wish to focus on the multidisciplinarydesign issues, however to insure appropriateimplementation, it may be necessary for the MDOcommunity to be actively involved in the evolution ofthis technology.

Approximate ProblemsThere is a large amount of heuristic and research

information suggesting that the way engineering designis efficiently conducted is that a limited amount of highquality, time consuming, expensive information iscollected and a simple approximation of thisinformation is constructed, either heuristically ormathematically. This simple model is exercised toidentify an improved design and this new design is thenevaluated using the high quality and expensive method.This process is certainly used in the heuristic and testmethod of design and the current analysis basedmethods. From the research standpoint it has been wellestablished in the structural optimization area that thisapproach reduces the computational effort by at least anorder of magnitude for moderately complex problems.We will propose that the ability to create and handleapproximations based on more refined data is required.

In some disciplines highly accurate, extremely fastanalyses may exist. From our standpoint these becomehighly accurate approximations for which no referenceto a more accurate analysis need be made.

Multidisciplinary Design StrategyGiven the above assumption that a set of

approximations will be available there are severalpieces to this strategy. First there will need to be someprocess to operate on the approximations to identify thenew and improved design. Since this will operate on thecheap-to-execute approximations, we will assume thatany strategy, including exhaustive search, could beused. In practice, exhaustive search many proveinefficient and the process probably would be selectedto take advantage of the nature of the approximations.The next level of the strategy is how the approximationsare generated. The final piece of the strategy is how theapproximations, the designs based on theseapproximations and the more detailed analysis areinterwoven. The three example problems showalternative implementations at this level of strategy.These examples suggest that one would not want toimpose a strategy a priori.

Clearly at the core is the concept of theapproximate models built on information in what mightbe called the results database. Therefore, we will beginour discussion with how these approximations might beconstructed and managed. We will develop the conceptof the IVDA results database throughout this discussion,but it is essentially where all of the relevant informationthat comes from the discipline analyses is located. In theCAD environment, many product data manager (PDM)systems anticipate storing a pointer to files of completedanalyses. However, to make use of this information in a


design system, some sort of data extraction is required.In the IVDA system, we required each discipline tosupply specific subsets of their output data to the resultsdatabase. The form in the results database wasstructured to meet the requirements of themathematically based design processes envisioned.Although we located all of the information in onedatabase, there is no reason that the information couldnot have been distributed in several disciplinedatabases. Part of this vision was that the resultsdatabase would live through several differentmultidisciplinary designs as opposed to one singleexecution of a precisely stated problem. Thus apersistent idea was that old analyses might be reused toconstruct new approximations for newly posedproblems. A conceptual layout of the results database isshown in Figure 4.

In the following sections we will discuss several typesof data stored in the results database and theirrelationships.

Response Approximations

The type of an approximation that we areconsidering here is a calculation that takes fractions of asecond to execute on whatever the current computeplatform is. These would generally fall into twocategories. The first is some sort of standardmathematical form that would be suitable for anydiscipline. Forms such as Taylor series, polynomialresponse surfaces, and neural nets fall into this category.The second category contains models developed for onespecific discipline. These could be spreadsheets,lumped parameter models, simple discrete models, orspecialized response surface or neural net models. It isanticipated that the generalized mathematical modelswould be generated from data that exists in the resultsdatabase. The discipline specific models would beeither generated or enhanced from the data in the resultsdatabase. Thus we clearly need to provide capabilitiesto interrogate the results database to create and updatethese models.

In IVDA we generate response approximationswith up to quadratic terms using two different methods.In the first approach, we use discipline generatedresponses and sensitivities at a single design point tocreate linear response approximations. The secondapproach uses only the discipline generated responsevalues at a number of points in the design space tocreate up to second order response surface

approximations. Both types of approximations arestored in the same relation in the results database.

Approximations Based on Discipline ResponseSensitivities

We chose to make the creation of an approximationa decision at the multidisciplinary level as opposed tothe discipline level. This is so the multidisciplinarydesign process would “understand” the approximationsit had available. On the other hand, we predicated theresults database on the idea that a decision to place datain it was made by the local discipline. This essentiallyplaced the burden on the local discipline to warrant thatthe data was correct and might have some potentialvalue. Thus it was necessary to create a location to storediscipline supplied sensitivities prior to the decision toelevate them to approximations to be used in themultidisciplinary design process. While response datacan be fairly compact (responses and associated designvariables), sensitivity data can be rather extensive andto store this information for each returned response maybe prohibitive. For that reason, in the IVDA resultsdatabase only one set of sensitivities for each disciplineis kept, the last one returned. In practice this has provedcumbersome because it requires that beforeapproximations are to be constructed, the “correct”response is the last one loaded. A module has beenimplemented which on command transfers thesensitivities for a particular response from thesensitivity relation to the approximation relation in theresults database.

Approximations Based on Response SurfacesMost response surface generation processes assume

that they are provided with a set of responses and theircorresponding sets of design variables. They then use aprescribed algorithm to develop an interpolation schemethat fits these points in some best way. Mostoptimization or design methods that use responsesurfaces presume that for every new problem, a startfrom no information is made and that the information todevelop the response surfaces is provided (perhaps n+1vectors of the n design variables for which theresponses must first be calculated). We wish to operatein a situation in which several designs have alreadybeen created (i.e. the results database is partiallypopulated) and we wish to use as much of the alreadygenerated information as possible. We do recognize thatat any given point in time there may be insufficientinformation and some additional analyses may need tobe executed, but we wish to minimize the amount of thisthat must be done.


Figure 4. Results Database and Associated Modules

To accomplish this a module was developed whichfor a given response will identify in the results databaseall designs which calculated a value for this response. Itthen identifies the values of the design variablesassociated with these responses. At this point theinformation is available to submit to a standardresponse surface generator. There is a question as towhether all of the responses in the database are “good”in that some could have been loaded and laterdetermined not to be valid. For that reason, it is possiblefor the user to specify which of the available responsesare to be included in the fit.

The main difficulty with this process is associatedwith the potential interrelationships among the designvariables. In a traditional regression approach, it isassumed that a unique set of design variables isidentified and remains constant throughout the process.In this case the only concern is that the designs used forthe response surface must not be linearly dependent inany fashion and there are standard methods to detectthis situation. In the process proposed above theseconditions cannot be guaranteed unless unreasonablerestrictions are placed upon the design process.

Design Variable Linking with Approximations inMultidisciplinary Design The fundamental concept in parametric modeling isthat the number of degrees of design freedom arereduced by relating potential degrees of freedom to areduced set of quantities by mathematical expressions.In the example problem described earlier the four pointsthat describe the rear of the vehicle (upper and lowercorners of the rear on each side of the vehicle) are allrelated to the total vehicle length. The process ofconstructing the relationships between potential designvariables and a reduced set of actual variables for agiven problem has been called linking in theoptimization literature and that term will be used here.In order to allow for some amount of generality andfuture changes, we anticipate that it will be desirable toretain access to this extended set of potential designvariables. Thus the basic set of design variables retainedin the IVDA results database is not the set of designvariables that are active on the current multidisciplinarydesign, but the complete set of design variables that isavailable in the template. For the examples shownearlier there are 3033 of these potential designvariables. Again taking the examples describedpreviously, the response surface module will identifythat four of these variables (upper and lower, right andleft) have been changed for any design that changes therear overhang. However, for this problem, the local

parameter description

parameter values { }response description

response values

linking histories

sensitivities

approximations

MDO problem description

Analysis 1Analysis 2..Analysis n

Multidisciplinarydesign and optimization

Approximatemodelbuilder

Parameter values Static values Analysis 1 . . Proposed new values Local design J . . MDO 1 MDO 2 . .

CAD Master parameter model

Design in progressparameter model

Parameter set upload module


linking specifies that all four quantities move the samedistance. Since the variables are not independent, weneed to fit only one variable, not four. Therefore someprocess must be implemented that recognizes thissituation and accounts for it. In the IVDAimplementation this is accomplished by creating ahistorical link table relation. This relation contains thelinks that are used in any set of responses that arereturned to the results database. It should be recognizedthat these links could have come from the centralvehicle description, or they could have been providedand/or modified by the local disciplines. The responsesurface generating module then checks to see if thesame linking was used for all of the responses to beused. If so, the number of design variables is reducedand the appropriate variables removed from theindependent variable list for fitting. This allows acorrect response surface approximation to be generated.

The difficulty here is that although severalvariables contributed to the total sensitivity calculation,all of the information is now attributed to one variable.This is fine as long as one wants to retain the currentlinking throughout the entire process. However, ifinformation from another discipline did not contain thislinking, and it was desired to allow linking changes inthe multidisciplinary problem, difficulties could arise.Therefore it is desirable to decompose the linked,aggregated sensitivity into the components of theunlinked design variables. This can be done in anapproximate sense by using the link relationships and achain rule. Then based on the linking used in themultidisciplinary problem, the appropriate linkedsensitivity can be reconstructed. For the specificexample used here, the sensitivities would be equallysplit among the four points. This capability has beenimplemented. The difficulties with this approach arealso obvious, since one might expect the sensitivities tobe equal from side to side, but the two lower pointscould be expected to have different sensitivities than thetwo upper points.

In looking at the current state of parametricmodeling implementation in CAD/CAE software, it isclear that many of these situations will arise here. It isrecognized that the discipline parameterization mustmatch the vehicle parameterization in terms of thequantities that are to be communicated. It is nothowever realized that the relationships among thesequantities will change and that an interpretable recordof these relationships may need to be kept. Just as itmay be appropriate for a discipline to propose a changeto a parametric dimension, a discipline may want topropose a change to the way these variables are linked,

and the decision making process needs a history ofthese proposals.

Multidisciplinary Analysis with Approximations

Although we are dealing with approximations, therelationships among the various approximations foreach discipline are the same as for the more complexmodules. Therefore the issues associated withexchanging information are the same. There are twopossible situations. Information from one discipline mayflow forward. That is the output from one disciplinemay be input for the next discipline. For instance theloads calculated by a suspension program might be theinput for a structural optimization program. Similarlythere may be a feed back of information in which theoutput of one program is needed to calculate the inputfor another program whose output is the input for thefirst program. For instance the mass calculated by thestructural optimization program is needed as input tothe suspension program which calculates the load inputfor the structural optimization program.

If there is only feed forward, it is fairly easy toenvision how a multidisciplinary design process wouldwork: by properly ordering the analyses (orapproximations), the outputs of the programs would beused as inputs for the following programs and allresponse properties could be properly calculated. Anoptimization capability could then be wrapped aroundthe feed forward package.

If there is any feedback present, the process is morecomplicated since there will need to be inner loopsaround these feed back loops to insure convergence ofthe responses before data is passed to the next step inthe process.

There are two approaches that might beimplemented here. Since we are working withapproximations that presumably execute very quickly,we could implement a full feedback and feed forwardprocess that would express all of the interactionsimplied in the approximations. This in general willrequire developing approximations of the outputquantities of each discipline module with respect to allof the input quantities, not just the design variables. Forexample in the example problem used here, anapproximation of the fuel economy with respect to Cd

and mass will be needed since these are the inputquantities needed by the fuel economy module (Figure3)

The second approach is through a mathematicalformulation. Mathematically this situation can be


expressed by what are called the global sensitivityequations.

dg

dx

g

x

g

g

dg

dx

j

i

j

i

j

k

k

i

= + ∑∂

∂

∂

∂ (1)

In this first order approximation sense theseequations allow us to create a complete approximation,dgj/dxi,to describe the effect of a design variable x onany output quantity g in terms of both feed forward andfeed back. To do this we need the traditionalderivatives, ∂gj/∂xi, (approximations) of any outputquantity with respect to its local input variables, plus thederivatives, ∂gj/∂gk, (approximations) of any outputresponse quantity with respect to any input responsequantity. This is essentially a set of linear algebraicequations that can be solved by standard matrixmethods.

Thus to implement either of the two approaches weneed the same types of additional information, i.e. eitherapproximations or sensitivities of response quantitieswith respect to input response quantities.

In the example problems, we had only a feedforward problem and response/response sensitivitieswere only needed for the fuel economy program. Weused a first order approximation, calculating thesensitivities by finite differences. This was implementedin the first approach, treating the sensitivities as anapproximation, and chaining the information throughthe approximations. This gives the same answer as theglobal sensitivity equations that in the case of feedforward reduce to a chain rule.

Multidisciplinary Design Strategies

As indicated earlier there are multiple levels to thisstrategy. We have proposed an approach in which thedesign or optimization strategy is applied to theapproximations. While virtually any optimizationpackage could be used, we used the feasible directionsstrategy in ADS [3] for the examples. Any of thesepackages require the availability of response andperhaps sensitivity information. Some process must bedevised to interface the optimization algorithms with theapproximation modules as described in the previoussection. While we developed a simple input format thatwould point to the appropriate modules, one of thenewer commercial MDO oriented packages could beadapted to the task. The output from the optimizationpackage would then be a proposed new design that mustbe reloaded into the high level description for re-analysis, if accepted.

This brings us to the relationships between the highlevel vehicle description and the other parts of theprocess. Both IVDA and the commercial CADpackages envision a high level description of thepresent state of the vehicle which is the common centraldescription which all analyses reference as their startingpoint. It is then envisioned that the local disciplines mayexplore alternative designs and propose a new set ofdesign parameters. Most of the current CAD vendorthinking is around the process of allowing onediscipline to upload its new set of parameters to theCAD model. It is unlikely, however, that all of adiscipline’s proposed changes would be accepted. Themore likely situation is that many disciplines maypropose conflicting sets of design parameters and theseconflicts must be resolved before a design can beuploaded to the central description. This is essentiallythe job that is handled by the multidisciplinaryoptimization process. Obviously, there needs to be someintermediate level of data storage to handle all of theseproposed new designs. The results database in IVDAstores these proposed designs from all the disciplines.No discipline can directly input its results to the vehicledatabase. The only way that the vehicle database (CADmodel) can be updated is through the results database.A module was created that will select a complete set ofdesign parameters in the results database and return it tothe vehicle database (Figure 4). This then treats theapproximate multidisciplinary optimization as justanother discipline that has returned a proposed designthat can be selected for return to the results database.

Although it was not implemented in the currentversion of IVDA, it is reasonable to assume that anintermediate copy of the vehicle database will beneeded to hold modifications of the design parametersthat are used to construct approximations required bythe design process, such as those required by the lengthvariable in the structural discipline in the examples.This is shown in Figure 4 with broken lines.

The remaining issue is how a high levelmultidisciplinary design strategy will interact with theapproximate design strategy and the complex analyses.In the example problems this was handled by directinteraction, implementing all of the necessaryexecutions through high level IVDA commands thatprovide for the input generation, execution, and transferof the results of the various disciplines. This makes thisprocess more time consuming than necessary, butbecause the exact series of steps is as yet undetermined,it seemed inappropriate to automate them, until suchtime as the rest of the system is more formalized. It does


suggest that a menu of appropriate actions should begenerated to guide the user through the process.

Summary

By first using some simple example problemswe have tried to motivate a view of a mathematicallybased design process that has parallels in the traditionalprocesses. This vision involves an interplay betweencomplex, time consuming analyses and approximationsbased on these analyses. It is unlikely that a singlemultidisciplinary design strategy will suffice for allproblems. Therefore a system must evolve that willhandle a number of different strategies. Three classes ofissues were discussed. First, a method to share acommon description of design parameters must beimplemented, Associated with this is a necessity to keeptrack of the linking relationships among the potential setof design variables as these may change throughout thedesign process. Second is the need to have for eachdiscipline a quickly executed approximation of aperhaps more complex behavior. These approximationscan either be supplied by the disciplines, for examplesensitivities if available, or they might be created at ahigher level by examining all of the available detailedanalysis results. The IVDA system was constructedexplicitly to examine the latter situation and requiredadditional sophistication to implement. Finally, if theprevious two capabilities are in place, a shared set ofdesign parameters and a shared set of approximations,the implementation of a design or optimization strategyis fairly straightforward and a wide range of strategiescan be implemented including heuristic andmathematically based strategies.

Acknowledgements

The authors would like to acknowledge the valuedcontributions of V. Sankar who did much of thedevelopment programming of the multidisciplinarycapability described here and who contributed to thediscussions of the basic strategy. We would also like toacknowledge Bob Lust for pointing out the use of thechain rule and the linking information to develop amore generalized approximation to the decoupling ofthe aggregated sensitivity information.

References

1. J. Bennett, et. al., “A Multidisciplinary Frameworkfor Preliminary Vehicle Analysis and Design,” GMResearch Publication R&D-8290, Feb 9,1995.

2. M. E. Botkin, et al, “Structural Sizing OptimizationUsing an External Finite Element Program,”Proceedings of 28th Structures, Structural Dynamics andMaterials Conference, AIAA, New York, 1987.

3. G. N. Vanderplaats, H. Sugimoto, and C. M.Sprague, ADS-1: A New General-Purpose OptimizationProgram,” Proceedings of 24th Structures, StructuralDynamics and Materials Conference, AIAA, New York,1983.


AIAA-98-4731

1


MDO TECHNOLOGY NEEDS INAEROELASTIC STRUCTURAL DESIGN

H.G. Hönlinger* J. Krammer† M. Stettner ‡

German Aerospace Center (DLR)Göttingen, Germany

Daimler-Benz Aerospace AGMunich, Germany

Abstract.

Increasing performance requirements andeconomical pressure to reduce aircraft DirectOperational Costs can no longer be met bytraditional design processes. In particular, the impactof aeroelastic effects on aircraft design demands theuse of multidisciplinary design concepts andoptimization (MDO) strategies to develop flutter-free structures and to ensure excellent multipointperformance characteristics. This paper describes theaeroelastic and aeroservoelastic MDO problem,presents a variety of production and research levelmethods for its solution, and highlights currentbottlenecks. Industrial MDO technology needs foraeroelastic structural design are identified byreviewing previous aeroelastic studies performed atDaimler-Benz Aerospace AG Military AircraftDivision (DASA-M) and results from a study inwhich several DASA-M staff members were askedto specify future analysis and MDO needs. A trendtowards loosely coupled approaches is detectedwhich is opposed by a current shortage of frameworksoftware and MDO algorithms specificallysupporting industrial implementation and use.Another obstacle is the lack of standardized toolinterfaces. Finally, cultural changes are required inindustry to exploit the full potential of MDO.

* .Director, Institute of Aeroelastics† Manager Loads/Dynamics , Senior Member AIAA‡ Research Engineer, Member AIAA

Copyright ©1998 by the American Institute of Aeronautics andAstronautics, Inc. All rights reserved.

General Aeroelastic Requirements for HighPerformance Aircraft Design

The whole spectrum of aeroelastic phenomenato be considered during the design process can beclassified by means of Collar’s well-knownaeroelastic triangle of forces illustrated in Fig. 1.Three types of forces - aerodynamic, elastic andinertial - are involved in the aeroelastic process.Generally aeroelastic phenomena can be divided intwo main groups:

- static aeroelastic phenomena, which lie outsideof the triangle, i.e. divergence of the structure,control effectiveness, and load distributioncreated by aerodynamic and elastic forces,flight mechanic stability.

- dynamic aeroelastic phenomena, which liewithin the triangle since they involve all threetypes of forces, i.e. flutter, buffeting, anddynamic response or dynamic flight stability.

All of these aeroelastic phenomena haveprofound effects on the aircraft design and can onlybe solved in concurrent consideration by alldisciplines involved.

FLIGHTMECHANICS

VIBRATIONS

AERODYNAMICFORCES

ELASTICFORCES

INERTIALFORCES

DYNAMICSTABILITY

STATICAEROELASTICITY

Fig. 1: Collar’s Aeroelastic Triangle

2


The flexibility of the aircraft structure isfundamentally responsible for a variety ofaeroelastic phenomena and related problems. Aslong as the strength requirements are fulfilled,structural flexibility itself is not necessarilyobjectionable. The aeroelastic deformations,however, may not only strongly influence thestructural dynamics and flight stability, but also theoverall performance and controllability of theaircraft. Therefore in the conceptual and preliminarydesign phase of a new aircraft, the application ofaeroelastic design criteria becomes imperative forthe structural design and optimization process. Thefollowing design criteria, among others, become thestandard for any aircraft design:

- The aircraft must be free of flutter, divergence,and aeroelastic instability within its flightenvelope

- the control effectiveness must be above a givenminimum to assure safe flight performancewithin the flight envelope

- the flight shape of the wing should haveminimum aerodynamic drag and sufficienteffectiveness for all configurations.

Despite all these design criteria in a wide rangeof applications the airframe design processtraditionally starts with a strength design andinteractions with any control system are notconsidered in early design phases. Aeroelastic designcriteria, however, are related to flexibility. Flutterstability of strength-designed metallic wingstructures therefore must often be ensured with a nonoptimal repair solution. During the repair process,areas on selected components like spars orattachments with beneficial impact on flutterdamping are identified and reinforced. This can beaccomplished by analyzing the sensitivity ofstiffness and mode shapes to reinforcement1, asshown in Fig. 2. Nodal lines of the critical modes aremoved, and the flutter speed increases (Fig. 3). Onlysmall changes to the existing design are necessary,but a weight penalty is always added - in thisparticular case 95 kg.

Concurrent consideration of stress and flutterconstraints in a complete re-design, as demonstratedin a recent European research project on MDO2

(„MDO-Project“), has the potential to yield afeasible design with a smaller weight increase. Toachieve this goal, symmetric and antimetricboundary need to be regarded simultaneously. Alarge number of structural optimization tools,however, do not permit this approach. As a result,full models must be used, which may render thedesign problem too large to handle3.

Fig. 2: Results of Sensitivity Analysis (Ref. 1)

Fig. 3: Flutter Speed Increase (Ref. 1)

Similar „repair solutions“ are used forimproving aeroelastic effectivenesses of liftingsurfaces. Knowledge of structural bending-twistcoupling, i.e. observation of the physical behavior ofthe structure, may be used to manipulate the stiffnessdistribution appropriately4. However, interactionsbetween different aeroelastic requirements are notobvious. An example from the above-mentionedMDO-Project shows that wing optimization withaileron effectiveness constraints may have anoticable effect on the symmetric trim of large jettransport aircraft. Hence, cross-couplings betweensymmetric and antimetric load cases, boundaryconditions, and design requirements exist. Again,limitations of current optimization packages oftenprohibit simultaneous consideration of constraintsfrom different boundary conditions.

3


Servoelastic Aspects for High Performance AircraftDesign

The integration of modern electronic flightcontrol systems (EFCS) in combination with fly-by-wire technology offer the design engineers a chanceto implement additional active control functions inorder to gain benefits for the airframe performance.The interaction between the aircraft’s flight controland additional active functions has emerged as animportant design potential. This field of merging thedisciplines structural dynamics, aeroelastic and flightcontrol system dynamics is calledaeroservoelasticity. The interactions are illustrated inFig. 4.

STRUCTURALDYNAMICS

AEROSERVOELASTICITY

UNSTEADYAERODYNAMICS

FLIGHT CONTROLSYSTEM DYNAMICSAEROSERVODYNAMIC

DYN

AMIC

AER

OEL

ASTI

C SERVO

ELASTIC

Fig. 4: Aeroervoelastic Triangle

In the aeroservoelastic triangle, the left legrepresents the dynamic aeroelastic interaction whichdoes not include inputs from an active system.Similarly, the lower leg of the triangle stands forclassical aeroservodynamic control system synthesis.Finally, the right leg depicts the important dynamicservoelastic coupling between the elastic modes ofthe aircraft and the active control system. Thiscoupling, together with the unsteady aerodynamicfeedback inputs from servo-actuated active controlsurfaces, then results in an aeroservoelasticinteraction which is generally known as „structuralcoupling“ and can be as dangerous as flutter. Toavoid dangerous instabilities aeroservoelastic designcriteria have been developed for active controlfunctions which take into account flight dynamics,structural stability, and performance as well.

All active functions (control systems) have to bedesigned to cover full rigid and flexible aircraftfrequency ranges with respect to the aircraft rigidmode and structural mode coupling stabilityrequirements for each control system loop. Thestructural coupling influences will be minimized by

notch filters or other measures and the controlsystem must be as robust as possible with regard toall aircraft configurations and to all kinds of non-linearities of the complete system „flying aircraftwith active controls“.

Additional requirements to be met within thedesign process are:

- minimization of impact on actuator fatigue- minimization of impact on actuator back-up

structures fatigue life to reduce weight penalties

The most important active control functionswhich are mature for implementation are:

- care-free handling- Maneuver Load Control (MLC)- Gust Load Alleviation (GLA)- Fatigue Life Enhancement- Deformation and Elastic Mode Control- Flutter Suppression- Ride Comfort Improvement

The first experimental applications of thesefunctions have been repair solutions in most cases tomeet aircraft performance specifications. The fullpotential of this technology however can only beexplored when it is used as design tool and fullyintegrated into the MDO process of active aircraftstructures.

At present the aeroelastic design of activeaircraft structures is still the task of the future.Currently, only a patchwork of methods is available:

Active flutter suppression and gust alleviationhave matured from an academic to almost anindustrial application level5. In case of a controllerfailure, however, current certification proceduresrequire an actively controlled aircraft to bedynamically stable with the same, significant safetymargins as a purely passively controlled aircraft.Significant structural weight reductions fromexploitation of active flutter suppression cantherefore not be expected. It might pay off if thecommon 20% dive speed to flutter speed margin isvalid for actively flutter-suppressed aircraft, but areduced safety margin is accepted in the case offlutter suppression controller failure .

The EFCS is usually designed and optimized forflight-dynamic stability and performance with well-established methods. The aircraft model in thesemethods is derived from rigid structure propertiesusing aeroelastic efficiencies. Since the actualaircraft properties differ from those assumed in the

4


model - in flexibility and dynamic behavior -expensive and time-consuming adjustments arenecessary. Seyffarth et. al.6 suggest a two-stepprocedure to correct this discrepancy. First, sensorlocations, sensor attachments, actuator transferfunctions, and control surface dynamic properties areoptimized. Second, notch filters and phase advancefilters are designed to eliminate sensor signals fromstructural vibrations which could affect EFCSperformance and stability. This design task is initself a challenging optimization problem, asmultiple loading and flight conditions must beconsidered7.

This traditional separation between low- andhigh frequency behavior, EFCS and structural designdecomposes the design problem into manageable,discipline-conform sub-tasks, but also poses anumber of costly integration challenges.

Taking into account that computational powerdoubles per year new approaches to an integratedflight control design and optimization with respect toflight dynamics, active functions and aeroelasticstability requirements seem to become feasible. Highperformance computing will not only speed up theMDO process, it will start with better aircraft modelsand will allow the representation of multipleboundary conditions in an „integrated MDOprocess“.

Previous Aeroelastic Optimization Applications,Development Trends, and Technology Gaps

Table 1 provides an overview of some publishedDASA-M studies in the field of aeroelasticity whichreferred to identified shortcomings in the area ofaeroelastic optimization at the time of theircompletion. The table is not meant to provide asummary of these studies, but a condensed accountof information pertaining to the topic of this paper.The interested reader is referred to the originalpapers for details. The following section summarizestechnology gaps which were identified in the courseof these studies, outlines actions taken to close thesegaps, and identifies current trends in thedevelopment of aeroelastic optimization capabilities.

Selected activities in the 1980s focused onidentifying and solving stability problemsencountered with existing designs. Flight testing of a1/3 scale model of the SB-13 tailless glider airplanerevealed a severe low-speed instability8. Theproblem was analytically traced back to couplingbetween the aircraft’s short period oscillation andthe first symmetric wing bending mode. Oneapproach to alleviation of the instability was to use

structural optimization. Limitations of the programsTSO and FASTOP, however, made modeling of thiscoupled flight dynamic/structural elasticity problemvery difficult. It was concluded that flight dynamicsmust be integrated in aeroelastic optimizationsoftware.

In 1986, transonic wind tunnel tests were usedto validate composite fin designs obtained fromstructural optimization9. In static aeroelasticity andflight dynamics commonly a linear dependencybetween a given aerodynamic load coefficient and acontrol surface deflection is assumed. Thisassumption is the basis of the notion of aeroelasticeffectiveness, a constant factor representing the ratioof a given aerodynamic load coefficient achieved bya flexible structure compared to that of a rigidstructure. This wind tunnel test, however, showed anon-linear relationship between rudder twist andstagnation pressure. The phenomenon was tracedback to geometric coupling of rudder deflection andload-induced fin box deformation. Hence, flightdynamic calculations considering structuralflexibility in form of stagnation pressureindependent effectiveness values may be unreliable,and trimmed aeroelastic equilibrium loadcalculations are required.

As seen from these two examples, early studiesattempting to avoid aeroelastic stability problems byautomatic computational design revealed that thestate-of-the-art optimization tools of that time lackedseveral important analysis and modeling features.Structural analysis programs, on the other hand, didnot have the desired open, multidisciplinaryoptimization features. In order to satisfy both needs,the package LAGRANGE

10 was developed at DASA-M. As a structural optimization utility by design itincludes both FE-based analysis capabilities andoptimization features, for example a host ofoptimization algorithms and analytical sensitivitycalculation for a number of constraints .

The software was successfully applied tooptimization problems throughout the 1990s. One ofthe first applications was weight minimization of theabove-mentioned small, simplified ACA-Fin modelsubject to strength, aeroelastic effectiveness, flutterand gauge constraints11. Based on success withexisting analysis capabilities, additional desirablefeatures were formulated (refer to the „TechnologyGap“ column of Table 1, third entry from above).Among these were buckling constraints, which werehence introduced into LAGRANGE.

A larger example was optimization of the X-31A composite wing, which already considered

5


buckling in addition to strength, effectiveness, andflutter constraints12. The resulting ply orientations,however, were not suited for compositemanufacturing. The concept of Constructive DesignElements was therefore introduced in LAGRANGE toallow addition of tape laying constraints13 and shapevariables, for example for optimal stringerplacement in composite panels under bucklingloads14.

Entire aircraft were also modeled andoptimized. The JPATS contender „Ranger 2000“ in199415 was checked in preliminary design foraeroelastic stability, and the potential for controlsurface flutter was detected. The problem was solvedby positioning masses on the control surfaces. Themass values were then optimized with LAGRANGE.

The largest application yet was a StealthDemonstrator model to be optimized for minimumweight subject to strength, buckling, effectiveness,and flutter constraints3. In order to consider tensymmetric and antimetric load cases simultaneously,a full model had to be used. With 22,000 degrees offreedom, 11,000 structural elements, 360 designvariables and 110,000 constraints the problem was atthe limits of reasonable size with respect to run timeand the possibility of physical interpretation. Theneed for techniques to solve such tasks with reduced(half-) models and multiple boundary conditions isobvious.

A 1992 study with the ACA-Fin16 showed thatconstraints like buckling may complicate thestructural design space significantly. In such a casethe choice of optimizer (or a sequence of optimizers)determines whether an optimum, or even a feasibledesign, can be found. To date, algorithms suitablefor buckling problems are still sought.

By the mid-1990s, LAGRANGE had been used tosolve most traditional aeroelastic optimizationproblems using Doublet-Lattice-Methodaerodynamics with sizing, shape, and fiberorientation design variables. At the same time, theneed for integration of structural optimization,aerodynamic analysis and optimization, flightdynamics and control systems design became apressing issue. With regard to jig shape, deformedaircraft drag, and flexible aircraft flight dynamics,aerodynamic methods at least comparable in fidelityto those used in preliminary aerodynamic designwere required for aeroelastic computations.Furthermore, the capability to alter the globalstructural layout numerically was desired so thatstructural data could be quickly generated inresponse to changes in aircraft configuration - like in

the course of wing planform optimization. Furtherextensions of LAGRANGE were considered to beprohibitively complex and costly, and efforts weremade to use the system as a stand-alone componenteither in tight coupling with a few other disciplinarytools, or loose coupling within larger, more generalarchitectures.

With regard to aeroelastic modeling, tightcoupling of LAGRANGE with a higher order panelmethod, HISSS17, is about to be completed. Thiscombination allows accurate load modeling at giventrim conditions. Optimization control remains withLAGRANGE; it communicates with a stand-alonecoupling component via shared memory and inter-process control.

LAGRANGE already supplied sensitivityinformation in a 1991 study on integratedaerodynamic, flight dynamic, and structural designof the ACA-Fin18. Models of four planform variantswere generated manually and the sensitivity ofaeroelastic side force effectiveness with respect totaper ratio, aspect ratio, and area were calculated.These data were then inserted into the GlobalSensitivity Equation, GSE, for derivatives of sideforce coefficient (flight dynamics), side force(aerodynamics), and effectiveness (structures). Dueto the lack of a software framework supportingloosely-coupled, GSE-based algorithms at that time,global sensitivities were used to determine newcandidate configurations, but not to drive anautomatic optimization.

In the MDO-Project2,19 an automatic modelgenerator was developed and proved to be veryuseful for rapid variant generation for large transportaircraft wings. In one demonstration application ofthe program suite, a two-level weight minimizationwas implemented with structural sizing variables atthe lower level using LAGRANGE, and planformvariables at the top level controlled by the MDOframework tool iSIGHT20, which had becomeavailable since the 1991 ACA-Fin study. The FEmodels for LAGRANGE were produced by the modelgenerator. Experience with this tool also underscoredhow important it will be to develop future generatorswhich are applicable to generic wing-typecomponents (transport, fighter, tail, etc.) andmultiple structural concepts, and ensure robustnesswith regard to model degeneration. Similar modulesare required for other structural components.

In another task of this project, it was necessaryto combine the specific capabilities of theaeroservoelastic optimization tool, AIDIA, atAermacchi in Italy, with those of LAGRANGE at

6


DASA-M in Germany. A simple approximation-based approach was used for this particular multi-site problem19. Approximations of flutter constraintson one hand, and weight, stress constraints andaeroelastic effectiveness constraints on the otherhand as functions of sizing variables were generatedfrom data produced during optimization studies atAermacchi and DASA-M, respectively. Theseapproximations were then integrated into iSIGHT,and a design satisfying both sets of constraintsapproximately was found. The availability of aframework tool facilitated implementation of thismethod. For future studies it is desirable to have off-the-shelve software for generating multi-dimensionalfunction approximations of several types, too.

This review indicates that until a few years agothe focus of in-house developments was onimprovement of disciplinary analyses and integrationof new constraint types in a tightly coupledoptimization package. More recently, the need tocombine existing disciplinary analysis capabilities inorder to solve multi-discipline design problemsshifted interest towards tight coupling betweenspecific analysis tools where appropriate andpossible. Also due to growing acceptance of MDOideas in the company, lose coupling via standardinterfaces, controlled by framework software isplanned and tested for the general case. Importantpieces are still missing for industrial application:Software for supporting generic model generation,software for design space approximation, MDOmethods for multi-site, multi-partner problems, and aproduct/process data standard to allowstandardization of disciplinary tool interfaces, toname a few.

Future Industrial User Requirements

In order to provide a comprehensive picture offuture trends in aeroelastic structural design and userrequirements, a catalogue of questions prepared byMr. Joe Giesing of McDonnell Douglas Corporationwas presented to seven DASA-M staff members inthe field of aerodynamics and structural dynamics,ranging from disciplinary experts to technicalmanagers. Questions and answers are listed in Tables2 and 3. The following paragraphs represent asummary of all responses. Question numbers refer tothe order used in Tables 2 and 3.

Major barriers to MDO in industry (question 1)are, in the field of structural optimization, the lack ofoptimization algorithms for topology/layout/materialdistribution, and Mathematical Programming (MP)tools for dual formulations. Most MDO technologiesstill not mature enough for industrial application, or

have not been implemented in mature softwareproducts. Integration of disciplinary analyses isdifficult since tool interfaces do not match (see alsoquestion 5). Organizational and cultural aspects arean important factor, since the concurrent nature ofMDO processes differs significantly from thetraditional sequential practice. No coordinatingposition for MDO is present in typical industrialhierarchies.

The typical design problem (question 2) is tofind a feasible, better, or locally optimal design in amostly continuous design space. The globaloptimum is of lesser interest.

Current software integration tools (question 3)have only recently been used in MDO applications.Improved support of design process organization andgraphic visualization is required. The latter refersspecifically to monitoring of optimization progresswhich lends itself to physical interpretation andidentification of typical features of a family ofdesigns.

The most significant integrated simulationchallenge (question 4) is nonlinear, aeroelastic,trimmed load calculation. The next important stepwill be inclusion of EFCS design for fully integratedloads and performance calculation. Current practicalchallenges are handling of multiple designconfigurations in load calculation, and considerationof manufacturing aspects like tape steering incomposite design.

Five barriers for using disciplinary analysisprocesses in MDO and design (question 5) werementioned: Tool robustness, automation level, easeof use and checking, lack of control by experts, andlack of interfaces to other disciplines. The last itemincludes both consideration of other discipline’sneeds and data format compatibility. Sinceoptimization is also considered a disciplinaryanalysis, another problem is the reliability of currentMP tools.

Tightly coupled methods for solution ofintegrated simulation problems (question 6) areneeded for specific problems with strong or highdata volume couplings. Loose coupling is preferredthough, since the systems are more transparent andflexible.

Analytical sensitivity derivatives are used withinLAGRANGE, and the current challenge is to integratethis package and other disciplinary tools. Sensitivityderivatives for existing tools (question 7) aretherefore not if immediate interest.

7


Automatic differentiation tools (question 8) willmost probably not be used in-house. It is more likelythat this work will be contracted to academia orresearch laboratories.

The most important obstacles for usingoptimization (question 9) are user familiarity/trainingand difficulty in interpreting results. Improvedgraphical monitoring tools would facilitateinterpretation (see question 3). A cultural aspect isthat optimization is considered to be timeconsuming, so that in considering product cost vs.possible product improvement the management go-ahead for efficient optimization often comes too late.

The primary reason for not usingdecomposition-based optimization algorithms(question 10) is the lack of demonstrated andvalidated software packages. Furthermore, anefficient implementation of such concurrent designmethods faces organizational obstacles (see alsoquestion 1).

In the previous paragraphs, responses werelisted irrespective of the questioned persons’backgrounds and positions, although the individualperspectives definitely permeate the responses. Staffmembers involved in the actual computational workare primarily concerned with practical issues likehandling of current tools (analysis and optimizationalike), solution of aeroelastic simulation problemstoday or in the near future, disciplinary toolcoupling, and interpretation of results. Responses ofindividuals in charge of project and departmentmanagement focus on topics like decompositiontechniques or design process organization, andorganizational challenges impeding MDOimplementation in industry. This polarization is mostobvious in answers to the last question asking for thethree MDO developments which would facilitate thedesigner’s job over the next 10 years (question 11).Assuming that the term „MDO developments“ refersstrictly to general-purpose MDO algorithms,methods, and implementations, then the followingitems can be extracted:

- reliable, demonstrated, and validated softwarepackages for industrial-size applications ofMDO algorithms from conceptual to detaileddesign, including graphical monitoring, designspace approximation, multi-criteria decisionmaking, and analysis integration tools;

- standardized tool interfaces and disciplinaryanalysis tools which are developed withinterdisciplinary interfacing in mind; thisrequires identification of each singlediscipline’s (or tool class’s) required inputs andgenerated outputs;

- MDO algorithms suited for optimization tasksto be performed by heterogeneous industrialconsortia.

Organizational aspects do not fit within thisstrict definition, but are nevertheless very important.The answers reflect the opinion that multi-disciplineand concurrent design thinking is not manifested intoday’s industrial design processes and companystructures.

Summary

The need for increasing integration ofaerodynamics, structures, and control system designin a Multidisciplinary Optimization environment isevident both from past design trends and industrialuser predictions of future directions.

The most pressing issue for the structuraldesigner’s daily work is the gap in fidelity betweenaerodynamics used in performance calculations,flight dynamics and aeroelasticity. Generic flow-structure interaction techniques are needed so thatEuler and Navier-Stokes Methods can be used inearly design for reliable load and maneuverperformance predictions. High performance/parallelcomputing will enable practical use of thesemethods.

In the very near future, however, looselycoupled MDO strategies will be used in theindustrial environment. Software framework toolssupporting these approaches exist but need to berefined, extended, and validated for productiveapplication. Reliable, robust software for genericmodel and design space approximation is missing.When this is accomplished, MDO methods likeConcurrent Subspace Optimization need to proveapplicability to industrial use. Successfulimplementation might be the key to the requiredcultural change in industry towards concurrentengineering.

8


Model Task Method Key Finding Conclusion Technology Gap

SB 13 (1985)8 eliminate flutter

within flightenvelope

manual optimization/trend studies withTSO/FASTOP

flutter source:coupling of firstelastic symmetricwing and short periodmode

need to changematerial, sparposition and wingplanform

aeroelasticity - flightdynamics coupling

ACA-Fin (1986)9 validation of

optimized compositedesigns

wind tunnel tests nonlinear relationshipbetween ruddercommand and loaddue to structuraldeflections

traditional design for"effectiveness"possibly unreliable

aeroelasticequilibrium loadcalculation

ACA-Fin (1990)11 weight optimization

with strength, aileroneffectiveness, flutter,minimum gaugeconstraints; designvariables: compositeply thicknesses

optimization withLAGRANGE

efficient methods forcross-disciplinesensitivities and app.optimization proce-dures; modeling:FCS, aeroelastic equi-librium and thermalloads, buckling,dynamic response ;fiber orientation, sparpositioning; multi-objectiveoptimization

ACA-Fin (1991)18 integrated planform

& sizing optimizationfor flight dynamicand stressrequirements

total sensitivityanalysis using 3x3GSE (flightdynamics,aerodynamics,aeroelasticity)

transparency of GSE:automated procedurereflects well-knowncouplings in processstructure

GSE useful for morecomplex problems

optimization proc.using GSE; applica-bility of existingsoftware to integratedoptimization

ACA-Fin (1992)16 1. buckling influence

2. performance ofdifferent optimizationalgorithms

optimization withLAGRANGE

1. buckling is designdriver; 2. mostalgorithms find onlyclosest local optimum

1. buckling must beconsidered;2. sequence ofoptimizers necessary

suitable optimizationalgorithms forbuckling problems

X-31A (1990)12 minimize weight s.t.

buckling, stress,effectiveness, flutter

optimization withLAGRANGE, materialtests

optimal design notsuited formanufacturing

composite materialsmanufacturingconstraint modeling

Ranger 2000

(1994)15

alleviate fluttertendency inpreliminary design

optimization withLAGRANGE; flutterflight test; groundresonance test

mass positioning oncontrol surfaces moreefficient than struct.reinforcement

Stealth Demonstra-

tor (1995)3

minimize weight s.t.strength, buckling,aeroelasticity;symmetric andantimetric loading

optimization with fullmodel in LAGRANGE

sequence of algo-rithms successful;structural model verylarge (cycle time)

large optimizationproblems can behandled

simultaneous con-sideration of multipleboundary conditionswith half models

„MDO-Aircraft“

(1998)19

1. optimize wing boxstructure (sizingvariables) subject tostatic and dynamicaeroservoelasticconstraints atdifferent sites2. test MDO methodsand framework tools

development ofMDO coordinationmethod based on sub-problemapproximations;application usingMDO frameworktool; optimizationwith LAGRANGE

(stress and effective-ness constraints)

1. leads to acceptableglobal design if app.are good; partnerstudies yield hints atlocation of globalsolution2. tightly and looselycoupled methodsrequired dependingon design stage

no weight savingsfrom active fluttersuppression undercurrent regulations;sensor location &actuation systemparameters needed asdesign variables;interaction betweenroll effectiveness andsymmetric trim

suitable MDOmethod; automated,generic, robust modelgeneration; product/process datastandard;approximationgeneration software;aeroelastic effects inperformance andflight dynamics

currently still open technology gaps are underscored

Table 1: Aeroelastic Optimization Applications and Conclusions

9


1. What are the major barriers and challenges to MDO in industry?Do they pertain to the state of the art in computer sciences, theavailability of a suite of robust, automated analyses of variedaccuracy, the need for robust optimization algorithms and tools, orthe organizational (cultural) challenges you are facing?

SO industrial processes in companies aresequential; MDO requires more concurrentengineering processes; tool interfaces donot match; MDO technologies not yetavailable for industrial use (maturity);no coordinating function (persons) for MDOin industry

SO lack of general MP tools on dual formulations; lack oftopology/ layout/material distribution algorithms

MO organizational/cultural: acceptance of MDO by disciplineexperts and management

2. What is your design problem and design goal? For example, isyour goal a better design or the best design? Do you want the codeto find the optimum or just show you the design space? Is youroptimization mostly continuous or mostly discrete? Do you havemultiple objectives to maximize?

SO to get a feasible design is most importantA multi-point design: feasibility,

reliabilityDE to get a better designSO to find the nearest local optimum; variables: mostly continuous

(MP algorithm for discrete variables desirable for composites)MO to find a feasible, better design in a mostly con tinuous design

space with a large number of design variables and constraintsand multiple objectives

AEO reduced time (model generation, results evaluation); include alldesign drivers; discover critical aspects early; model close tomanufacturing (affordability, final weight); continuous processfrom definition to product

3. Has the current state of software integration tools helped yourimplementation of integrated design and analysis processes? Dothese processes require more than your current software tools candeliver in: Database management; distributed computing; analysisand design graphic visualization; analysis and design processorganization, integration, monitoring and control

SO analysis and design process organizationA expert systems for guidance and easy

implementation of newapplications/analysis tools; multi-criteria decision making tools

DE all itemsSO all itemsMO in decreasing order of importance: analysis and design process

organization, graphic visualization (monitoring!)SD visualization/monitoring: extraction of characteristic features of

a family of designs4. There is significant research nowadays directed to, for example,the multidisciplinary simulation for aeroelastic, fully nonlinear,multiple control surface, trimmed load calculation. Do youfrequently encounter different integrated simulation challengeswhich you believe require additional research and development?

SO integration of FCS design and structuraldynamics (including aeroelastics andflight dynamics)

A virtual aircraft in full flightSO tape steering subject to manufacturing aspectsAEO multiple configurations (fuel, stores, actuator failure modes)

5. What are the major barriers in the use of disciplinary analysisprocesses in MDO and design? Consider the following areas: Cycletime, automation, robustness, fidelity, ease of use and checking, andapplicability for MDO, loss of control by technical experts, or other.

SO robustness, usability and applicabilityfor MDO, tool interfaces

DE automation, ease of use and c heckingSO in decreasing order of importance: robustness, loss of control

by technical experts, ease of use/checking, reliability of MPtools

MO in decreasing order of importance: applicability to MDO(interfaces), loss of control, ease of use

AEO single discipline models too complex, not considering otherdisciplines’ requirements (e.g. statics FEM: wing mass,stiffness, DOFs); lack of understanding other disciplines’ needs

6. One can solve integrated analysis and simulation problems usingeither a tightly coupled or a loosely coupled approach. A tightlycoupled approach is a very efficient method but somewhatmonolithic and it requires a new simulation code development.Instead, the loosely coupled approach is less efficient, but moremodular and requires the integration of existing simulation codes.Most, if not all integrated analysis and simulation problems in placenowadays are of the loosely coupled variety. Would you consider theuse of a tightly coupled method?

SO loosely preferred due to complexit y oftightly coupled systems

A loosely coupled: allows for easilyexchanging analysis tools

SO both is needed!MO tightly coupled only for special problems (with strong

coupling); loosely coupled preferred due to flexibility

SO department manager, MDO expert MO MDO expert, aeroelasticianA R&D project manager, aerodynamicist AEO aeroelastician, structural optimization expertDE conceptual designer SD structural dynamicistSO structural optimization expert

Table 2: Responses to „MDO Requirements“ Questionnaire (1)

10


7. Sensitivity derivatives are available for a number of commercialand government-supplied simulation codes. Are there othersimulations for which you wish you had sensitivity derivatives?Also, would you consider using that information even in other thanan optimization setting? If so, do you have specific requirements onsuch sensitivity capabilities?8. Are you likely to invest time and effort as a user of automaticdifferentiation tools to produce your own sensitivity analysissoftware or are you looking to academia/ government researchers touse those tools and generate sensitivity analysis software for youruse?

SO academia/labsSO academia/labsMO academia/labs

9. What are the single most important obstacles to your use ofoptimization? User familiarity and training, optimization codeperformance, reliability/robustness, ease of use, difficulty informulating an optimization problem representative of the designproblems you face in your day-to-day applications, difficulty ininterpreting the resulting designs or in validating them, or other?

SO user familiarity and trainingSO performance, reliability/robustness, difficulty in interpreting/

validating resultsAEO management go-ahead for efficient optimization often comes

too late; external opinion: optimization costly, increasesproduct cost ("... for a 1% weight saving")

10. Few, if any of the currently available multilevel/multidisciplinary(CSSO, CO...) optimization algorithms based on decomposition havebeen used in industry. Do you attribute that to: The fact that onedoes not need in reality such general purpose methods, thecomplexity of the methods, the lack of maturity of the methods, or thelack of demonstrated and validated software packages?

SO lack of demonstrated and validatedsoftware packages

MO too complex/immature for industrial applications, also hardlyknown or understood (organizational aspects); lack of software

11. In order of decreasing priority, what are the 3 MD developmentswhich would help you do your job better, as a designer over the next10 years?

SO 1. process and company organization;2. standardized tool interfaces;3. demonstrated and validated MDO softwarepackages

A optimization strategies for heterogeneousprojects (with partners from variousindustry branches)

SO 1. conceptual design optimization tools; 2. more general MPtools; 3. easy-to-use monitoring tools

MO 1. product/process model standard (data format) and interfacescatering to it; 2. software (MDO algorithms, approximations);3. MDO strategies for multi-partner, multi-site optimization

AEO 1. completeness of single-disciplines' „set-of-needs “(automatic, integrated load case generation); 2. efficient aero-structures coupling mechanisms (generation, reliability,modifications);3. formulation of active a/c optimization approach: What is theoptimum deformed structure shape? How can it be achieved atminimum „cost“ (energy, mass of actuation system)? What isthe optimum passive structure and control system design toachieve an overall optimum design?

SO department manager, MDO expert MO MDO expert, aeroelasticianA R&D project manager, aerodynamicist AEO aeroelastician, structural optimization expertDE conceptual designer SD structural dynamicistSO structural optimization expert

Table 3: Responses to „MDO Requirements“ Questionnaire (2)

11


References

1 Snee, J.M.D., Zimmermann, H.; Schierenbeck,D., Heinz, P., „Simultaneous Stress and FlutterOptimization for the Wing of a TransportAircraft Equipped With Four Engines,“ Bath, UK1991. AGARD-R-784.

2 Allwright, S.A., „Technical Data Managementfor Collaborative Multi-discipline Optimisation,“6th AIAA/NASA/ISSMO Symposium onMultidisci-plinary Analysis and Optimization,Bellevue, Washington, September 1996. AIAA96-4160.

3 Krammer, J.M., and Lemmen, G., „IntegrierteStrukturauslegung mit dem Strukturoptimie-rungsprogramm LAGRANGE am Beispiel desfliegenden Technologieträgers,“ DGLR Jahrbuch1996, pp. 625-634. DGLR-JT96-079.

4 Haftka, R.T., „Structural Optimization withAeroelastic Constraints: A Survey of USApplications,“ International Symposium onAeroelasticity, Nürnberg, Germany, 1991.

5 Hönlinger, H., „Active Flutter Suppression on anAirplane with Wing Mounted External Stores,“Structural Aspects of Active Control, Paper 3,April 1977. AGARD-CP-228.

6 Seyffarth, K., Lacabanne, M., König, K., Cassau,H., „Comfort in Turbulence (CIT) for a LargeCivil Transport Aircraft,“ Forum Int.Aeroelasticité et Dynamique de Structure,Strassbourg, France, 1993.

7 Becker, J., Luber, W., „Flight Control DesignOptimization with Respect to Flight- andStructural Dynamic Requirements,“ 6th AIAA/NASA/ISSMO Symposium on MultidisciplinaryAnalysis and Optimization, Bellevue,Washington, September 1996. AIAA-96-4047.

8 Schweiger, J., Sensburg, O., and Berns, H.J.,„Aeroelastic Problems and Structural Design of aTailless CFC- Sailplane,“ InternationalSymposium on Aeroelasticity and StructuralDynamics, Aachen, Germany, April 1985.

9 Hönlinger, H., Schweiger, J., and Schewe, G.,„The Use of Aeroelastic Wind Tunnel Models toprove Structural Design Methods,“ 63th Meetingof the AGARD Structures and Materials Panel,Athens, Greece, September 1986

10 Schweiger, J., Krammer, J., and Hörnlein,H.R.E.M., „Development and Application of theIntegrated Structural Design Tool LAGRANGE,“6th AIAA/NASA/ISSMO Symposium on Multi-disciplinary Analysis and Optimization,Bellevue, Washington, September 1996. AIAA-96-4169.

11 Sensburg, O., Schweiger, J., Gödel, H., andLotze, A., „The Integration of StructuralOptimization in the General Design Process ofAircraft,“ 17th Congress of the InternationalCouncil of the Aeronautical Sciences,Stockholm, Sept. 1990.

12 Lonsinger, H., Günther, G., and Schweiger, J.,„Enhanced Fighter Manoeuverability Aircraft(X-31A) Wing and Thrust Vectoring VaneDesign,“ 11th American Society of MechanicalEngineers Winter Annual Meeting, Dallas,Texas, Nov 1990.

13 Schuhmacher, G., Multidisziplinäre, fertigungs-gerechte Optimierung von Faserverbund-Flächentragwerken, Dissertation, Universität-Gesamt-hochschule Siegen, FOMAAS, March1995. TIM-Bericht Nr. T07-03.95.

14 Eschenauer, H., and Weber, C., Stiffened CFRP-Panels Under Buckling Loads - Modeling,Analysis, Optimization, DE-VOL. 82, 1995Design Engineering Technical Conferences,Volume 1 ASME 1995, pp. 233-239.

15 Weiss, F., Schweiger, J., and Hönlinger, H.,„Flutter Flight Test of the RANGER 2000Aircraft,“ Meeting of the AGARD Structures andMaterials Panel, Rotterdam, The Netherlands,May 1994.

16 Hörnlein, H.R.E.M., „Overview of BenchmarkProblem MBB Fin,“ Final Report of theGARTEUR Action Group on StructuralOptimisation SM(AG13), Volume 3, Section C,Defence Evaluation and Research Agency,Farnborough, United Kingdom, February 1997.

17 Fornasier, L., „HISSS - A Higher-Order PanelMethod for Subsonic and Supersonic AttachedFlow about Arbitrary Configurations“, PanelMethods in Fluid Mechanics with Emphasis onAerodynamics, Notes on Fluid Mechanics 21,Vieweg Verlag Braunschweig/Wiesbaden, 1987.

18 Schneider, H., Krammer, J., and Hörnlein,H.R.E.M., „First Approach to an Integrated FinDesign,“ 72nd Meeting of the AGARDStructures and Materials Panel, 1991. AGARDReport 784.

19 Stettner, M., and Basso, W., „Multi-SiteCoordinated Aeroservoelastic SubtaskOptimization,“ 7th AIAA/USAF/NASA/ISSMOMultidisciplinary Analysis and OptimizationSymposium, St. Louis, Missouri, September1998. AIAA-98-4836.

20 ISIGHT Designer’s Guide, Engineous Software,Inc., Raleigh, North Carolina, 1998.


1

Multidiscipline Design as Applied to Space

Charles F. Lillie*, Michael J. Wehner and Tom Fitzgerald

TRW Space & Electronics GroupOne Space Park, Redondo Beach, CA 90278

Abstract

The objective of this paper is to look at thespacecraft design process and see how thatprocess balances desired spacecraft featureswithin an imposed set of operational and costconstraints. The constraints often show up ascompeting multidiscipline interactions, which intheir resolution lead to practical spacecraftdesigns. This paper gives examples of how thedesign process was implemented in a feasibilitydesign study for NASA's proposed NextGeneration Space Telescope (NGST), anddescribes how the project organization was usedto effectively deal with multidiscipline design.Orbit selection, spacecraft propulsion, stationkeeping, and overall mechanical and thermalsubsystem designs are discussed as examples ofmultidisciplinary design optimization. The finalsection is an across-the-board discussion ofmultidiscipline design optimization, what itsbenefits are, what are the negative points andwhat can be done to improve the process.

Introduction

This paper deals with work performed by theTRW-led study team under National Aeronauticsand Space Administration CooperativeAgreement No. NCC5-137, awarded May 24,1996 by the Goddard Space Flight Center, forresearch entitled: Next Generation SpaceTelescope Feasibility Assessments. The report________*Senior Member AIAACopyright © 1998 by TRW, Inc.Published by the American Institute ofAeronautics and Astronautics, Inc. withpermission.

was presented to the NGST Integration Team atGSFC on August 20, 1996.

The study was to involve industry, universitiesand/or non-profit organizations in the earlyplanning for the NGST in a search for the bestideas for accomplishing the mission. TheNGST project office felt that it would benecessary to go beyond simple parameter tradesto non-linear thinking in order to break thecurrent cost-aperture paradigm to achieve the$500M goal for NGST development, with atotal life-cycle cost of $900M in 1996 dollars..

This paper describes some of the major featuresof our approach to developing the NGSTspacecraft, launching it, and operating it for 10years. The paper includes the missionrequirements which we derived from theDressler Committee’s “HST and Beyond”report, and examples of the trades and analyseswhich we performed to develop a missionconcept and baseline configuration for theNGST, a development plan for enablingtechnologies, a cost estimate and arecommended management approach.

Figure 1 shows the organization of our studyteam and each team's responsibilities. Ourorganization paralleled that of the ongoinggovernment study to facilitate the integration ofour results with those from the other teams.

During the study the Integrated Product Teams(IPT's) responsible for the Optical TelescopeAssembly and for the Science Module workedclosely together to define an integrated payload

2American Institute of Aeronautics and Astronautics

with an optimum partitioning of functionsbetween the two assemblies. The Spacecraftsystems team was responsible for the classicalsubsystems as well as thermal shields, vibrationcontrol, and the fine pointing system. TheOperations team was responsible for the end-to-

end data flow, including the ground systemarchitecture and partitioning flight and groundsystem functions. The System Engineering teamhad responsibility for design integration as wellas requirements definition, mission analysis, andinterface definition

Figure 1. Study Organization

The responsibilities of the study team memberorganizations are summarized in Figure 2.TRW personnel led the NGST study IPTs andtook the lead in the system engineering, designintegration, science module, spacecraft bus andoperations activities. HDOS led the opticaldesign activities and supported the NGST studyin requirements development, materialsselection, performance modeling, active opticalsystems, and mirror assembly concepts. Swalespersonnel supported our NGST study in thermaldesign, contamination control, structure/mechanism design, science module design, andoperations.

Swales worked closely with personnel fromGoddard Space Flight Center with experience in

optics, structures, electromechanical devices,thermal control, cryogenics, contaminationcontrol, instrument design, and operations, whowere members of our IPTs and supported ourstudy activities. Additional support to the IPTswas provided by scientists and engineers fromthe Langley Research Center with expertise inanalysis and control of flexible structures, activestructures, active materials, isolation systems,and spacecraft analysis and modeling. The studywas also supported by scientists from severaluniversities who worked with their industrialcounterparts to define the system requirements,develop conceptual designs for the instruments,assess system performance and review theoutputs of our study.

Systems Engineering

Operations Systems

Spacecraft Systems

Science Module

Optical Telescope Assembly

• Optics • Light Baffles • Structure • Mechanisms • Cryogenics • Materials • Figure Control

N. Wallace, TRW

• Detectors • Cryostat • Cryocoolers • Cameras • Spectrometers • Wavefront Sensors • Deformable Mirrors • Steering Mirrors

T. Fitzgerald, TRW

• Electrical Power • Communications • C&DH • ACS • Propulsion • Structure and Mechanisms • Thermal Control • Vibration Control

K. Biber, TRW

• Ground System Design • Staffing • Data Reduction and Analysis • Planning and Scheduling • Contingency Operations • Normal Operations

D. Werts, TRW

• Systems Requirement Definition • Mission Analysis • Interface Definition & Control • Optical, Electrical, Mechanical, & Thermal Design Integration • Contamination Control • Launch Vehicle Interface

M. Wehner, TRW

Payload

NGST Study

C. Lillie, TRW

Science Advisors/Working Group

Cost Modeling

E. Wood, WW

B. Marcus, TRW LOB Manager S. Savarino, TRW Marketing


Most interaction between team members wasaccomplished by weekly telecons and individualphone, fax, and e-mail communications. Wealso established an Internet homepage for thestudy team to facilitate the flow of information,

including direct file transfers. This approachworked reasonably well, once the tools were inplace and face-to-face introductions of teammembers had been accomplished.

Figure 2. Team Responsibility Matrix

Opening a website at TRW to external teammembers required the development of newnetwork security procedures, which weresuccessfully implemented midway through thestudy. Once established, this website was veryuseful for the disseminating data to the team andarchiving the results of the study, as well asproviding pointers to other relevant informationon the Internet.

NGST System Design Process

We used our proven system engineering processon this project in defining mission requirements,deriving system requirements and developingsystem concepts. Several of these processes areiterative. The design features are balancedagainst the cost, risk and complexity of theconcepts to produce a baseline concept. As theconcept evolves the system requirements arefinalized. The final product is a baseline NGST

design and the associated technologydevelopment necessary to implement the design.

Design Reference Mission

We developed a Design Reference Missionbased on the Dressler Report and our ScienceTeam's expertise in astronomy. The NGSTsystem was optimized to provide high qualityinformation for investigating the early universeformation (using a large aperture and IRimaging). NGST would also continue theHubble telescope role of determining theHubble constant via Cepheid variables and othertechniques. NGST would have very significantcapabilities in ‘ordinary’ astronomy involvingstellar evolution, galactic structure, planetaryastronomy, etc.

Team TRW HDOS GSFC/SWALES LaRC/Science Team

1 SystemEngineering

Lead, system requirements,trades, analysis, designintegration

Requirements Development,optical performance modeling

Thermal design,Contamination control

Science team modelssystem performance fortypical targets

2 OTA,Includingstructures/mechanisms

Lead, deployable structure,mechanisms

Optical design, materialselection, modeling, assemblyconcepts

Support for structure andmechanisms design

LaRC supports activestructures design,technology roadmapdevelopment

3 ScienceModule

Lead, system design,payload accommodation,

Wavefront sensor, fineguidance sensor, active optics

Instrument design Science team supportsinstrument design

4 SpacecraftBus

Lead, classical bus design,vibration control, finepointing

Identify Enablingtechnologies, alternativedesigns, attitude control

LaRC supports vibrationcontrol, spacecraftanalysis and modeling

5 Operations Lead, ground system.design, mission operations.planning

Operations plans andscenarios, communicationslink trades and analyses

Science team supportsmission scenario prep.,MO&DA planning

6 Science/

MO&DA

Coordinate science advisor,working group activities

Science Support Science team reviewsstudy results,


Technology

Performance

$

Concept Assessments• Performance• – Science return• – Requirements• – Integrated modeling• LCC• Cost drivers• Risk• Schedule

Cost Performance

Outputs

System Trade Studies• Payload• Observatory• Spacecraft• Orbit• Launch vehicle• Ground station

Advanced instrument technologies

Primary mirror technologies

Instrument studies technologies results

Precision deployable technologies

Advanced spacecraft subsystem technologies

Mission concepts

NGST design(s)SI and mission operations requirementsArchitecture modelsPerformance assessmentTechnology roadmapLCC estimate and descope optionsDevelopment planAlternative approachesTechnology development

a)b)c)d)e)f)g)h)i)

Top-down requirements definition and bottoms-up technology application procesess produce an optimized design for NGST

Program Requirements• Cost• Schedule

Science Requirements• Wavelength• Aperture

Mission/Observatory Requirements• Design life• Launch vehicle

Revised requirements/cost allocation

Requirements and life cycle cost targets allocations

Requirements

Figure 3. System Design Process

• Early Universe Investigation (Z ~4 to 10)– ~50% of NGST observing time– 100 to 200 survey fields at high galactic

latitudes– Integration times ~10e3 to 10e5 sec

• Foreground Galaxies (Z ~0.5 to 3)– ~20% of NGST observing time– Observation of Cepheids, supernovae, etc.

(Hubble Constant)– Integration times ~10e3 to 10e4 sec

• Local Galaxy (including Local Group)– ~10% of NGST observing time– Stellar evolution, brown dwarfs, etc.– Integration times ~10e3 to 10e4 sec

• Solar System Objects– ~10% of NGST observing time– Planets, comets, asteroids, Kuiper Belt objects– Integration times ~10 to 100 sec

• Targets of Opportunity– ~12 to 24 hour response time

Figure 4. DRM

Mission Requirements

Based on the DRM and our Science Teamsguidance, we developed a set of MissionRequirements for NGST. These requirementsare essentially concept independent, demandingonly that NGST be a large aperture, imaging andspectroscopic IR optimized space telescope.Note that there are four graduations ofimportance in the requirements: 1) required, 2)highly desired, 3) desired and 4) goal. These areguidance to the concept designers as to theimportance of these requirements. We placedsome emphasis on targets of opportunity. Ourdesign incorporates features dedicated to this.We believe that such flexibility is essential toprovide data on comets and transient targets,such as supernovae.


The Dressler report and the DRM are directlyresponsible for the ‘quality’ requirement on thispage. Early universe objects are highly red-shifted, which reduces the need for visible lightobservations. Therefore, we designed the NGSTfor diffraction limited performance at 1 µm.Note also the required bands correspond to theDressler reports recommendations, but it wasconsidered advantageous for NGST to exceedthis band range if possible and cost effective.

Slit spectrometers are required. It was alsodesired that an imaging spectrometer be added,if feasible. We did not want NGST to belimited in stare time by design features.Therefore, a very long (~28 hours) requirementfor stare time was included.

The agility requirement of 30° in 15 minutes isexpected to not be stressing from a designviewpoint, and to provide a reasonably smallloss in total observing time. Given that themajority of observations are long exposures (~2hours based on the DRM), this implies that thetelescope is repositioned ~10 times per day,resulting in down-time of 2.5 hours out of 24,which is roughly 10% down-time.

Field of view of the imager has been a parame-ter much discussed. Larger is of course better,but has significant cost implications in requiringlarge numbers of pixels and stresses the opticsdesign. The value chosen is the same as the cur-rent Hubble Wide Field camera (if the squarewas filled).

Lifetime• 10-year Mean Mission Duration (MMD) (required)• 13-year design life (required)

Targets• High redshift objects (required)• Local area galaxies, clusters (required)• Milky Way objects (required)• Solar system objects – Planets (desired), outer solar system objects (highly

desired) – Near-earth comets/asteroids (goal) – Targets of opportunity within Field of Regard

Observations• Multi-color imaging (required• Spectroscopy (required)• Polarimetry (highly desired)• High speed photometry (desired)• Astrometry (desired)• Response times• Scheduled observations: � 1 month (required• Targets of opportunity: 24 hours (required; 12 hours

(goal)Aperture

• � 6 m (required)• � 8 m (highly desired)

Quality• Optics have diffraction limit (1/14 wave RMS) at 1µm

(required)• Nyquist sampled at lower end of each octave range

except for bands < 1 µmImaging Spectral Bands

• 1 to 5 µm (required)• 0.5 to 10 µm (highly desired)• 0.5 to 20 µm (desired)• 0.35 to 40 µm (goal)

2-D Spectrometer Bands (Slit Spectrometer)• l/� l = 1000 selected imaging band (but no greater than

0.5 to 20 µm) (required)• l /� l = 10000 in 0.5 to 20 µm band (highly desired)

3-D Spectrometer Bands (Simultaneous 2-D SpatialSpectroscopy)• l/� l = 50 in all bands (required)• l /� l = 1000 in 0.5 to 20 µm band (desired)

Stare Time• No system limitations up to 1E5 sec (required)• Sufficiently short such that bright targets not over exposed

(required)Agility

• Slew and settle a nominal distance (30°) within 900 sec(required)

• Sufficient to follow planets and outer solar system objects(required)

• Sufficient to follow fast moving comets (e.g., CometHyakutake) (highly desired)

– 0.5 arcsec/sec (highly desired)– 2.0 arcsec/sec (goal)

Pointing Stability• Total short-term jitter and long-team drift during exposure

results in � 20% larger diffraction image (note that isdependent of diffraction limit selected) (required)

Imaging Field of View• � 2.5 x 2.5 arc minutes (required)• � 4 x 4 arc minutes (highly desired)Spectroscopic Field of View• 2D slit � 30 arcsec (required)• 3D array covering � 0.5 x � 0.5 arc minutes(desired)

Coverage• 4 � steradian coverage of the celestial sphere (required)• Coverage of any solar system object greater than 1.5 au

from the sun, when projected onto the ecliptic plane(required)

Field of Regard• 1 steradian (required)• 2 � steradian hemisphere centered 180° from the sun

(hemisphere zenith pointing anti-sunward) (highlydesired)

Figure 5. Mission Requirements


Coverage is defined as the region which can beviewed by NGST over an extended time (likeone year). Field of Regard (FOR) is the regionwhich can be viewed by NGST over a short time(like one day). Field of View is the region thatcan be viewed by NGST instantaneously. Withthe coverage requirement defined, NGST will beable to view all parts of the celestial sphere andthe outer parts of the solar system. The highlydesired FOR enables target of opportunitydetection over half the celestial sphere at anyone time. The required FOR corresponds to a20° annulus perpendicular to the sun vector.This is commensurate with an NGST designwithout an elevation gimbal.

Baseline Concept

When stepping from the realm of missionrequirements to system requirements, it isnecessary to have a baseline system concept.This chart and the one following show theNGST baseline as of the conclusion of the threemonth study. In this paper we show the keytrades and requirements flowdown which led tothis baseline.

NGST is in a Lissajous orbit at the Lagrangian L2 point,placed there by an Atlas II AS (specified by thegovernment) which follows an Earth-Moon flybytrajectory. Communication to earth is via X-Band.

Visible IR Telescope

Space Vehicle

Visible IR Telescope

SpaceVehicle

9601255.014.SA043

X-Band Transponder/TT&C2 to 4 kbps

X-Band High Gain10 Mbits/sec

11 m Antenna

To Sun

Atlas II AS

6000 lb Space Vehicle

Lissajous Orbit

L2 Point

Moon

Figure 6. Mission Concept

A small (11 m) X-band antenna on the groundwill provide low cost support to the NGSTSpace Vehicle (SV). A dedicated ground stationwould schedule and operate the SV.

Figure 7 presents the configuration which wedeveloped for the NGST space vehicle. Notethe sun and thermal shields are cut away forclarification. The spacecraft bus is located at

the center of the shields, separated from theinstrument module and telescope by a boom.The space vehicle is kept oriented such that theshields shade the telescope from the sun andearth. An optional shield sized to shade thetelescope from the moon was considered, butrejected (shield size approximately doubled,from ~200 m2 to ~400 m2). The thermal load


from the moon is negligible; the impact ofsunlight reflected off the moon needs furtherinvestigation. The shields are supported bystruts, attached to the spacecraft. Note thesymmetry of the shield. This is to counteractsolar pressure. Note also the placement ofelectrochromic patches, which are used as trimtabs to balance the pressure with the spacevehicles center of gravity.

The telescope primary mirror is deployableusing TRW’s HARD (High Accuracy ReflectorDevelopment) technology. The telescope iscoarsely pointed with an elevation gimbal.After thermally stabilizing, fine pointing isachieved by ‘nodding’ the space vehicle androtating in azimuth about the sun line. A finepointing mirror provides final pointing and

tracking of the targets.

As an illustrative tool and as a guide to our tradespace, we present the key trades we performedthroughout this study. Note that some of theoptions are in italics and lined out. These arepotential solutions that were rejected. Thehighlighted options have been baselined.

Key trades

Spectral Band Options: It was decided that aUV capability for NGST would be costly andnot in keeping with the Dressler guidelines.Fabricating UV optics is expensive, andcoupling that with deployable optics wasconsidered too extreme. Similarly, to achieve

Figure 7. Key Design Features

Elevation Gimbal

Deployable Boom Separating Hot and Cold Regions

Silvered Teflon Sun Shield

Imbedded Amorphous Silicon Solar Array

Electrochromic Patches for Momentum Dumping

Smart Strut Booms Twist For Propeller Momentum Dumping

Rotate SV for Azimuth Control

Passively Cooled FPAs Passively Cooled Telescope f1.25 ~8 m Primary

4 Thermal Shields

10° Tilting of SV for Fine Elevation Pointing

Primary & Secondary Tip/Tilt Deployable Secondary Steering Cables for Elevation Gimbal

Deformable Primary and Wavefront Mirror

Deployable Struts Rigidly Support Shields

Biprop for Transfer Orbit

H2 Resistojets for Stationkeeping

Shield Cp Balanced with Cg

9602155.013.SA043


Figure 8. Payload Pointing

40 µm capability, we found that the optics wouldhave to be cooled below reasonable levels (nextfigure). Later we will show that due to costreasons, the 20 µm band was also rejected.

Transfer Orbit Options

A number of options are available to deliver the spacevehicle to L2. The selected baseline, lunar flyby withphasing loops, offers a large launch window with goodthrow weight. Direct transfer is advantageous as it hasa large launch window, but has the least throw weightof any of the options. Direct lunar flyby has the samethrow weight as the selected option, but has a very shortlaunch window. Integral propulsion is attractive as it

Figure 9. Key Trades

±10° SV Tilting

0° to 80° Elevation Gimbal

360° Azimuth Pointing

NGST Trade Tree (1/3)Spectral Band Options

• 0.1-0.5 µ m

• 0.5-1.0 µ m

• 1.0-5.0 µ m

• 1-10 µ m

• 1-20 µ m

• 1-40 µ m

Orbit Options

• L2 Lissajous

• L2 'Exact'

• L2 Halo

• L1 Orbits

• Drift Away

• L4,L5 Orbits

• 3 AU Heleocentric

• .1-.3 Heleocentric

Cold optics, FPAs

L2 Transfer Orbit Options

• Direct Transfer

• Lunar Flyby

• Lunar Flyby with Phasing Loops

• Integral Propulsion + Lunar Flyby

L2 Lissajous Orbit

Transfer Orbit

Propulsion Options

• Bi Prop

• Mono Prop

• Cold Gas

• None

Contamination Control

• Material Selection

• Propellent Selection

• Active Cleaning

Stationkeeping

Propulsion Options

• None

• Hydrazine

• Bi Prop

• Arcjets

• Pulsed Plasma

• Resistojets

• Solar Sailing


has the best throw weight of the options, butrequires a large amount of propellant. Due tolaunch vehicle size constraints, we do not havethe room to accommodate this additionalpropellant.

Figure 10 illustrates why NGST was notdesigned to operate at 40 µm. Operating at thispoint would require very cold mirrortemperatures, which are beyond a reasonabledesign capability.We allocated to thermal control the objective ofpassively cooling optics to ~30 K. Thispreserves the option of including a 20 µm band.The cost of achieving this temperature is verymodest, only requiring the inclusion of anadditional thermal shield layer. Therequirement for cold optics drives the SVconfiguration.

Assumptions• f/15 telescope• Zodiacal spectral radiance

– 10e-11 W/cm2 µm Sr• Bandpass: 0.1 µm• Irradiance: 3.5e-15 W/cm2• Mirror emissitivity: 0.1

70

60

50

40

30

208 10 12 14 16 18 20

Mirr

or T

empe

ratu

re, K

Cutoff Wavelength, µm

Pas

sive

Opt

ics

Coo

ling

Fea

sibl

e

Figure 10. Temperature vs. IR Wavelength

Orbit Options

Orbit Selection Summary

• L2 Lissajous requires no insertion ²V, low stationkeeping ²V

– Low meteoroid, solar flare flux

– Negligible thermal from earth and sun

– Good launch window, throw weight with “Lunar Assist + Phasing Loops” transfer orbit

• L2 halo requires insertion ²V

• L2 exact orbit requires high stationkeeping ²V

• L4/5 have very long communication ranges (1 AU)

• Drift-away orbit limits life; long communication range

• 1 AU heliocentric orbit needs further investigation

• 3 AU heliocentric orbit has lower throw weight; not needed for our bands

Near-Earth (and moon) orbits have a stressingthermal environment. Therefore, only orbitssome distance from the earth were considered.The Lissajous L2 orbit was baselined.Attractive features of this orbit are: short rangeto the earth, low station keeping requirements,

and no insertion DV to enter the orbit. L1 orbitshave no advantages and the disadvantage ofhigher solar flux and having the earth shining inthe telescopes field of regard. The drift away,L4/L5 and 3 AU heliocentric orbits are at longranges from the earth and have minimal


advantages in the primary band of interest (1-5µm ). Halo and L2 ‘exact’ orbits have highstation keeping and transfer DV requirements.The 1 AU Heliocentric orbit located 0.1-0.3 AUfrom the earth is still under investigation. Thisorbit may be able to be station kept at areasonable earth distance.

One of the advantages of the selected Lissajousorbit is that no burns are required to enter, and itrequires low DV to maintain. Also, this orbit isvery large (300,000 km by 600,000 km axes),and only needs maintenance occasionally. TheDV required to meet this the station keepingrequirements is 2-4 m/sec/year, or 20-40 m/secover the mission life.

Station keeping at L2

Considerations

• L2 is an unstable point, sostation keeping is required

• Serious contamination concerndue to cold optics temperatures

Station keeping Requirements

• Delta V: ~2 m/sec/year

• Station keeping maneuver timeliness �3 months

Contamination Concerns

• Contamination is a major concern in cryogenic optical systems

• Acceptable contamination levels have not yet been determined for NGST

• Water, oxygen, argon, nitrogen, etc. can freeze out on cold surfaces

• Contamination control approaches

– Select low outgassing materials for construction

– Exercise contamination control pre-launch

– Protect optical surfaces during launch and during early time on-orbit

– Perform vacuum bakeout and use molecular absorbers to reduce outgassing rates

– Minimize vapor and gas flux to cryogenic surfaces

– Periodic heating of surfaces to remove contamination

• A major potential source of contamination is the propulsion systems

– Prudent selection of the propulsion systems will reduce contamination issues

• An additional source of contamination is from launch vehicle fairing during ascent

Contamination Concerns

Contamination concerns have driven ourselection of the propulsion systems for NGST.As detailed design progresses, contaminationconcerns will significantly affect materialselection and will require designing in ventpaths and baffles. Our ~30 K optics will be coldtraps for volatile materials to condense on. Ofparticular concern are the effects frompropulsion systems. Some propulsion systems

are very dirty. Others are relatively clean, butproduce by-products such as water which cancondense onto the cold optics. On the followingcharts we present the propulsion trades andexplain how contamination concerns were adriver.

Transfer Propulsion Trades

A number of options were considered as atransfer orbit propulsion system. Such a system,


assuming a lunar flyby with phasing loopstrajectory, requires ~ 100 m/sec DV (includingmargin). Multiple burns are required, extendingover weeks after launch. Due to contaminationconcerns, we considered first using a cold gaswith no contamination concern, such ashydrogen. However, we found that due to thelow ISP and large DV required, it was notpossible to package this system in the allowablevolume. Electric propulsion was considered andrejected, mainly due to low thrust levels that

were not compatible with the mission. Solidswere rejected as too dirty and impractical due torestart requirements. Liquid propulsion wasselected, specifically a dual mode system.Weight of the system is ~170 lb., using availablethrusters. Contamination products are mostlywater. This led us to delay deployment of thetelescope and sun/thermal shields until after thetransfer burns were completed. This wouldallow time for the propulsion system products todisappear.

Transfer Propulsion Trades

• Propulsion system requirements for lunar assist with phasing loops transfer orbit

– ~100 m/sec total ²V

– Phasing maneuver ²V at launch + days

– Mid-course maneuver ²V at lunar flyby + weeks

– NO ²V REQUIRED FOR L2 INSERTION

– Low contamination system required

System Cold Gas Solid Liquid Electric Propulsion

Advantages • No contamination with right gas • Inexpensive • Simple • High thrust • High ISP • Restart capability • High thrust • Very high ISP

Disadvantages • Heavy, very large storage tanks needed • Low ISP, thrust • Serious contamination potential • No restart; multiple engines required • Contamination control must be considered • Very low thrust • High power requirements

Station keeping Propulsion Options

Contamination was the driving concern inselecting the station keeping propulsion systemwhich led us to reject the liquid system used fortransfer orbit. This is unfortunate, as only a fewextra kilograms of fuel would suffice to providestation keeping over the mission life. Theproducts (water, etc.) would likely be majorcontaminants on the cold mirror and othersurfaces. Therefore, only non-contaminatingfuels were considered further.

Cold gas systems are attractive due to theirsimplicity. However, the low ISP means thathundreds of kilograms of H2 would be neededover the mission life. There is not enough

weight margin or volume to accommodate sucha system.

Electric propulsion (resistojets, arcjets, Halleffect thrusters) is attractive, but often entailssignificant cost and requires high power.However, resistojets are a simple electricalsystem with great promise. This technology isflight proven, and TRW has past experiencewith these systems. Resistojets are very small(couple of inches long) and light weight (only~10-20 kg of H2 needed). They use ~ 250 W ofpower each, and have a high ISP. As will beseen in the Space Support Module (spacecraft)discussion, resistojets make a lightweightattractive system. Operationally, due to lowthrust, they would have to burn for hours. This


would probably mean shutting downobservations, but as burns are only needed everyseveral months, this is not an issue.

Note the location of the resistojets on theconcept description chart. Thrusters shouldoperate though the Cg of the space vehicle. Theresistojets are located on the boom at theapproximate Cg of the system.Our early baseline contained a cryostat (forinstrument cooling) of solid Hydrogen.

Interestingly, the amount of H2 needed in thecryostat for a ten year mission is about the sameas needed for station keeping. We expendedsome effort to try to utilize the cryostat boiloffas fuel for the resistojets. Unfortunately, the H2in the cryostat is at very low pressure (<<1 psi),and we could find no practical way to pressurizethis gas to the 10s of psi required. Lack ofsynergy with the resistojets contributed to thedemise of the cryostat.

Stationkeeping Propulsion Options

Options

Mono or Biprop

Cold Gas

Arcjet

Hall Effect

Resistojet

Solar Sailing

Advantages

• Synergistic with transfer orbit propulsion system

• Very simple system• No contamination concerns - H condenses at 5 K - He condenses at <<1 K - N condenses at 30 K

• Can use H or hydrazine• Very high ISP

• Can use H , N , Zenon (inert gasses)• Very high ISP

• Can use H , N , Zenon• High ISP

• Utilizes our sunshade• No contamination

Disadvantages

• Serious contamination concerns

• Requires 100s of kg of gas• Very large tankage required

• Complex system• Requires high power (>1 kW)

• Complex system• Requires high power (~1 kW)

• Requires moderate power (~500 W)

• Tilting increases shade size• Very low thrust. Sufficient?

2

2

2 2

2 2

2

Launch Vehicle Capabilities

This list of current and anticipated expendablelaunch vehicles potentially suitable to the NGSTmission indicates the relative performanceparameters and fairing volume constraints. Theforeign vehicles are listed for completeness andcomparison, and could be of interest should theprogram become an international effort. Thecapabilities of future systems are listed withpublic performance specifications to protectcompetition sensitive contractor actualestimates. The trend of all planned futurevehicles is increased performance at reducedcosts. Fairing dimensions are inside payload

usable volume. Approximate (~) performancesare not based on specific mission estimates butare extrapolated from GTO capability. Allestimates are for optimum inclination for eachlaunch vehicle and launch site. The Atlas IIARand Delta III vehicles currently undercommercial development with contractor fundsand are planned for first flight in 1998.Although details are still considered propriety,both contractors have plans to expand thesevehicles into a family with increased capability.It is reasonable to expect the commercial marketto stimulate substantial performanceimprovements in the medium and heavy classbefore NGST is ready for procurement.


Launch Vehicle Capabilities

C3=-2.3(kg)

2840

~2990

~5220

~2850

~1310

~9400

~3620

~7600

~3670

4100

~5450

3560

Notes:

1) Fairing dimensions are inside payload usable volume.

2) Approximate (~) performances are not established values but are estimated.

3) Performance is for optimum inclination for each launch vehicle site.

4) Atlas is currently doing design trades to develop a 5-meter diameter (outside) fairing for the

Atlas IIAR series and the above. For special unique missions they could change the existing fairing

ring and stringer design to get to a ~3.8-meter inside payload usable diameter.

5) Atlas IIAR series performance is assumimg a 3-foot stretch of the fairing as indicated.

Launch

Vehicle

Atlas IIAS

Atlas II AR

Ariane 5

Delta III

Delta II

EELV Heavy

H IIA (initial)

H IIA (growth)

Long March 3B

Proton D1e

Proton M

Zenith 3 SL

GTO (kg)

3700

3900

6800

3810

1800

>12247

4700

9900

4800

6700

7100

5200

C3=0(kg)

2710

~2850

~4980

2722

~1200

~8970

~3450

~7250

~3500

4100

~5200

3400

Dia (m)

3.65

3.65

4.6

3.75

2.8

4.5

5.1

Unknown

3.65

4.1

Unknown

3.75

Length(m)

4.2

5.1

9.2

4.3

-

12.2

4.9

Unknown

4.7

7.5

Unknown

4.9

Cylinder

Length(m)

9.7

10.6

15.2

8.9

-

Unknown

10

Unknown

6.5

7.6

Unknown

8.5

NGST Trade Trees (2/3)

The following figure provides a roadmapthrough additional trades used to define ourbaseline. Here we concentrate on issues relatedto space vehicle design. Key to concept

development is the realization that we have avery limited volume to package a very largestructure. The launch vehicle constraints andthe thermal considerations drove ourconfiguration.

NGST Trade Tree (2/3) Spectral Band Options

• 0.1-0.5 µm

• 0.5-1.0 µm

• 1.0-5.0 µm

• 1-10 µm

• 1-20 µm

• 1-40 µm

Launch Vehicle Options

• Atlas II AR

• Atlas II AS

• Arianne 5

• Delta II

• Delta III

• EELV Heavy

• H IIA (initial)

• HIIA (growth)

• Long March 3B

• Proton D1e

• Proton M

• Zenith 3 SL

Cold optics, FPAs

Telescope Thermal

Options

• Parasol Shield

• Piggyback Shield

• Payload-on-a-Stick

Optics Packaging Options

• Fold Up/Down

• 'HARD'* Stacking

• Modified 'HARD' Stacking

Secondary Mirror

• Fixed Secondary

• Deployed Secondary

Sun/Thermal Shields

• MLI Shields

• Single Sheet Shields

• Boom Stabilized

• Inflatable Shields

Secondary Support

• Single Strut

• Two Struts

• Three Struts

Optics f# Options

• f 0.9 primary mirror

• f 1.25 primary mirror

Expandability

Desirement

Solar Pressure Compensation

• Momentum dumping by turning SV

• Momentum dumping by tilting shield

• Control tabs

• Symmetric shields

• Electrochromic panels

• Boom twist for anti-propeller*HARD: High Accuracy Reflector Development


Deployable Mirror Concept

The small fairings available in the Atlas classdrove our selection of the deployable mirror.Two general classes were considered, foldableand stackable. The fold up-down is attractive as

it is simple. However, it wastes a great deal offairing volume, limiting the room left for thespacecraft. Packaging studies indicated that wehad insufficient volume left for the spacecraftand instruments. The alternate concept, basedon the TRW developed HARD deployment

Stacked Versus Fold Up-Down Mirror Configuration Designs

• Space vehicle packaging trade hinges on the mirror configuration

• Fold Up-down is potential simpler mechanically, but has limited volume, also has smaller growth potential

• HARD concepts package more efficiently, increasing useable fairing volume

– Has significant growth potential

9601255.009.SA043

83.98

42.50

235.80

Atlas IIAS Fairing

80.00

80.00

66.71

71.58

(399.96)

336.769601255.006.SA043

121.50

110.94

Atlas IIAS Fairing

Fold Up-Down Concept HARD Concept

Mirror Deployment


concept, is much more compact, and leaves thelower part of the fairing free for spacecraft andinstrument packaging. TRW has demonstratedthe HARD concept for large deployable RFantennas. Another very attractive feature of thisconcept is that it is expandable (see next chart).We have baselined the HARD concept.

Support of Secondary Mirror

The support structure for the mirror secondaryhas evolved significantly throughout our study.The first concepts had three fixed struts holdingthe secondary. Unfortunately, due to heightlimitations in the fairing, this required a veryfast (f 0.9) primary mirror, which wasconsidered very difficult to build and toosensitive to mechanical disturbances. Once thedecision was made to have a slower mirror (f1.25), we went to a single deployable boomholding the secondary. Analysis showed thatthe allowable deflections in the secondarylocation were very small (55 µm perpendicularto the optical axis, 300 µm in axis).Dynamically, when the telescope slewed, wewere very concerned that vibration andhysteresis effects would exceed these values.While the secondary has five axis positioncontrol, it is desirable to not have to recollimateafter every slew. Additionally, even with a verylow CTE material, we found that temperaturedifferences had to be kept at ~ 1°F both acrossthe boom diameter and along the boom length.The temperature deltas along the length isconsidered challenging.We considered supporting the secondary betterby placing the boom in tension and adding guywires. Deployment of the wires was difficult,and dynamically not much stability was added.Two struts were considered briefly. They werefound to offer little additional stability.Three struts is the current baseline. Two of thestruts fold out of the way during mirrordeployment and then fold back up to catch the

secondary. This provides a rigid tripodstructure. Thermal considerations remain,which is the primary reason for only moving theelevation gimbal periodically. At a constantgimbal angle, even with the SV tilting 10°, thethermal environment is stable.

HARD Mirror Concept Expandability

The HARD technology allows easy expansion tomuch larger surfaces. With hexagonal petals,two rings of petals can be deployed. The entirestack of petals pivots about one corner of thelast petal deployed and then drops into place.The remaining petals now pivot about the newpetal, continuing the process.

Generic Options for Space Vehicle Design

We examined three generic concepts for theSpace Vehicle design. The first two, parasoland piggyback, have the spacecraft behind thesun/thermal shields. Operating a spacecraft in a~30 K environment is beyond the state of theart. The payload-on-a-stick concept permits thespacecraft to stay warm while the instrumentcompartment and telescope are behind theshade, staying cold. We examined options forthe boom separating the regions. Able has aFASTmast that looks acceptable. The mast iscollapsible into a compact package one foot inheight, and is stored in within a 47" canister inthe spacecraft central cylinder. As it isdeployed, the longerons, diagonals and battenssnap into place. The boom can be constructedof low CTE material such as T300 graphite,resulting in only ~70 milliwatts conduction fromthe Spacecraft to the instrument module.Dynamically, the boom is rigid and stable. Evenafter a slew the boom returns to position veryaccurately - errors between the star trackers(located on the S/C) and the fine guidancesensors (located on the P/L) are ~arcseconds.


HARD Mirror Concept Expandability

Generic Options for Space Vehicle Design

Science

InstrumentsSpacecraft

Advantages :

• Integral spacecraft-payload

• Lightweight shield

Disadvantages:

• Operation of spacecraft at cryo

temp

• Heat sources near SI and OTA

Parasol Shield Payload-on-a-Stick

Advantages:

• Easier spacecraft thermal

• No heat sources near

payload

Disadvantages:

• Complicated dynamics

Piggyback Shield

Advantages:

• Compact design

• Simple structures

Disadvantages:

• Limited FOR

• Spacecraft heat

sources near OTA

and SI

Predicted Temperature Distribution

The following chart presents the NGSTtemperatures with the telescope located at anelevation of 0°. Note that the mirror

temperatures are <30 K, the desired value. Theleft side of the instrument compartment, wherethe IR instrument passive radiator is located, isat 25 K, adequate to cool the FPAs to ~ 30 K fornear infra-red (NIR) imaging.


Predicted Temperature Distribution

• Assumptions

– Instrument Module dissipation of 1.004 W

– Parasitic heat load of 0.1 W

89135 123

79

145

98

9601255.004.SA043

2334

5

25 24

23

260 206

6356

202 193

49

324 322 322 321 321 324

182

105

65

40

260

72

46 Q = 1.1 WQ = 4 MW

Sun/Thermal Shield Design Options

The baseline is a sun shield of two mils silveredTeflon, followed by four shields of 1 mil mylarwith vacuum deposited aluminum on both sides,with an angle of 5° between the shields. Thispermits the cavity between the shields to radiateto deep space.

Deployment of the shields was a major issue.Early versions had inflatable shields. However,we had serious concerns over the additionalweight of the bladders, gas for inflating, andhow to rigidize the structure. Outgassing anddeployment were other issues of concern. Webaselined a strut deployment which would thenpull out the sun and thermal shields.

The size of the shields is sufficient to preventeither the sun- or earth-shine from striking thetelescope and to accommodate a 10° tilt in theentire SV for pointing.

Early versions of our shields had an asymmetricdesign (since the telescope gimbals in only one

direction) to minimize shield size.Unfortunately, we found that the reactionwheels would saturate in ~11 hours due tounbalanced solar pressure. This led to thepresent symmetric shield. Eventually residualtorque will spin up the wheels anyway, somethods of dumping the momentum weredeveloped.

We considered trim tabs on the edge of the sunshields, but it is difficult to keep the telescopefrom seeing the hot tabs. An option that lookspromising is to use panels covered withelectrochromic materials that change reflectivitybased on the voltage applied. This changes theresultant momentum by a factor of ~ two.Issues remain on material selection.

Another effect that must be compensated for isspin momentum buildup. Any mismatch inshield symmetry will cause it to act like apropeller. This could be stopped and the wheelmomentum dumped by twisting the struts tochange the pitch of the propeller.


Sun/Thermal Shield Design Options• Multi-layer insulation was first considered as shields

– Weight of spacers between layers added considerably to mass

• Analysis showed that single sheets with an angular difference between them was as efficient and saved considerable weight

• Inflatable shields were considered and rejected

– Added weight for the inflated portions (double thickness) and inflation gas

– Concern on how to ridigidize the inflated spokes and rims

• Concern that UV hardening required thermal shields that could withstand direct solar heating

• Concern that hardening compounds might outgas

– Concern on rigidity of structure after tilting or twisting

• TRW has demonstrated deployable booms/arrays

– Booms can be easily applied to this task

– Wire rigging can pull out the shields

Other Key Trades

We point the telescope coarsely by moving theelevation gimbal, and then by tilting the SV androtating the entire SV about its axis. Reactionwheels will accomplish this. We examined themoments of inertia of the system and found that

the required 30° slews can be accomplished inwell under the 15 minutes required (30° Az slewin ~8 minutes, 10° El pitch in ~9 minutes).Additionally, we examined the vibration modesof the system and found that the lateral bendingmodes of the spacecraft/mast/payload are about1 Hz. The sunshield modes will likely be lower

Other Key Trades (3/3)

Ephemeris,

Pointing,

Control &

Guidance

Fine Guidance Sensor

Fields of View

Spectral Band Options• 0.1-0.5 µm

• 0.5-1.0 µm

• 1.0-5.0 µm

• 1-10 µm

• 1-20 µm

• 1-40 µm

Fine Guidance Sensor

• Hubble-like Mechanical

• CCD Arrays

• Hubble GSC (14.5 mag)

• New 19th Magnitude GSC

Instrument

Cooling

• Passive

• Cryostat

• Cryocooler


in frequency. These are anticipated to damp outquickly, and any residual motion can beaccommodated by fine pointing mirror in theoptical train of the telescope. Reaction wheelsare biased to spin at 10 Hz or higher.

Fine Guidance Sensor Options

The mission requirements state that pointingmust be stable enough to not increase thediffraction blur by <20%. At 1 µm , thiscorresponds to an AIRY disk diameter of 0.03arcsec, and with 20% jitter, requires a pointingerror of less than 6 milli-arcsec. Since apractical blur centroiding algorithm will providelocation to ~1/5 of a pixel, this leads to a fineguidance sensor pixel size of 30 milli-arcsec.Given a FOV of 2x2 arcmin (see next chart),this results in an array size of 4000 x 4000pixels, an easy value to achieve, with 18thmagnitude, adequate signal-to-noise ratio existsto permit centroiding.

Three Fine Guidance Sensor options have beenconsidered. One is like Hubble, which used alarge field of regard field of regard but fewpixels. Hubble used a mechanical arm to movea very small field of view within the. Given the

limited Field of regard that we need at 18thmagnitude, and given that we can readily buyenough pixels to cover this field of view, werejected the Hubble concept. Separate guide

Fine Guidance System SensorRequirements

• Mission Requirements– Diffraction limited optics at 1 µm (1.2 l/d)– Pointing stability �20% of diffraction blur

• Pointing System Requirement– Diffraction blur: 1.2 x 1e-6/8 = 0.15 µrad = 0.03

arcsec– Allowable jitter/drift: 6 milli-arcsec

• With adequate SNR, can use centroiding to locate astar

– ~1/4 to 1/10 of the FGS pixel size– Assuming 1/5 => FGS pixel size is ~30 milli-

arcsec• Given FOV requirement of 2 x 2 arcmin (see previous

chart): 4000 x 4000 pixels required• Adequate SNR exists

– Flux from 18th magnitude star: 58,000photons/sec (8 meter telescope)

– Image blurred to cover ~4 pixels: 14,500photons/sec/pixel

– SNR (1 sec): ~sqrt (14,500) = ~120– SNR (0.1 sec): ~sqrt (1450) = ~40– (Note: Quantum efficiency of pixels assumed to

be ~1)

Fine Guidance Sensor OptionsHubble-Like

• Use outer edges of main telescope FOV

• Operate in visible

• Use mechanical pickoff mirrors to locate guide stars

• Relay starlight to an interferometer

• FOR of FGS magnitude dependent (see following chart)

Large Arrays

• Use outer edges of main telescope FOV

• Operate in visible with 19th magnitude guide star catalogue

• Pave a sufficiently large area with FPAs such that high probability that star is in FOR

- ~3x3 arcmin FOV

Separate Guide Telescopes

• Two 45 cm Cassegrain visible light telescopes

• 4 x 372 x 372 arcsec FOV (~150 arcmin 2 )

• 14.5 magnitude guide star catalogue required

• Located at right angles to each other and to the main telescope axis

Main FOV

FGS FOV

69 arcmin2 FOR

5"x5" mechanical

pickoff mirror

Main FOV

Main Mirror

FGS Telescopes


telescopes were considered and sized. Werejected this concept based on limited volume inthe fairing and the potential for misalignmentsbetween telescopes. Instead, the Large Arrayconcept uses the existing main telescope field ofview.

Multidiscipline Design Optimization.

Our paper describes the process used in theaerospace industry to develop design conceptsfor space science missions, using our NextGeneration Space Telescope FeasibilityAssessment Study [1] as an example.

The process begins with articulation of the needfor a mission, a definition of its objectives andan estimate of the funding which is available.For NGST, the need and objectives wereprovided by the report of the "HST and Beyond"committee [2], while the funding level wasdetermined by the savings which NASA couldachieve by discontinuing HST maintenanceactivities after the 2003 servicing mission.

A mission concept and spacecraft design arethen developed by a multi-disciplinary teamorganized by function or spacecraft element intoIntegrated Product Teams. These teams identifydesign options which meet the missionobjectives, and select the most promisingalternatives through a series of trades andanalyses. Their selection criteria include systemperformance as well as cost and risk. If nodesign solution is found, the requirements aremodified and new technologies [3] areintroduced until an "optimum design" isachieved.

Design optimization is an iterative process, withmore detailed designs and analyses generatedduring each iteration. For the NGST CAN studywe used relatively simple thermal, dynamic andoptical models to assess system performanceand utilized existing spacecraft designswherever possible. Much more detailed modelsand designs were generated during our current

mission architecture study, including anintegrated model to assess the end-to-end opticalperformance of our baseline design in thedynamic and thermal environment predicted forNGST.

The use of highly integrated system performancemodels for design optimization is a new trendin spacecraft design, made practical by recentadvances in computer technology. Ideally,integrated models can be used to determine thesensitivity of our design to key parameters andfind an optimum configuration. To date,however, high fidelity simulations are bestobtained by linking existing stand-alone"industrial-strength" software tools with specialpurpose "translators". And building a detailedsystem performance model is a labor-intensiveprocess which can only begin when a detaileddesign of the spacecraft is available.

Low fidelity integrated models using linkedspreadsheets running on PC's are now beingused by integrated design teams at manyaerospace companies and governmentlaboratories. It is important to have a welldeveloped set of mission requirements and welldefined mission concept before going to adesign center; however; since a typical designeffort a week of effort by 10-12 highly skilledengineers and scientists.

The use of multidicipline design teams is apowerful tool for exploring the design space tofind a optimum solution, since experts in all ofthe relevant areas are readily available. Theymust be used judiciously, however; to controlcosts. Powerful analytic tools are available fordesign optimization, but more work needs to bedone to link them together. Before a designcenters or integrated models can be usedeffectively, much effort must be expended torefine the mission requirements and "zero-in" ona feasible mission concept and baseline design.


Summary

The objective of this paper was to look at thespacecraft design process and see how thatprocess balances desired spacecraft featureswithin an imposed set of operational and costconstraints. The constraints often show up ascompeting multidiscipline interactions, which intheir resolution lead to practical spacecraftdesigns. This paper gives examples of how thedesign process was implemented in a feasibilitydesign study for NASA's proposed NextGeneration Space Telescope (NGST), anddescribes how the project organization was usedto effectively deal with multidiscipline design.Orbit selection, spacecraft propulsion, stationkeeping, and overall mechanical and thermalsubsystem designs were discussed as examplesof multidisciplinary design optimization. Thefinal section discusses multidiscipline designoptimization, what its benefits are, what are thenegative points and what can be done toimprove the process.

Acknowledgments

This work was supported by the NationalAeronautics and Space Administration underCooperative Agreement No. NCC5-137.

References

1.TRW-Led Next Generation Space TelescopeFeasibility Assessment Study Results, ed. C.F.Lillie, TRW, 1996.

2. Exploration and the Search for Origins: AVision for Ultraviolet-Optical-Infrared SpaceAstronomy, Report of the "HST and Beyond"Committee, ed. A. Dressler, AURA, 1996.

3. The Next Generation Space Telescope, ed.P.Y. Bely, C.J. Burrows, and G.D.Illingworth, STScI, 1989.

4. Visiting a time When Galaxies Were Young, ed. P.Stockman, STScI, 1997.

AIAA 98-4704


MULTIDISCIPLINARY DESIGN PRACTICES FROM THE F-16 AGILE FALCON*

Michael H. Love†

Lockheed Martin Tactical Aircraft Systems

* Copyright 1998 by Lockheed Martin Corporation.Published by the American Institute of Aeronauticsand Astronautics, Inc. with permission.† Engineering Specialist Senior, AIAA Senior Member

ABSTRACT

An advanced version of the F-16 called the AgileFalcon was studied and a preliminary design wasdeveloped in the late 1980’s. Multidisciplinarydesign issues were addressed through trade-offsat the conceptual and preliminary design levels.Trade studies and associated approaches from aperspective of how they effected the course ofthe design process are discusssed. The interestof the Agile Falcon was directed at a balance ofmultirole capability. The results of the studiesfocused the airframe toward an F-16 typetrapezoidal wing. Ensuing studies involvedoptimization of the wing to maximize themultirole capacity while constraining/minimizingimpact to existing hardware. The redesign of thewing touched all aspects of the airframe andsubsystems.

INTRODUCTION

In the early 1980’s General Dynamics FortWorth Division (now Lockheed Martin TacticalAircraft Systems) conducted studies toinvestigate the incorporation of advancedtechnologies into an F-16 with a larger wing.The interest was directed at maintaining abalance of multirole capability. The results ofthe studies focused the F-16 variant, called AgileFalcon, toward an F-16 type trapezoidal wing.Ensuing studies involved optimization of thewing to maximize the airplanes multirolecapacity while constraining/minimizing impact tothe fuselage and empennage. The redesign of thewing however touched all aspects of the airframeand subsystems.

Multidisciplinary, multi-objective design issuesdrive aircraft design. For example, the AgileFalcon program was focused to enhance the F-16’s current state of agility. The agility measureincludes multi-objectives of maneuverability andcontrollability. Difficulties in design decisionsarise from the uncertainties of what one might

categorize as the weighting factors of a system-level, multi-objective function. In other words,priorities of the multiple objectives in a systemdesign are usually not clear. The Agile Falconprogram1 attempted to address these issues in asystematic approach in the predevelopment stageprior to full scale development. Figure 1 depictsthe Agile Falcon at the end of its predevelopmentphase in 1989.

Figure 1 Agile Falcon At Completion ofPredevelopment Program

Methods used in the data development to supportthe Agile program have since evolved. Forexample, a combination of computational fluiddynamics analyses (CFD) and wind tunnel testingwould be used in lieu of extensive wind tunneltesting for performance and stability and controldata acquisition. In this paper, methods andprocesses used in the Agile program areexamined and compared to those that might beused if the Agile Falcon were being developedtoday.

Agile Falcon Objective

The F-16 was born in the 1970’s from the lightweight fighter program. Over the last 20 years ithas provided the Air Force both air-to-air and

AIAA 98-4704


Figure 2 Characteristics of Agilityair-to-ground combat capability. Its light weightand efficient aerodynamic design have providedoutstanding agility characteristics. Advancedversions of the F-16, however, are less agile thanits earlier versions. Increased capabilities inareas such as pilot awareness have led toincreased vehicle weight. Studies were initiatedin the 1980’s to regain F-16A agility.

Many papers in the 1980’s discussed the topic ofagility2,3 In reference 1, agility in a fighteraircraft sense was defined as “performanceneeded to win and survive close-in combat.”Furthermore, maneuverability and controllabilityas they are related to agility are discussed asshown in Figure 2. “Maneuverability is thequality that changes the flight path vector of anaircraft. It results from the sum of forces (lift,weight, thrust, and drag) that cause a change inthe speed and direction of the flight path.Controllability is the ability to guide flight pathchanges.” Maneuverability leads to suchmeasures as turn rate, acceleration anddeceleration. Controllability leads to suchmeasures as rates and accelerations of aircraftstates.

Together, controllability and maneuverability ina fighter aircraft allow its pilot to win “dog fight”encounters with opposing aircraft. A pilot will

call on the aircraft, for example, to turn,accelerate, turn again, decelerate, fire a missile,and accelerate suddenly to gain the advantage onanother aircraft and win a “multi-bogey”engagement.

In order to address maneuverability andcontrollability, the Agile Falcon program focusedon the development of an advanced wing andwing/strake/fuselage integration. Trade studieswere performed to develop informationmeasuring agility as defined throughcontrollability and maneuverability metrics asrelated to geometric variations of the wing andwing/strake/fuselage integration.

DEVELOPMENT APPROACH

A predevelopment program was executed toimprove turning performance, increase the AOAcapability, maintain adequate controllability inthe roll axis throughout the AOA envelope, andminimize impact to existing systems on the F-16.The turning performance and AOA capability areconsistent ingredients to maneuverability.Controllability in the roll axis was emphasized athigh AOA to allow sudden changes in flightpaths while allowing maximum maneuverability.All existing systems on the F-16 were evaluatedto constrain/minimize cost impact fromwing/strake/fuselage modification.

AIAA 98-4704


Two of the airframe studies will be used toillustrate how agility was addressed during theAgile predevelopment phase. One study involvesthe overall synthesis of a baselinewing/strake/fuselage configuration; the secondillustration encompasses development of thewing design within the context of the baselineconfiguration.

Baseline Wing/Strake/Fuselage Configuration

Prior to the predevelopment program, conceptsizing studies were performed to define aneighborhood for potential Agile Falconsolutions. These studies included traditionalparametric databases for weight, costs, andaerodynamics. These databases were founded onthe F-16 and provided stable measure forsensitivity studies. The study-results led toselection of a matrix of wings and strakes tobuild a more accurate parametric space andprovide refinement to a baselinewing/strake/fuselage configuration. The selectedconfigurations are depicted in Table 1 and Figure3.

This matrix of configurations included 3 strakesin combination with 7 wings. The Baseline wingwas derived during the aforementioned synthesisstudy. Data was developed for theseconfigurations with regard to agilitycharacteristics and structural integration.

The agility characteristics were studied throughthe combination of wind tunnel tests followed byanalyses. The structural integration studiesincluded airframe layout studies combined withpreliminary level aeroelastic synthesisevaluation. The data developed in these studieswas combined in a qualitative evaluation.

Table 1 - Candidate Wing ConfigurationsConfig.

# 1# 2# 3# 4# 5# 6

Baseline

Span (ft)

37.5035.0735.0735.0733.5437.5037.50

Area (sq ft)

375375410328375375375

Sweep

40.0°37.5°37.5°37.5°37.5°37.5°34.3°

Span Trade

Area Trade

Sweep Trade

Figure 3 Three Planform Trades

Agility CharacteristicsFigure 4 presents the flow of wind tunnel testsand analyses performed in the matrix study. Twoseries of tests were performed to provide ascreening process for the later more expensivetransonic tests. The configurations tested were“full-up” F-16-like models (1/9th scale). As seenin the figure, the first set of tests concentrated onan understanding of characteristics in extendedregions of AOA where basic lateral directionalstability and

AIAA 98-4704


7 WINGSx

3 STRAKES

Aeroanalysis Aeroanalysis

Stability andControl

Stability andControl

CGAOA

CGAOA

• Polar Shape

• Trim CLMAX

• Drag CLMAX

• Flap Sched.

• Trim CLMAX

• Drag Improv

• Aft CG Limit

• AOA Limits

• Lat/Dir Stab.

• Aft CG Limit

• AOA Limits

• Lat/Dir Stab.

Wing/StrakeSelection

ConfigurationSelection

• Compromise

AOA & Polar

• Mid & High

AOA Polars

• Good High

AOA Stab.

Low Speed Low Speed & Transonic

• Polar

Improvement

• Highest

CLMAX

• Good High

AOA Stab.

Figure 4 Aerodynamic and Stability & Control Screening Process

CLMAX could be evaluated. These parameters arekey in the maneuverability and controllabilityarea. A Taguchi experiment was performedduring the wind tunnel testing to reduce follow-on low speed and transonic testing. From theresults of the first tests, a down-select to onestrake/4 wings was made for the follow-oncombined low speed and transonic testing.

In the testing, no one configuration providedsuperior performance in combined CLMAX andstability in extended AOA. However a cluster of3 configurations seemed to be the bestperformers: 2, 3, and 6. General conclusionsfrom a stability and control viewpoint included(1) minimizing span and (2) moving the wing aftfor balance. Conclusions from anaero/performance viewpoint included increasingL/D withspan and increasing CLMAX with area. Bothdisciplines also recommended continued tailoringof the wing/strake area as key.

Structural IntegrationStructural evaluation of this matrix involvedquantitative and qualitative studies. Structuralsizing issues needed to be evaluated as well assystem integration issues. Benefits from anyaerodynamic configuration selected should not

be impeded by structural weight increases, wingdeformation characteristics, or system changes.

The Wing Aeroelastic Synthesis Procedure,TSO, was used to evaluate structural sizingissues4,5. All of the parametric variations in wingspan, wing area. and wing sweep were studied.Design optimization was performed in each casefor a variety of objective and constraintfunctions. In addition to the planform variations,a study of wing t/c and material properties wasincluded.

Typical optimization results for varying conceptsof aeroelastic tailoring are shown in Figure 5.The wing box skins were designed in eachconfiguration for three different designgoals/concepts. A minimum weight “StrengthSized” design was achieved with three aircraftsimulated maneuvers (two symmetrical pull-upsand one asymmetric rolling pullout). In thesecond concept, a flutter requirement and anaileron roll control effectiveness requirementwere added to the strength requirements(“Aeroelastic Sized”). The third concept addedan aeroelastic twist requirement to the strengthand aeroelastic requirements. The aeroelastictwist provides lift-to-drag efficiency at thesimulated turn maneuver point.

AIAA 98-4704


33 34 35 36 37 38400

600

800

1200

WING SPAN (FT)

A/C

SK

IN W

T (

LB

S)

1000

AEROELASTIC SIZED

DRAG SIZED

STRENGTH SIZED

LIF

T-I

ND

UC

ED

DR

AG

CO

EF

F.

33 34 35 36 37 38950

1000

1050

1100

1150

0.08

0.09

0.10

0.11

WING SPAN (FT)

AIR

CR

AF

T S

KIN

WT

(L

BS

)

DRAGSKIN WT

Figure 5 Optimization Study ExaminedWeight, Design Concepts, and Performance

The top part of Figure 5 displays the sensitivityof the wing box skin weight with respect to theconcepts and span. The span study (shownabove) provided the greatest sensitivity while thesweep (not shown) provided the least. Thebottom part of Figure 5 provides the sensitivityof the aeroelastic drag to the wing skin weight forthe “Drag Sized” concept. Interestingly, the areastudy (not shown) indicated that as areaincreased, the weight decreased to a point beforebeginning to increase. This observation wasrationalized by the increase of wing depth for afixed t/c allowing for gains in structuralefficiency up to a point. Therefore,Configurations 2 and 3 provided interest forfurther study.

Airframe layout studies were performed toexamine system interface issues. Considered inthese studies were landing gear placement,engine and engine accessories placement,interface of fuselage-based wing control surfaceactuation subsystems, interface of wing/fuselagefuel systems, and wing/fuselage interface loads.Structural arrangements studies involvedplacement of wing spars and ribs as well asfuselage carry-thru bulkheads. Qualitativeassessments were made with regard to ease ofintegration. Configurations 2 and 3 were thehighest ranked.

Quantitative assessments were made in terms ofmass properties estimation for eachconfiguration. Although these estimations wereparametrically based, the aforementioned TSOstudies (a subset of the estimates) substantiatedthe findings. The results were provided tovarious analysis groups to evaluate performanceand stability.

Selection of New Baseline ConfigurationDerivation of a new baseline from thisinformation was performed through a qualitativeanalysis. Stability and Control considerations ledto the conclusion that the baseline span of 37.5feet needed to be reduced. Aerodynamicperformance considerations lead to theconclusion that although increased span over theF-16 provides substantial improvements in L/D,increased span with increased area might provideenhanced stability with no degradation in L/D.Figure 6 illustrates this in showing the sensitivityof CLMAX and CD at CLMAX. The increased areaprovides for the aft shift of aerodynamic centerfor stability considerations while allowing theincrease span for L/D

0.6

0.7

0.8

0.9

1

1.1

1.2

1.3

1.4

300 328 375 410 440

Wing Area (Sq. Ft.)

CD

@ C

L MA

X

1.4

1.5

1.6

1.7

1.8

1.9

2

2.1

2.2

300 328 375 410 440

Wing Area (Sq. Ft.)

CL

MA

X

Figure 6 Sensitivity of L/D @ CLMAX With Wing Area

AIAA 98-4704


performance. Figure 7 shows data from the TSOstudy indicating that an increase in area withfixed t/c could offset an increase in span in termsof structural weight.

475

525

575

625

675

725

AIR

CR

AF

T S

KIN

WT

(L

BS

)

SPAN WT (LBS)AREA WT (LBS)SWEEP WT (LBS)

SPAN (FT)

AREA (SQ.FT.)

SWEEP (DEG.)

33.5 35.0 37.5

328 375 410

34.3 37.5 40

Figure 7 Weight for Area Increase at FixedSpan Offsets Weight for Span at Fixed Area

Increase

These pieces of information coupled with thesystem interface studies led to a newconfiguration baseline. The process fordetermining a new configuration involved anintegrated product team approach considering thepositives, negatives and sensitivities of theaforementioned studies. While the systemintegration studies are not shown, they providedindications to ease of design and manufacturingassembly, as well as costs of the Agile Falcon.The new configuration was determined through acombined selection of wing-span, wing-area,wing-sweep, strake, and wing-placement withrespect to the fuselage.

Impact of Design Technologies On ApproachWhile the study to establish a refined baselineinvolved development of multidisciplinarysensitivities, the number of data pointsestablished was few; and the ability to establishan accurate parametric connection of the data toagility was not there. For example, each of thewings studied in the matrix allowed forintegrated computation of turn rate performance,which involves L/D, CLMAX, and airframe weight.There was not enough time or information,however, to integrate controllability measures.Time history maneuvers would have allowed

characterization of the vehicle’s full agility.Finally, three points in sweep, three in span, andthree in area as considered in the wing matrixstudy, allowed characterization of a second ordercurve of information. However, there wasalways question on information distant from anyof these points.

With the current capability of computationalfluid dynamics, enough wing/strake/fuselagecombinations could be evaluated andtransformed into response surfaces to allowconsideration of a design space, rather than asampling of the space. Similarly, the abilityto develop structural finite element models andperform ASTROS-like design optimizationstudies6,7 would allow structural evaluation on afiner level. Response surface techniques lend todesign of experiment approaches8. Given suchmethods, syntheses can be performed that allowexamination of many configurationsapproximated through the response surface.Agility metrics involving controllability andmaneuverability could be evaluated and factoredinto battle scenarios.

Parametric modeling of aerodynamics andstructural configurations is imminent. Design ofexperiment approaches may occur in anautomated fashion in the future. A missing linkis the development of techniques for control lawmodeling to allow parametric time historyevaluations in rapid fashion.

In the case of an active aeroelastic wing,redundant controllers can be used withaugmented control objectives where forceimbalance constraints are combined withmaneuver load control metrics to achieve controlsurface gearing per maneuver9. A genericcontrol approach10 may also provide initialthrough-put for DOE in the design of airframe.The issue lies in the overall vehicle synthesis, itsmission scenarios and overall vehicle class (e.g.subsonic attack vs. supersonic air superiority).Often the metric for design is not clear. Thequestion to be answered is how a vehicle will beused and how it will respond in a combatenvironment. To perform such simulation,integrated measures of agility are required.Response surfaces can allow rapid evaluations ofinputs to the agility measures such as turn rateover a wide range of geometric variables.

AIAA 98-4704


Much of the data developed with regard tosystem integration was qualitative, requiring“man in the loop” to evaluate the manypossibilities. Genetic algorithms combined withobject oriented modeling languages may serve toautomate systems integration. Object orientedapproaches to conceptual design are beingexplored11,12.

Not presented here is any approach to bringaffordability into the decision process. Thismetric is a function of many discrete decisionsthat are linked to materials and manufacturingprocesses. Historically, we have relied onweight-based cost. It is conceivable that objectoriented approaches may enable rapid evaluationof activity based costs as functions of geometricparameters and inclusion of such data as anindependent variable.

Wing Design

The wing design integrated three studies towardenhanced agility for the Agile Falcon. Anaerodynamic performance study focused on thedevelopment of the wing twist and camberdistribution for maximum maneuverability. Theobjective of the study centered on a balance inhigh-g turn objectives and 1g accelerationobjectives. Controllability studies focused ondefinition of the control suite of the wing tosatisfy low speed (high AOA) and high speed(structural flexibility) handling qualities. Anoutboard aileron was considered in addition tothe F-16 baseline flaperon (inboard trailing edgesurface). The structural studies included anassessment of aeroelastic tailoring strategies thatwould best complement the maneuverability andcontrollability initiatives. The evaluation criteriafor the three studies consisted of measurement in(1) turn rate, (2) roll performance, (3) structuralweight (wing and fuselage), (4) impact tofuselage structure and fuselage based systems,and (5) airframe producibility. The presentdiscussion of the wing design is presented fromthe bias of the structural studies and where theyinterfaced with the aero/performance andstability and control studies.

Focus on Structural StudiesThe baseline material for the Agile Falcon wingskin was advanced graphite composites.Extensive material trades were performed.

Within these trades was a study of aeroelastictailoring. Three concepts were derived: (1)Washout - minimum weight including aconstitutive tendency of the wing to twistnegatively with positive bending; (2) Washin -minimum weight including a constitutivetendency of the wing to twist positively withpositive bending; (3) Strength - minimum weightwith the requirement that the wing only meetgeneral strength integrity. The objective of thestudy was to find the aeroelastic tailoring conceptmost suited to benefit the aero/performance andcontrollability studies. Included in the sizingwere detailed requirements such as buckling,bolted joints, fuel pressure, and wing skinproducibility.

The process of sizing and evaluation is shown inFigure 8. The TSO program13 had been used inthe context of internal loads development forover ten years at the point of this application.Interface tools were developed to allow themapping of TSO results to a finite elementmodel. The MODGEN program was tailored tothe quick development of wing finite elementmodels. The process of a TSO skin developmentstudy and a wing finite element model at thistime was approximately an eighty hour task. Thewing model was attached to a stick fuselagerepresentation allowing fast evaluation of flexibleaerodynamics in the FLEXLODS code14. Acritical loads study was performed for some timeon the Agile program, so therefore, identificationand mapping of a critical loads case simplyinvolved derivation of the aeroelastic tailoringconcept and associated aeroelastic increments tothe model. Each concept was then uniquely sizedwith an in-house tool known the CompositePanel Analysis Package (CPAP). A new set ofaeroelastic analyses were conducted for theresized concepts. Aeroelastic deformation datawas provided to the aero/performance groupallowing integration of aeroelastic increments tothe drag polars developed for candidate rigidwing distribution shapes (rigid camber and twist).Flexible-to-Rigid ratios were provided to thestability and controls group and applied in a 6-DOF simulation.

The aeroelastic tailoring concepts were selectedfor detailed study for various reasons. TheWashout concept

AIAA 98-4704


TSO

Skin Optimization

CONCEPT

MODGEN

Internal Arrangement

NASTRAN

Stiffness Generation

FLEXLODS

Flexible Aerodynamics

LODSUM

Critical Loads Selection

NASTRAN / CPAP

Internal Loads & Sizing

CONTROLLABILITY

6-DOF Simulations

MANEUVERABILITY

High g Turns, 1g Accel

DEVELOPMENT

SYSTEM INTEGRATION

Weight & Producibility

EVALUATION

Figure 8 Aeroelastic Tailoring Concepts Were Systematically Evaluated

demonstrated, in the Validation of AeroelasticTailoring program through wind tunnel tests, a23% reduction in lift-induced drag over rigidaerodynamics. The Washin concept is noted forits propensity to maximize lift and controlsurface effectiveness. The Strength conceptallows for minimum weight and presumes thatenough control effectiveness is available throughredundancy. Each concept has valid benefits.

The ranking of the concept results in the study ispresented in Table 2. The Washout conceptprovides the best overall performance to thedesign metrics.

Table 2 Ranking of Aeroelastic TailoringConcept Results

Concept Maneuver Control Weight Produc-ibility

Washout 1 1 2 2Washin 2 2 3 1Strength 1 2 1 3

In the Maneuverability category, analyses wereperformed for loiter, maneuver, and acceleration.Wing deformation information was provided in asemi-empirical, linear superposition code thatwas tuned to rigid wind-tunnel data. Therefore,analysis credit was acquired for aeroelasticincrements. The distinguishing characteristicsinvolved the negative bend/twist coupling of theWashout concept, allowing minimum jig-shapecamber and twist. The Washout concept thenexcelled in sustained turn rate and acceleration.

The distinguishing feature of the Washout wingin the Controllability metric is its relief of rolldamping while retaining roll control. The rollcontrol of the Washin and Washout iscomparable. The damping behavior of theWashout and Strength concepts is comparable.Figure 9 illustrates the difference in roll rates forthe three concepts in 1-DOF simulation. Thedata are normalized to the Washout concept.The project also compared the Washout, Washin,and Strength concepts in 6-DOF simulations.Other 6-DOF simulations were performed forconfigurations with outboard aileron combinedwith the inboard flaperon. These controllabilitystudies were performed at high speed / highdynamic pressure, and the results wereconsidered in combination with low speedhandling quality studies where wing flexibility isnot the issue. At the time the Agile Falconprogram was canceled, the baseline configurationconsisted of a single inboard flaperon with theWashout concept.

The weight metric includes the wing weight andimpact to fuselage weight. Considering wingweight alone, the Washout concept is theheaviest. However, due to the load relief anddistribution of load at the wing/fuselageinterface, the Washout concept surpasses theWashin concept in minimum weight. The loadrelief and load distribution of the Strengthconcept is similar to that of the Washout conceptand is the lightest weight concept to begin with.

AIAA 98-4704


0.5

0.6

0.7

0.8

0.9

1

1.1

0 10 20 30 40 50

Altitude (1000 ft)

Ro

ll R

ate

Fra

ctio

n o

f W

ash

ou

t

Washout

Washin

Strength

0.5

0.6

0.7

0.8

0.9

1

1.1

0 10 20 30 40 50

Altitude (1000 ft)

Ro

ll R

ate

Fra

ctio

n o

f W

ash

ou

t

Washout

Washin

Strength

0.9 Mach 1.2 Mach

Figure 9 Roll Performance of Aeroelastic Tailoring Candidates

Producibility is measured by the gradient ofthickness changes per orientation over the entirewing skin. For manufacturing, the wing skinneeds to be dividable into areas or zones ofconstant thickness per orientation. The Strengthconcept was derived from the gradient-basedoptimization of TSO, and it was the mostcomplicated laminate wing skin definition. Whilethe Strength concept was developed throughdesign optimization, a more structured approachof “pre-zoning” might be taken to improve itsproducibility. The same might be said for theWashin and Washout approaches. The Washindesign has the fewest number of zones becauseits percentage of thicknesses per orientationremains approximately constant throughout thewing skin. The Washout concept could bebroken into a producible number of constantpercentage zones and overall thicknesses. TheStrength concept, as it was derived, wouldrequire a large number of constant percentagezones.

Impact of Design Technologies On DesignApproachLike the vehicle synthesis phase of the AgileFalcon program, the approach to achieveintegration would probably be the same today asin 1988-89. The differences in the overallprocess would be in the tool selection fordeveloping the data and the amount of datagenerated to perform the needed evaluations.

Recent directions in development of ASTROS15

and NASTRAN16 allow that there is little needfor TSO in this phase of design. Designoptimization with nonlinear aerodynamics (suchas CFD-based pressures) is becoming a reality.However, the aeroelastic increments would still

be computed with linear aerodynamic influencecoefficients. Codes such as ISMD from BoeingNorth American9 even make it possible toconsider the aerodynamic design of wing camberin the structural design process. Thecomputation of accurate lift induced drag iscomplex, however, and the trends at best are theonly thing believable. A design-of-experimentsapproach could be used with a modal- baseddesign optimization17 to arrive at optimal camberand robust structural design18. In addition toderiving optimal camber, ASTROS and ISMDcould be used in an active aeroelastic wingapproach to evaluate interaction with controllaws with redundant control effectors.

In the Agile Falcon approach, only the wingstructure was sized per concept. The loaddistribution at the wing/fuselage interface wasconsidered qualitatively in a weight measure forthe wing skin concepts. However, the truemeasure is in the sizing of the fuselage structure.The wing is a very small percentage of the basicdesign flight gross weight. Saving weight isimportant, but the center fuselage is denselypacked with systems and loads. It is important tobe able to quantify the benefits of redistributingloads across the fuselage, which aeroelastictailoring accommodates. Today’s technologyallows for this.

Maneuverability evaluations could be developedtoday in the CFD realm with aeroelasticdeformations superpositioned on the rigidgeometry and the trim state provided at 6-DOFtrim conditions to create a “rigid” CFDconfiguration for analysis. These shapes couldbe used for CFD-based drag computations. Ofcourse, the test-anchored linear superposition

AIAA 98-4704


approach could be used again. There stillappears to be no tool that can adequately handlean iterative CFD-based nonlinear aeroelasticsolution for a full aircraft configuration, althoughmany are pursuing such a tool.

The process for Controllability studies would belittle different today. The time to achieve thisanalysis would be shortened, and the number orconditions evaluated would be greater. Designsin the near future might aggressively pursue anactive aeroelastic wing approach, which wouldnecessitate a tight connection between thestructural design and the control law design. Inother words, the robustness of the control systemwould depend on the robustness of thestructure18, since an active aeroelastic wingapproach consists of a “strength” concept forcomposite tailoring. The design of the structureis tightly coupled to the assumptions of thecontrol laws. There is currently little feedback ofrequirements from the control law group untilafter the structure is designed. Today,conservative assumptions are made to ensure thestructure covers all reasonable usage of controleffectors in the development of loads.Minimizing loads and minimizing structuralweight drives the control laws to a tentative state.A key area of technology development is aprocess and tools for performingcontrols/structures feedback early in the designprocess that allows the designer to focus onrobustness issues.

Affordability is the metric of the day, and ittypically factors in producibility. ASTROS andNASTRAN have design variable definitionoptions that allow the user to maintain control ofthickness gradients over the topology duringdesign optimization. The design results wouldthen be mapped to electronic CAD datasets forfurther evaluation. Tools such as PICASSO19

were developed during the Agile Falcon era tobegin to address these issues. PICASSO mapszones of constant percentages and thicknessesinto composite ply tables that interface from zoneto zone. This tool allows the rapid deploymentof tailored laminates to producibility evaluationtools. In addition, a study today would includemapping the manufacturing data back on theinternal loads model for an analysis iterationprior to sizing convergence.

As we look further to the future, parametric andassociativity concepts will allow us to considermore items simultaneously in the design study.Structural arrangement versus system integrationmay play greatly into the structural weightcomputations. As was mentioned, in the AgileFalcon approach, the load distribution at thewing/fuselage interface was consideredqualitatively in a weight measure for the wingskin concepts. In the future, resizing of thefuselage structure could be considered forvarious structural arrangements thataccommodate subsystems in the overallconfiguration.

SUMMARY / CONCLUSIONS

The Agile Falcon program was a programfocused on multidisciplinary design optimization.The objective was to maximize the agility of theF-16C while minimizing cost to do so. Theobjective was decomposed into developing adesign focused on enhancing maneuverabilityand controllability while minimizing impacts onaircraft weight and subsystems.

This paper examined two central studiesperformed in the course of the program; (1)refinement of a wing/strake/fuselageconfiguration, and (2) development of the wingdesign including structures definition,aerodynamic jig shape, and selection of thecontrol effector suite. These studies requiredcoordinated efforts to bring data together at keydecision points. Decisions were made in theconfiguration development on the basis ofquantitative and qualitative assessments. Noformal recomposition of the design metrics wasperformed to evaluate whether an optimum wasachieved. However, it was determined that theproduct concept was improved at the completionof the predevelopment program.

If the design were being performed today, theemphasis on higher resolution would drive thenumber of data points considered.Computational capacity continues to grow interms of accuracy and turn-around. Tighterintegration is evident in many areas, allowingcloser evaluation of multidisciplinary couplings.However, it seems that to truly usemultidisciplinary design, a system levelevaluation must be maintained to recompose sub-level studies into system level payoffs.

AIAA 98-4704


Although it is obvious, one would be remiss tonot make a statement on the importance ofculture. The nature of the Agile Falcon and thepersonalities involved allowed the programapproach. Integrated design is a conscious effortof tasking processes to develop essentialknowledge allowing strategic decisions thataccount for all design requirements. It is missiondependent. For instance, a design more prone toflutter requires more flutter analyses during thecourse of design. It relies on trade studies. TheLMTAS integrated philosophy is to ensure thatessential requirements are considered during thetrade study process. The strength of LMTASintegration is derived historically from thecoordination skills of our Design function20.New design technologies may well redefine“Design,” but they will not be accepted until theculture accepts them.

ACKNOWLEDGEMENT

The efforts reported in this paper were supportedunder Air Force Contract Number F33657-84-C-0247 CCP 4563 titled F-16 Agile Falcon / MLU.The author gratefully acknowledges the supportof the U.S. Air Force.

REFERENCES

1) Franks, J.M., Timpson, K.G., F-16 AgileFalcon / MLU Final Report, Volume I,Airframe / Subsystems Studies, Air ForceSystems Command, F-16 Systems ProgramOffice, 90PR064, 11 December 1989.

2) McAtee, T.P., “Agility - Its Nature and Needin the 1990’s,” presented at the Society ofExperimental Test Pilots Symposium,September 1987.

3) Hodgkinson, J., Skow, A. et al,“Relationships Between Flying Qualities,Transient Agility, and OperationalEffectiveness of Fighter Aircraft,” AIAAPaper 88-4329.

4) Lynch, R.W., Rogers, W.A., and Braymen,W.A., “An Integrated Capability for thePreliminary Design of AeroelasticallyTailored Wings,” AIAA Paper No. 76-912,Aircraft Systems and TechnologyConference, Dallas, Texas, September 1976.

5) Love, M.H., Bohlmann, J.D, “AeroelasticTailoring in Vehicle Design Synthesis,”presented at the

AIAA/ASME/ASCE/AHS/ASC 32ndStructures, Structural Dynamics, andMaterials Conference, April, 1991.

6) Love, M.H., Barker, D.K., and Bohlmann,J.D, An Aircraft Design Application UsingASTROS, WL-TR-93-3037, June 1993.

7) Barker, D.K. and Love, M.H., “An ASTROSApplication With Path Dependent Results,”presented at theAIAA/USAF/NASA/ISSMO Symposium onMultidisciplinary Analysis andOptimization, September 1996.

8) DeLaurentis, D.,Mavris, D.N., Schrage,D.P., “System Synthesis in PreliminaryAircraft Design Using Statistical Methods,”Presented at 20th International Council ofthe Aeronautical Sciences (ICAS).

9) Zillmer, S., Integrated MultidisciplinaryOptimization for Aeroelastic Wing Design,Wright Laboratory TR-97-3087, August,1997.

10) Ausman, J. and Volk, J., Integration ofControl Surface Load Limiting intoASTROS, presented at the 38thAIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and MaterialsConference, April 1997, Paper No. AIAA-97-1115.

11) Blair M. et al, Rapid Modeling withInnovative Structural Concepts, presented atthe 39th AIAA/ASME/ASCE/AHS/ASCStructures, Structural Dynamics, andMaterials Conference, April 1998, Paper No.AIAA-98-1755.

12) Zweber, J. and Blair, M., Structural andManufacturing Analysis of a Wing UsingAdaptive Modeling Language, presented atthe 39th AIAA/ASME/ASCE/AHS/ASCStructures, Structural Dynamics, andMaterials Conference, April 1998, Paper No.AIAA-98-1758.

13) Love, M.H., Milburn, R.T., and Rogers,W.A., Some Considerations for IntegratingAeroelasticity in CAE, presented at theASME Winter Annual Meeting, December,1987, Paper No. 87-WA/Aero-10.

14) Hoseck, J.J., Lyons, P.F., and Schmid, C.J.,Development of Airframe Structural DesignLoads Prediction Techniques for FlexibleMilitary Aircraft: Theoretical Development,AIAA Paper No. 81-1696, AIAA Systems

AIAA 98-4704


and Technology Conference, August 1981,Dayton, Ohio.

15) Love, M.H., et al, “Enhanced ManeuverAirloads Simulation for the AutomatedStructural Optimization System - ASTROS,”presented at the 38thAIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and MaterialsConference, April 1997, Paper No. AIAA-97-1116.

16) Whiting, B., and Neill, D.J., “InterfacingExternal, High Order Aerodynamics intoMSC/NASTRAN for Aeroelastic Analysis,”presented at the MSC Aerospace User’sConference, November 1997.

17) Karpel, M, Moulin, B., and Love, M.H.,“Structural Optimization with Stress andAeroelastic Constraints Using ExtendableModal Basis,” presented at the 39thAIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics, and MaterialsConference, April 1998, Paper No. AIAA-98-1868.

18) Zink, P.S., Mavris, D.M. Love, M.H.,Karpel, M., “Robust Design forAeroelastically Tailored / Active AeroelasticWing,” presented at theAIAA/USAF/NASA/ISSMO Symposium onMultidisciplinary Analysis andOptimization, September 1998.

19) Wang, B.P., Twu, M.J., Costin, D. , andEisenmann, J.R., and Norvell, R.G.,“Laminate Ply Stacking Sequence and PlyTermination Selection,” presented at the30th AIAA/ASME/ASCE/AHS/ASCStructures, Structural Dynamics, andMaterials Conference, April 1989, Paper No.AIAA-89-1295.

20) Love M.H., “Integrated Airframe Design atLockheed Martin Tactical Aircraft Systems,”presented at the AGARD 82nd Structuresand Materials Panel Meeting - Workshop onIntegrated Airframe Design Technology(Sesimbra, Portugal), May 1996.


THE F-22 STRUCTURAL/AEROELASTIC DESIGN PROCESS WITH MDO EXAMPLES

Nick Radovcich Division Manager Aeroelastic Analysis Lockheed Martin Aeronautical Systems AIAA Member&

David Layton Flutter & Dynamics Engineer Lockheed Martin Aeronautical Systems AIAA Member

AbstractDocumented experiences of MultidisciplinaryOptimization (MDO) applications during theengineering, manufacturing, and design phases offighter aircraft programs are not numerous.Documentation is even rarer for aircraft that haveflown. This paper describes in general terms the overalldesign experience of the F-22 fighter, and rapidlyfocuses on the aeroelastic/structural considerationswhere MDO like processes were employed. Central tothe design process is the Air Vehicle Finite ElementModel ( A/V FEM). The A/V FEM is the commonelement to link design requirements and processes forloads, flutter, stress, dynamics, and control law design.Multidisciplinary aspects of the interdependentprocesses includes stiffness tailoring for meeting flutterrequirements, control law tailoring for redistribution ofexternal loads, flex to rigid tailoring for satisfyinghandling qualities, stress sizing and aeroservoelasticfilter design within the general subject of aeroelasticoptimization. The investment of using a controlled A/VFEM for loads, stress, flutter, dynamics, control lawintegration, weight estimation, etc., was to a significantmeasure responsible for the excellent stiffness andloads tailoring which resulted in a minimum weightdesign while satisfying the airplane performancerequirements & allowing for the structural designparameters to be successfully iterated. The large A/VFEM was manageable in terms of configuration control,integration with specific discipline analysis processes,overall tracking/storing, and processing terabytes ofdata. The recovered cost of using a large model wasreturned many times over by savings in man-hours thanif structure decomposition/back transformation methodshad been employed. A very detailed loads grid, fueltank fuel-vapor boundaries matched to maneuverattitude and g loading, and detailed internal andexternal pressure loading were other challengessuccessfully achieved to satisfy the Integrated ProductTeams (IPT) requirements. The procedure formodifying panel flexible pressure loads to reflect non-linear wind tunnel rigid pressure distributions,especially due to control surface deflections, provided ahigh degree of fidelity to the flex to rigid and flex loadscalculations. Finally, the computer access for the usersdrove all the necessary MDO like processes. Thecomputational power and ease of use provided acapability to successfully manage the terabytes of dataacross wide area networks and many types ofcomputing platforms. Additionally, the storage of

results in relational databases provided fast and directanswers to questions with real time qualifications.

IntroductionThe road to a production F-22 fighter started withconcept studies during the mid-1980’s and a prototypefly off under the banner of Advanced Tactical Fighter(ATF) which was concluded in December of 1990.Participants which included competing teams andmulti-company collaborators had a number of rolechanges as the project came from behind the tightlyclosed doors during the concept days and into a morevisible prototype days. The project is in theEngineering, Manufacturing and Development (EMD)phase. Full envelope expansion is planned to start inMay 98 for ship 4001 at Edwards Airforce Base after asuccessful series of first flights conducted in Mariettaduring the third quarter of 1997.

The deciding milestone for the project came on theaward of the EMD contract to Lockheed in first quarter1991 after the conclusion of the prototype flight testprogram. The Lockheed prototype design demonstratedadequate performance, LO, and maneuveringcharacteristics. With the external geometry basicallyfixed, the focus of the design shifted to internalarrangement and design developments to satisfymaintainability, supportability, etc. requirements withweight as the principal metric for satisfyingperformance requirements.

Late in 1991, a number of trade studies were integratedinto the design to help manage the challenging weightconstraints. These studies foreshortened the fuselage bytwo feet and set the main landing gear configuration inthe wing. There were also minor changes to theplanform of all lifting surfaces and control surfacesbased on refined wind tunnel force models.

There are many interrelated requirements andconstraints, which enters into the design process andconsequently the evolution of the design. This paperwill focus on the design to data development, whichwas required to evolve the structure concepts anddesign.

Six areas were available to define the basis for thestructural design:


• Basic geometry; materials initial structuraldefinition.

• External loads driven by Airplane SimulatorResponses due to Maneuvers defined in theLoads’ Criteria and Weight.

• Flexible to Rigid Ratios.• Stiffness/Mass distribution for Flutter Margin

Requirements• Vibroacoustics environmental definitions and

high cycle fatigue design• Flight Control Laws and Aeroservoelastic

SSStttaaabbbiii lll iii tttyyy Requirements.

The integration of various disciplines represented bythe foreshortened list of six is largely governed by theconstraints imposed by many competing requirements.Ideally, full derivatives would be derived for aircraftperformance, LO signature, weight, equipmentplacement, maintainability, affordability, external loads,stiffness requirements, etc. with respect to each of avery large number of design variables.

Structures decided during the EMD proposal phase thatan approach would be pursued which would return tothe Project the greatest value for the resourcesexpended. The core issue for this approach was theutilization of a single vehicle FEM for all deriveddesign to data used to design the Structure:

• Vehicle loads (external, internal, internalpressures, etc.),

• Flutter and dynamics assessments,• Flexible to rigid ratios,• Extraction of material design allowables,• Aeroservoelastic analyses.

A balance in the vehicle FEM detail between accuracyand affordability was driven by the followingrequirements:

• The vehicle FEM had to have sufficient detailfor internal loads definition

• Model size could not overwhelm:• Databases for tracking and managing the

many FEM configurations (symmetric,anti-symmetric, left, right, and controlsurface deflections)

• Data management and computer usagerequirements for using the vehicle FEMwithout alternation by Flutter and Loads

Time lags in data availability appear due to the variousprocesses schedule requirements and the sequentialnature of the inter-related processes. In addition, somedesign decisions must be done early into the designprocess before a good definition of the structure isknown, such as locating the flight controls sensors.

The summary of the process flow for structure design todata is found in Figure 1. The data flow shows thatloads and flutter analyses are performed using a FEM (-1) which is one design (model) behind the FEM (0).More importantly, there are lags up to 3 design cyclesfor new flexible to rigid ratios and loads tailoring datato be incorporated into an updated flight simulator.Changes like loads tailoring had to first go throughcontrol law development cycle. Stress allowables,which define fatigue life requirements, may lag theprocess by 2 or more cycles. As bad as this may appear,as measured by external loads, stiffness requirements,and control law developments, the process didconverge. The major perturbation to the process wasthe changes coming from the Detail Design box. Herethe variability in the sizing and model grid and elementchanges caused significant changes in internal loads fora near equivalent external load definition.

In addition, the process was further removed from thedesired MDO approach because not all of the IntegratedProduct Team’s (IPT) budget profiles matched therequirements of the Process Flow Chart for an orderlyconvergence. With minimum weight requirementsdominating the structural design concepts, the IPTsdependence on fine grid structural sub-models grew.“Small” variations in load redistribution sometimescaused major shifts in margin calculations. This was aconsequence of forcing mathematical zero margins in afine grid FEM where large derivatives of internal loadchanges were possible for small changes in sizing orgrid definition.

The efficient computing and data management systemsemployed in the F-22 design development may haveproduced a downside or two. The IPTs decided to askfor redistribution of external loads on fine grid FEMsub-models. This permitted the using of a modelwithout going through the pain of understanding howthe structure really works up through ultimate load.The computer showed how a particular FEM could bemade to work without the proper controls on how wellthe FEM itself represented the structural concept.Good design concepts, which work on the hardwareairplane, are the deciding factors for establishing anefficient structural system that are lightweight, robust,and cost effective while avoiding single criteriaminimum weight solutions traps.

Statement of Problem Documented experiences of MDO applications forfighter aircraft during the design development phasesare not numerous. For aircraft that have flown,documentation is rare. The technical community knowsthe power of MDO and not having a cradle to graveexample has been a continual source of frustration, as


voiced by AIAA MDO technical committee membersover period of years.

Scope and Methods of ApproachThis paper describes in general terms the overall designexperience of the F-22 fighter, and rapidly focuses oniterative aeroelastic/structural design processes (Figure1) to highlight MDO like processes which were used.Central to the structural design process is the AirVehicle Finite Element Model (A/V FEM). The A/VFEM is the common element for loads, flutter, stress,dynamics, and control law design to processes.Multidisciplinary aspects of the interdependentprocesses includes stiffness tailoring for Flutterrequirements, control law tailoring for redistribution ofexternal loads, flex to rigid tailoring for handlingqualities, stress sizing, and aeroservoelastic filterdesigns within the general subject of aeroelasticoptimization. Finally, there are lessons to be learntfrom this exercise and in particular the specialrequirements of a fighter where volume is a premiumand structural concepts may be inherently non-optimumshapes as opposed to transport aircraft where thevolume permits fundamentally optimum shapes andconcepts.

Team Interaction and PoliciesTo achieve a minimum weight design while meeting theperformance goals required close coordination betweenthe customer and contractor as well as among thecontracting team members. As a result of this closecoordination a tailored design criteria was establishedto keep the design constraints specific and relevant tothe F-22. This entailed defining in close concert withthe customer a structural criteria document that wasspecific to the F-22 usage and performance.

The team integration was achieved by institutingpolicies and guidelines that each of the tri-companyteam members would be required to follow. Theseincluded developing a common set of materialproperties, conducting analysis with common orequivalent software tools, and building an Air VehicleFinite Element Model (A/V FEM). Additionally,significant effort was expended to ensure that theengineering design and analysis was closely integratedwith ground and flight-testing. This was accomplishedby developing detailed test plans in coordination withthe customer that was specific to the F-22.

Air Vehicle Finite Element ModelThe A/V FEM provides the foundation for the overalldesign process by providing a common basis forconfiguration control and analysis. The A/V FEM isthe common interface for many disciplines as shown inFigure 2) to develop design to data. This single model

is used to compute internal and external loads, flex-to-rigid ratios, flutter design requirements, andthermodynamic response. Figure 3 illustrates the size,complexity, and the number of configurations trackedfor this single model. The individual super elementswere built by the F-22 team member responsible for thestructure and then assembled for analysis by the primecontractor Lockheed Martin Aeronautical SystemsCompany (LMASC). A very detailed set of guidelineswas established and documented early in the program toensure compatibility among the organizationsdeveloping the model. These included defining thenumbering convention, definition of acceptable elementtypes, and the use of defaults and parameters.Additionally, the document included definition of anyrequirements defined by the functional disciplines tosupport their independent analysis tasks. An examplein this document was the requirement that thecomposite laminates be explicitly defined in thecomment statements to facilitate aeroelastic sensitivityanalysis at the composite ply level. The A/V FEM wasmanageable in term of configuration control,integration with analysis routines, overall tracking ofthe design, and storage/processing of terabytes of data.The cost of using a large model to generate aeroelasticdesign to data was insignificant compared to thesavings in man-hours achieved by using one verifiedmodel whose configuration control and responsibilityfor accuracy was vested in one group.

External LoadsThe air vehicle flight simulator drove computation ofexternal loads for transient maneuvers defined in theloads criteria report. The rigid air loads were based onextensive wind tunnel pressure model test data. Whilethe flexible incremental load distributions were derivedusing linear panel load methods, the panel loads wereadjusted on component basis based on wind tunnel rigidintegrated load values. The process permittedadjustments for non-linear effects especially near thecontrol surface hinge line. Another unique feature ofthe load process was the computation of the fuel tankpressure distribution consistent with the fuel freesurface orientation for the specific maneuver and fuelload distribution that was consistent with the loadcondition. Finally, hammer shock inlet pressuredistributions were used based on computational fluiddynamics (CFD) analytical codes and test data.

A major milestone during the first year was the releaseof a full set of design loads based on CFD data. Theloads latter agreed with the wind tunnel data to within 5percent. The CFD released loads were for complete setof control surface deflections.


Load tailoring by Maneuver Load Control wasestablished early in the EMD design phase. How muchcould the ailerons be used to dump the load inboard wasa function of two design considerations. The first wasthe effectiveness of the ailerons and the second was theimpact of the increased drag on performance. Thepoints in the sky where the maneuver load control(MLC) could be most effectively utilized, however, wasalmost on top of the maximum performance point.There was aggressive tailoring of the control surfacegain schedule to achieve weight benefits with MLCwhile holding the performance degradation to aminimum.

Load tailoring was achieved by minimizing adverseairplane responses during critical load’s maneuvers.Close coordination with developers of flight controllaws and quick turnaround for potential solutions on theflight simulation program were just two of the criticalprocess that lead to successful closure. Load’sengineers take six or more time hacks during eachmaneuver on the flight simulation. Critical loads areidentified for reduction and the time hack andassociated maneuvers are identified. Negotiationsbetween Flight Controls and Handling Quality (HQ)engineers and Loads engineers establish proposedchanges to the flight controls to tailor the loads. Thecycle is complete when the changes appear in the flightsimulation and a full load’s analyses and a completeHQ studies show that the tailored loads have beenachieved without introducing new issues for either HQor Loads.

Loads and flex-to-rigid tailoring through ply lay-upoptimization was attempted after the basic design wasestablished. Studies were conducted for the wing andvertical fin surfaces. Derivatives for each of the plydirections did not show large gains without impactingother constraints. The ply directions for the wingproved to be near optimum for basic loads. The winglayout naturally encourages efficient ply directionallocation because of the planform geometry. The zeroplies run parallel to the elastic axes for the outer wing.This is also true for the vertical fins. Bucklingmechanism is another significant factor for each ofthese surfaces. During the prototype trade studies,predominant buckling mode improvements could beachieved if ply lay-ups had non-traditional orientationsof (0,45,90). This is impractical from a materials testingpoint of view because of the costs associated with agreatly enlarge data base requirements. In each of theseareas the weight penalty due to low derivative valuesrequired other options to be pursued.

Internal Loads and Margin of SafetyAt Lockheed-Martin in Marietta, external loads formaneuvers and fatigue were processed through thevehicle FEM and the resulting internal loads wereloaded into Oracle relational database. The designer andstress analyst had immediate access not only to thecurrent loads released but also to past releases. Theanalyst then could compare what changed or work ondifferent releases of the drawings.

With weight a significant factor in the design process,many parts had zero margins of safety when released.With changes in the internal loads, some of those zeromargin areas could no longer support the new internalload distribution. In the course of the process thatfollowed, the question was raised, “what is the flightenvelope for the aircraft with negative margin?” Acomplex and data intensive methodology evolvedwhere point analysis programs generated margin ofsafety values for some 3000 load cases and thenthrough interpolation of flight conditions, contours ofzero margin of safety were derived in the Mach andAltitude plane. Then Aircraft Operating Limits (AOL)were then determined for the aircraft within thestructural capability and the derived limits based onwhat structural testing was completed up to that point.This margin of safety versus flight envelopemethodology will be a significant aid as the airplaneexplores the testing envelope where critical loadconditions exist.

Temperature EffectsTemperature distribution affects structural design in theselection of materials and in the introduction of thermalinduced stresses. Material allowable for composites isa function of maximum temperature and amount ofmoisture saturation. Hot-wet properties for compositesdominate the maximum temperatures allowed in thedesign. For aircraft structures constructed withdissimilar coefficient of expansion materials, such asmechanically joining of aluminum with compositecomponents, thermal strains must be accounted for inthe internal load definitions.

FlutterDefinition of the air vehicle flutter margin and thenecessary design to data lagged the detail design by nomore than a single design iteration and significantchanges were brought back an iteration to implement inthe aeroelastic model. Analysis metrics was establishedto facilitate tracking of the detail design. This includedthe definition of a procedure to compute, for eachcontrol axis, the total control loop stiffness, detailedweight estimates of control surface hinge-line inertiaand center of gravity, and unit loads on the A/V FEM totrack the structural flexibility.


The aeroelastic requirements were derived fromsensitivity and optimization of the design parameters.The design variables consisted of three primary types:percent changes to physical properties such as cross-sectional area and skin thickness; composite laminateproperties such as the addition of a single ply at a givenorientation angle; laminate material axis sweeps wherethe material axis for an entire surface is rotated.Table 1.0 lists a breakdown of the variables on a per-surface basis. To facilitate defining requirements interms of true sizing variables accurate and automatedsensitivity analysis to aeroelastic parameters isrequired. The F-22 program utilized “in-house”specialized software for sensitivity analysis.Additionally, a powerful Convex computer wasavailable with over a terabyte of disk and 10 terabytesof tape capacity.

Multiple complex analysis models and optimizationwas utilized to determine if a synergistic solution wouldprovide a decrease in weight or increase inperformance. For example, as part of the aeroelasticoptimization process a strength heuristic constraint wasimplemented. The heuristic approach defined theamount of material that can be removed in an area whenadditional material is added while not violating strengthrequirements. For example, if the optimization calls foradding plies to a laminate at +/-45 degree’s then either0 or 90-degree plies can be removed, the heuristicalgorithm constrains the amount to be removed.Additionally the process implements rules defined inthe structural policy document such as keeping thepercentage of plies at a given orientation angle withinspecified limits. F-22 structure effected by this type ofsizing includes the vertical fin and rudder.Interestingly, material added above the strength sizedesign for aeroelastic reasons at one design iterationturned out to be necessary in some areas for strength onthe next design iteration.

AeroservoelasticAeroservoelastic stability margins were defined byrunning a coupled analysis of the A/V FEM, theaeroelastic mass distribution, unsteady aerodynamics,and flight control laws. This multidisciplinary task wasaccomplished by Flutter organization by computingaircraft responses in the frequency domain and thencoupling these responses with control law’s supplied byFlight Controls. Both the control laws and the aircraftresponses were computed for a set of mach/altitude/fuelloading /maneuver load conditions that spanned theflight envelope with a heavy concentration in criticalregions. The process did iterate and converge byFlutter defining bandpass/lowpass filter requirementsfor each control law release. These changes were thenimplemented and reflected in a subsequent release of

the control laws. The sensitivities of the location ofboth the rate-gyros and the Nz accelerometer wereexamined. However, moving the sensors were notrequired as structural filters in the control laws providedadequate stability margins.

Both open loop and closed-loop ground testing wascompleted prior to first flight to obtain data that couldbe correlated with the analysis. Minor tailoring of thefilters was required after these tests.

DynamicsThere are two principal focuses with respect tostructural dynamics. The first is the definition of thevibration environment to support the design of bothairframe structure and equipment installations. Thebasis of this environment was flight test data acquiredduring the YF-22 (prototype) flight test program. Largedatabases of acoustic and acceleration data wereassimilated into the Environmental Criteria Documentto support detail design. The second focus was thevibration environment to predict and test the highcycle/low cycle fatigue life of structural sub-systems,equipment, tubing, avionics, etc.

Flexible To Rigid RatiosThe flexible to rigid ratios are computed by LoadsDepartment and is forwarded to the AerodynamicsDepartment for integration with rigid aerodynamicsdatabase. These data are used directly by the controlsdepartment to generate inputs to the flight simulationmodel, which in turn is used by Loads to determinemaneuvers critical to establishing design loads.

DADT /Stress AllowablesCrack growth analysis was the backbone forestablishing durability limits for the aircraft. Parts weredesigned for 8000 hours of life. Durability Analysis andDamage Tolerance (DADT) established working stressallowables throughout the structure. Point analysis wasperformed to support MRB (manufacturing reworkboard) activities using the same databases andtechniques established in the basic design.

Detail DesignThe major issue in detail design was the enormouspressure to meet allocated weight targets. Continuoustrade studies absorbed manpower and scheduleresources and as a consequence made the task of gettingFEM updated with best if not forward looking data avery low priority task. Since the FEM is the pivotalconnection to all facets of generating design to data, theinaccuracies in the FEM had serious impact of the rapidconvergence of the design process.


FEM ChangesThe process of building a finite element model for acomplete vehicle is complex and time varying. Rapidconvergence of the model configuration and propertiesrequires the team to look into the future to where themodel arrangement and the individual finite elementproperties will eventually converge. The challenge to beahead of the actual detailed design is made morecomplex when three groups in three differentcompanies attempt to operate as a single unit andovercome the different cultures, which by traditionoperated as a single unit within each company.Significant organizational tasks were required toassemble a model with many interfaces. Thisintegration task almost becomes an end to itself. Whatwent into the model in so far as material properties,sizing and grid point selections was by its very natureless visible and therefore less likely to be challenged. Inthe end, the devil was in the details for specification ofsizing data, grid point selection, and material properties.Near the end of the design iterations, the biggestvariation in internal loads was in FEM property changesand not the external loads.

Typical Processes During IterationThe basic design iteration was a process that essentiallycreated data sequentially. For example, a FEM wasrequired before basic load process could start. Allexternal loads must be computed before internal loadscould be established and loaded into the relationaldatabase. All of the internal loads were required beforesizing of aircraft parts could start. And finally, theaircraft parts had to be designed before the FEM couldbe updated. Within this basic design loop, stiffnessrequirements were established using FEM and massdistributions together with unsteady aerodynamicrepresentations, which in turn were supported by windtunnel flutter model testing. Stiffness requirementsoften worked inside the basic design cycle at a rate of 2or 3 iterations to one full design cycle iteration. Thedesign iteration would not work practically unless eachgroup in the design process worked with models anddata that were one or more iterations behind the currentcycle. Also, strategic short cuts had to be taken duringsome of the iterations to get forward looking modelsand designs to leap frog the full design iterationschedule. Additional short cuts were required whenrequirements had to be updated to support long leadmanufacturing schedules. This required analyst toaccept or specially modify what ever the vehicle systemanalysis maturity was available at that time. In somecases the requirements were limited to only subsystems.The actuator stiffness loop requirements were decidedyears ahead of the 90% drawing release dates becauseof the long lead times for the control surface actuatordevelopment and testing required for flight. The flight

controls development was planned for late softwarereleases because handling qualities was dependent onextensive wind-tunnel testing and the integration ofstructural flexibility effects into the simulation modelfrom which the processes of control syntheses soheavily depended. But external loads was committed tousing maneuvers from the same HQ simulation modelto determine in-flight loads as they occurred and notarbitrary maneuvers based on specific criteria such asmaximum control surface deflections.

The process flow of specific tasks was more like a quiltthan a simultaneous interacting derivation of design todata. Figure 1.0 illustrates the basic interactions and thebox show the iteration cycle lags that some of theprocess-generated data entered into the design. Thecomplete design iteration cycle included external loadsto internal loads to design update to the FEM update forthe next iteration. The initiation of complete cyclewhich included fatigue design to data generation wasmajor commitment of program resources. During thismajor design cycle, there was many timely injection ofstiffness requirements. The stiffness and high cyclefatigue requirements often short circuited the outer loopwith 2 or more updates within one overall large loads,design and FEM update cycle. Another iteration loop,which operated inside the main loop, was load tailoring.This was particularly true during the last phases of thedesign development. Load tailoring will probablycontinue during flight-testing.

Rather than being a well ordered sequence of events,the team injected updated design to data where theleverage to impact the design had the most benefits interms of the resolving next most critical milestone. Inthis role, the team interpreted what the programrequirements were, and even if a moving target,provided design to data with the best rate of return andstill remain within the budget constraints of eachIPT/Design to data function support.

Vehicle Level Results

Stiffness RequirementsThe control loop actuation stiffness requirements foreach of the flight control surfaces namely the rudder,horizontal stabilizer, aileron, flaperon, and the leadingedge flap was directly imposed on the IPTs. Thedefinition of how to compute the loop stiffness for eachcontrol axis was defined in an Interface ControlDocument. This metric was used to allow the IPT’s todetermine the minimum weight design that satisfied thestiffness requirement. Typically, three IPT’s wererequired to determine the stiffness allocation among themain surface, actuator, and control surface. Table 2.0lists the breakdown in stiffness for each control axis.


Loads TailoringWith the design drawings basically released tomanufacturing, load tailoring via control laws surfacescheduling changes provided the tool to keep theexisting design within the existing structural capabilitybox while retaining the performance and HQrequirements.

Design To DataStructure organization provided 90% of the design todata for the F-22. The effective management of internalload data permitted the controlled phased releases ofdrawings with manageable audit trails. The processprovided flexibility when design updates required torelease two different airplanes designs known as Block1 and Block 2. The process kept the airplane designweight within the contract performance specifications.

Design to data as issued from the Flutter organizationconsisted of defining true design data such as percentchanges to physical properties such as cross-sectionalarea and skin thickness, and composite laminateproperties such as the addition of a single ply at a givenorientation angle. Figures 4.0 illustrate how data wastransmitted to the appropriate Integrated Product Team.The important point here is that the Air Vehicle FEMwas used as the vehicle to transmit design to data. Thisallowed for “checking” the design as to theincorporation of the requirements and for keeping ahistory of the requirements. Aeroelastic sizingrequirements were defined for the horizontal stabilizerskins, vertical fin skins, rudder skins and substructure,flaperon skins, and wing mounted pylons. . Prior totransmission of the design to data coordination andagreement was reached between the functionalorganization and the IPT that these design changescould be accommodated.

Summary of Important Conclusions

The investment of using a controlled A/V FEM forloads, stress, flutter, dynamics, control law integration,weight estimation, etc., was to a significant measureresponsible for the excellent results for stiffness andloads tailoring for minimum weight design whilesatisfying the airplane performance requirements. Thestructural design was successfully iterated during fourmajor design cycles. For this type of aircraft, rapidconvergence was achieved by: 1) satisfying externalload strength and life requirements; 2) then iterate forstiffness and dynamic sizing requirements. Theseprocedures generated critical design to data, which wasrequired by the analyst and designer to provide insightinto the available design space and the direction formoving the design. These studies provided data foruncoupling certain design parameters during the design

iterations. The large A/V FEM was manageable interms of configuration control, integration with specificdiscipline analysis routines, overall tracking, storing,and processing terabytes of data. The recovered cost ofusing a large model was return many times over bysavings in man-hours as compared todecomposition/back transformation approaches. Thecommon basis for communication and changes to themodel made the MDO like processes affordable andmore to the point, feasible. A very detailed load grid,fuel tank fuel-vapor boundaries matched to maneuverattitude and g loading, and detailed pressure loadingwere other challenges successfully achieved to satisfythe IPT’s requirements. The procedure for modifyingflexible panel pressure loads to reflect non-linear windtunnel pressure distributions especially due to controlsurface deflections provided a high degree of fidelity tothe flexible to rigid ratios and flexible loadscalculations. Finally, the computer access for the usersdrove all the necessary MDO like processes tosuccessfully provide and manage the data across widearea networks, using many types of computingplatforms, relational database storage of results for fastand direct answers to questions with real timequalifications.


Minimum Weight DesignLO Constraints

Equipment PackagingMaintainabilityLife Cycle costsManufacturing

StressAllowables

Process FlowProcess Flow

FEMExternal

Loads

FEM

Detail designStress; StabilityLife ; Stiffness

Material

ElementChanges

Grid PointChanges

SizingChanges

FEMUPDATE

FlutterDynamics

FlightSimulator

ControlLaws Dev.

Flex/RigidRatios

LoadsTailoring

StiffnessRequirements

InternalLoads

MissionProfilesExternal

Loads

FEMInternalLoads

UsageSpectrum

LifeAssessment

-1

0

-1

0

-3

-2

-2


FUNCTIONAL DISCIPLINES

FLUTTER & DIVERGENCE AEROSERVOELASTICITYEXTERNAL LOADS INTERNAL LOADS FLEX-TO-RIGID RATIOS

AIR VEHICLEFEM

INTEGRATED PRODUCT TEAMS (IPT’s)

MID FUSELAGE WING AFT BODYCOCKPIT EMPENAGE FORWARD FUSELAGE

EDGES LANDING GEAR

DESIGN TO DATA

Figure 2 Discipline / IPT / FEM Relationship

Figure 4 Communicating Design ToData

Figure 3 Air Vehicle FEM


AllowableFreeplay(Degrees)

Stiffness/FreeplayDriverSurface

RequiredLoopStiffness(in-lb/Rad)

FrequencyRange(Hz)

LoopRequirementImpact

R.S.S. At LifePins 0.0183 0.060 Classical Flutter

StabilizerBearings

21.4 23 – 30 Weight = 79 Lbs.0270 .069 LCO

Rudder 5.86 27 – 35 Weight =42 Lbs .0344 .175BuzzLCO

Flaperon 5.40 21 – 28 Weight = 6.0 Lbs 0.1060 .0300Classical FlutterLCO

Aileron 1.60 33 – 40 N/A 0.0810 0.274BuzzLCO

Actuator # 1 3.58 23 – 30Number of slices& Backup Stiffness

< 0.82 0.82Classical FlutterLCO

Actuator # 2 1.72 30 Backup Stiffness < 1.21 1.21 LCOActuator # 3 1.46 30 Backup Stiffness < 1.21 1.38 LCOActuator # 4 1.41 30 Backup Stiffness < 1.38 1.38 LCO

LeadingEdgeFlap

Actuator # 5 1.29 30 Backup Stiffness < 1.38 1.38 LCOFin See Rudder N/A Weight = 60 Lb. N/A See Rudder

Table 1.0 Loop Stiffness Impact & Freeplay Requirements

Surface Type QuantityRudder Skins 118

Spars 6Ribs 6

Vertical Fin Skins 138Spars 10Ribs 5

Flaperon Skins 132

Aileron Skins 72

Tail boom Skins 19

HorizontalStabilizer

Skins 162

Spar 15

Wing Skins 195Spars 8

Total = 886

Table 2.0 Design Variable Distribution

AIAA-98-4734


A COLLABORATIVE OPTIMIZATION ENVIRONMENT FOR TURBINE ENGINE DEVELOPMENT

Peter J. Röhl*, Beichang He*, Peter M. Finnigan†

General Electric Corporate Research and DevelopmentSchenectady, NY 12301

Abstract

A MDO scenario for the design and manufacturingprocess of gas turbine engine disks is developed. High-fidelity engineering analysis and process simulationtools are integrated into an optimization environment.While different formal MDO approaches are discussed,a sequential optimization approach seems to be bestsuited or this specific problem. The forgingoptimization results in a minimum-weight forgeabledisk that meets all constraints in terms of processparameters. The optimization of the heat treatmentprocess reduces residual stresses while maintainingrequired cooling rates through the modification ofsurface heat transfer coefficients. Optimization of boththe forging and the heat treatment process individuallyhas been successful, but the complete MDO scenariostill faces a number of obstacles. Parametric CAD toolsare not as robust for complicated geometry as it wouldbe necessary in an automatic optimization environment.The same applies to the interface between CAD andCAE tools. Computational resources constitute anotherbottleneck - formal MDO algorithms tend to be slow intheir convergence behavior, which makes them less wellsuited for problems requiring high-fidelity analysiscodes with their long execution times. Despite all theseobstacles, though, progress towards a comprehensivedisk MDO environment is apparent.

Introduction

The design and manufacturing of gas turbineengines is a highly coupled multidisciplinary processinvolving design of the thermodynamic cycle, flow pathand airfoil design, rotordynamics, and thermo-mechanical design for life prediction. An importantaspect is the design and optimization of themanufacturing process of the mechanical components,requiring detailed simulation of forging, heat treatment,and machining processes. With the economic pressureswhich exist today, the need to develop affordable, high-

* Staff Engineer, Member AIAA † Manager, Mechanical Design Methods and

Processes ProgramCopyright © 1998 by P. Röhl, B. He, and P. Finnigan.Published by American Institute of Aeronautics andAstronautics, Inc., with permission.

performance defense systems, with shorter productdevelopment cycle times has never been greater.Propulsion systems are no exception considering theirintrinsic complexity and strong system coupling withtheir associated aircraft or launch vehicles. Thesuccessful development of integrated propulsionsystems is critically linked to our ability to performsystem, subsystem, and component-level simulations ofthe design and manufacturing processes. Today, theproblems are compounded because of thegeographically distributed intra- and inter-companypartnerships, including second and third tier suppliers,which are formed out of economic, technical, andproduct necessity. The ability for industry to develop,and cost-effectively deploy these systems, is predicatedon its ability to rapidly simulate both products andprocesses to achieve globally optimized designs. Tothat end, there are a number of key technologies whichare being developed and demonstrated under theDARPA-funded RaDEO (Rapid Design Explorationand Optimization) program1 as part of the propulsionscenario. Under the RaDEO contract, the GE Researchand Development Center is teamed with EngineousSoftware, Inc. (ESI) to develop a collaborativeoptimization environment based on iSIGHT2,Engineous Software’s optimzation framework. Onefocal point is the development of an optimization toolkitwhich enables the user to easily formulate an MDOproblem and cast is into the form of one of the “formal”MDO algorithms supported by this toolkit. Another isthe extension of the iSIGHT environment to facilitatethe integration of CAD and CAE systems with the helpof two toolkits, the Product Modeling Toolkit (PMTK)and the Discrete Analysis Modeling Toolkit (DMTK).The engine disk design problem is one of theapplication demonstrations to be addressed in thisproject.

The Engine Disk MDO Problem

The individual steps of the disk design process,broken down into the mechanical design andprocess/manufacturing aspect, are shown in fig. 1. Eachof the five steps in the process can be further subdividedinto a number of individual sub-steps with analyses atvarying levels of complexity. The thermo-mechanicaldesign, for example, starts with a simple 1-d analysis toobtain a rough thickness distribution of the disk. Asknowledge about the design increases, more complexanalysis models are created up to a full 3-d finite

AIAA-98-4734


Design Manufacturing

Requirements

MechanicalDesign

Forging HeatTreatment

LifePrediction Machining

FinishedPart

Die(s)

Billet

ProcessParameters

ProcessParameters

Finished Part

HCF/LCFData

• Geometry• Material

• Part Life • Residual Stress• Distortions

• Residual Stress

• Material Properties

ProcessParameters Near-net Shape

Disk Forging

Fig. 1: Engine Disk Design and Manufacturing Process

element analysis with tens of thousands of elements. Athermal transient analysis is performed on the disk tosupply the mechanical design group with thermal loadsfor the different points in the design missions. Thesethermal loads are iteratively adjusted as the designprogresses. Objective during the mechanical designphase is first and foremost the determination of the finaldisk shape, as early as possible in the design timeline inorder to be able to release the forgings which tend torequire a long lead time. A shape is to be determinedthat meets mission requirements at minimum weightand/or minimum cost. A detailed simulation of themanufacturing process is necessary in order todetermine both residual stresses and final distortions ofthe finished part after machining operations. Theseresidual stresses, in turn, are used in the subsequent lifeprediction of the part. Objective during the simulationof the forging process is the determination of the dieshape on the one hand and of an optimum forgingprocess on the other that ensures proper die fillingwithout compromising mechanical properties of the diskthrough the violation of stress, strain, strain rate, ortemperature limits. The subsequent heat treatmentprocess is designed to generate acceptable mechanicalproperties in the forged disk. A simulation of themachining process results in the final disk shape withaccurate residual stresses and distortions. If thedistortions are within acceptable limits, an accurate lifeprediction of the part will be performed. Otherwise, the

heat treatment or forging process need to be improvedin order to achieve acceptable distortions. If that is notpossible, the finished disk shape needs to be changedand the mechanical design - at least in parts - berepeated. The same applies in the case that the designdoes not meet life requirements.

As this description demonstrates, an integratedprocedure that addresses both mechanical design andmanufacturing processes is absolutely necessarybecause of the iterative nature of the process and theprohibitive costs involved if changes become necessaryonce actual parts are being produced. Simulation toolsfor each individual stage are available and widely used.But opportunities for mathematical optimization of theindividual process steps are currently not fully utilized,and an integrated procedure which is the ultimate goalof this research is missing altogether.

If we try to recast the disk design problem in theform of a formal MDO problem, weight can beconsidered as the overall system objective, and thedifferent objectives of some of the individualsubsystems can be formulated as constraints. Weighthere would be the billet weight of the forging, which, ofcourse, also includes the weight of the final part, both ofwhich need to be minimized. Since the forging billetweight is inherently much larger that the final partweight, a linear combination of the two in the followingform could be considered as the system objective:

AIAA-98-4734


( )F W Wfinal billet= + −α α1 (1)

Mechanical

Thermal

Forging

Heat Treatm.

Machining

Life

opt.

Fig 2: Multi-Discipline Feasible Formulation

Mechanical

Thermal

Forging

Heat Treatm.

Machining

Life

opt.

Fig 3: Individual Discipline Feasible Formulation

In a multidiscipline-feasible type scenario (fig. 2),each of the disciplinary analyses would contribute anumber of constraints to the system level optimizationproblem. In an individual discipline-feasible typescenario (fig. 3), the feedback loops from life analysisto mechanical design and from thermal analysis tomechanical design are severed at the cost of additionalsystem level constraints accounting for theinterdisciplinary discrepancies introduced. Both ofthese standard formulations are not very satisfactory forthis type of problem out of several reasons. First, alarge number of system level constraints would beintroduced which are purely disciplinary in nature. Itmakes no sense for the system level optimizer to bebothered with all the intricacies of the forgingoptimization problem, for example. Additionally, theheat treatment problem is an optimization problem initself, but it does not directly contribute to the overallobjective, weight, but rather addresses producibility andthe satisfaction of constraints for distortion and materialproperties. Therefore, the disk design problem calls for

an approach where the optimization itself is distributed,and where each disciplinary optimization problem doesnot necessarily contribute directly to the overallobjective. Both the Concurrent Subspace Optimization(CSSO)3 and the Collaborative Optimization (CO)4

methods have been looked at as possible solutions, butit seems that neither one of them really captures thesalient features of the disk design problem. CSSOassumes a common objective that each discipline issomehow contributing to, and requires anapproximation of the non-local states in each discipline.This means one would have to create an approximationof the forging problem inside the heat treatmentproblem and so on, which is not very practical. COintroduces artificial non-physical objectives for thedisciplinary optimization problems so that for thedesigner it is somewhat difficult to follow the progressof the optimization from a disciplinary point of view.Besides, slow convergence rates in conjunction withlong analysis times (in the order of several hours peranalysis for the forging problem, for example) renderthis approach impractical. Therefore, it seems that inthis case a sequential optimization within an integratedframework seems most promising (fig. 4), where westart with the mechanical design problem and simple 1-D and 2-D axisymmetric tools to obtain an initial diskshape, use this to design a near net shape forgingprocess, then optimize the heat treatment process, andfinally perform the machining and life analyses.

Thermal

Machining

Life

Mechanical

opt.

Forging

opt.

Heat Treatm.

opt.

Fig 4: Sequential Optimization Approach

In subsequent loops, full 3-D analysis tools areused in the mechanical design phase. This sequentialapproach is possible in this specific case because thenear-net shape forging optimization will notcompromise the thermo-mechanical minimum weightdesign, and the optimum heat treatment process has noinfluence on either one of the two.

AIAA-98-4734


Simulation and Optimization Tools

Optimization Framework

The basic framework for the optimizationenvironment under development is iSIGHT, a softwareproduct developed and marketed by ESI. iSIGHT isconceptually a follow-up product to Engineous5,6, theoptimization framework developed at GE CR&D duringthe 1980s. Both products facilitate easy integration ofboth commercial and company proprietary software intoan overall optimization environment which makes useof the concept of interdigitation7 where the user has asuite of optimization tools available, including gradientbased and heuristic search techniques, geneticalgorithms, and simulated annealing, which can be usedin any combination during the optimization process.Experience over the years has shown that oneoptimization strategy alone is often unable to solve aproblem, but that a combination of different strategiesleads to improved results. iSIGHT enables the user toformulate a sequence of different optimizers and thenapply this sequence to the optimization problem.Another strong point of iSIGHT is the ease with whichanalysis programs can be integrated into the framework,including large-scale engineering applications like finiteelement codes. These codes can reside on theirrespective platforms, irrespective of where iSIGHT isinstalled, an important point with respect to softwareleasing and maintenance cost for software which may belicensed only on a certain workstation. A number oftoolkits are under development in conjunction with theRaDEO project, among them the Product ModelingToolkit (PMTK) to support product data modeling andthe interaction with commercial CAD software, and theDiscrete Analysis Modeling Toolkit (DMTK), whichfacilitates the interaction of analysis models of differentdisciplines and levels of fidelity. Both of these toolkitsare heavily leveraged in the engine disk design scenario.

Process Simulation

Nowadays advanced process simulation tools arebecoming more and more available for all stages of thedisk design and manufacturing process. Simulationtools such as DEFORM8 and ABAQUS can accuratelypredict the mechanical behavior and properties duringthe manufacturing process. Therefore, these tools havebecome the state-of-the-art and are widely used. Incombination with numerical optimization techniques,these tools offer the opportunity to improve individualsteps in the overall process9. In this application,DEFORM was chosen as the tool to be applied in theforging and heat treatment optimization procedure.

CAD Tool

The CAD tool of choice is Unigraphics10,developed by EDS, which is the adopted CAD softwareat GE Aircraft Engines. Parametric master modelscontrol the geometry and engineering analysis “views”which support analyses at different levels of fidelity.These analysis-“views” - or context models - are de-featured models capturing the essential geometry for therespective analysis. They can also contain additionalinformation necessary to generate the analysis modelslike boundary conditions and load and meshinformation. PMTK will allow the user to graphicallypick geometric design variables from the CAD modeland automatically link them with the optimizer fortopology optimization.

Mechanical Analysis

Finite element analyses are performed usingANSYS, with model preparation done partly in ANSYSand partly in PATRAN. PATRAN’s P/THERMALmodule supplies the required heat transfer data andtemperature boundary conditions for the stress analyses.Different approaches are being evaluated for automaticanalysis model generation from the CAD representation.One is the use of “tags” in the CAD model, where theCAD model would house all the information necessaryto generate the analysis model. Another is the use ofscripts that are reusable. This approach relies on aconstant topology of the geometric model and entityconsistency of the geometry import into the CAE tool.

Current Status

MDO Algorithms

Implementations of both the CSSO and COalgorithms inside iSIGHT have been developed andtested11,12. Since convergence of the CO algorithmtends to be very slow, its usefulness in detailed designapplications requiring high-fidelity engineering analysisremains doubtful. One promising possiblity is thecombination of the CO algorithm with response surfacesin order to reduce the number of analyses at thesubsystem level. An initial implementation of theCSSO algorithm has been validated both with standardtextbook-type example problems and with two morerealistic problems representing a welding design and anidealized turbine blade. This CSSO implementation iscurrently being evaluated in connection with the diskdesign problem.

AIAA-98-4734


Forging Shape Opitmization

A procedure has been developed to address theforging shape optimization problem, integratingUnigraphics and DEFORM with iSIGHT, leveragingfunctionality of the product modeling toolkit. Reference13 describes the system in greater detail than iswarranted here. The objective of the forging shapeoptimization problem is the design of a minimumweight forged shape that satisfies constraints on bothforging press capacity, strain and strain rates, die filling,and minimum coverage of the final part shape.

In the present study, forging is modeled as a time-dependent, plastic-deforming, either isothermal or non-isothermal process. Since the forging simulation isconducted in an optimization environment, some of theprocess and geometry parameters are modified in eachDEFORM run. Therefore, it is necessary to regeneratethe mesh and redefine the boundary conditions.Furthermore, it is necessary to post-process the analysisresults and extract information on optimizationobjective and constraint functions. Several moduleshave been developed that drive DEFORM toaccomplish following tasks:

• import geometry and regenerate die and billetmeshes,

• create appropriate boundary conditions,• start DEFORM simulation in batch mode,• monitor DEFORM runs, and• postprocess simulation results to extract maximum

press load, strain, temperature, etc.

Each of these modules acts like a separateexecutable, or “simcode” in iSIGHT terminology.iSIGHT executes these “simcodes” in a pre-definedsequence, including potential looping and branching.

Consider the forging shape optimization of ageneric turbine disk. A cylindrical billet is forged into adisk of the shape shown in figure 5. The die geometry iscaptured in a Unigraphics parametric model. Severalfillet radii R1-R6 have been chosen as design variables.Both invalid geometry and intrusion into the minimumcoverage over the so-called “shipped” shape, theintermediate shape in which the forging vendor suppliesthe part, and which is used for testing purposes, can beprevented by putting simple bounds on the designvariables. It should be noted that simple bounds maynot be sufficient to guarantee geometric validity in amore general situation. They work in this case becausethere is no coupling among the selected designvariables.

Figure 5: Turbine disk and its cross-section

Thus, the optimization problem is formulated as

min V,s.t. R R R (i , )i lb i iub≤ ≤ = 1 6K, ,

P Pub≤ ,where V is the volume of the work-piece, P and Pub arethe maximum press load and its upper bound,respectively, and Rilb and Riub are given lower and upperbounds of the fillet radii, respectively. The mostaggressive shape, which corresponds to the lowestvolume V, has been chosen as the initial design. It isrelatively easy to get this shape from the specified diskdesign by adding a minimum cover. However, the pressload constraint is usually violated for this design, andthus the fillet radii Ri have to be increased which resultsin a larger volume. Subject to the press load constraintPub, the optimizer will choose the optimal values ofdesign variables Ri . As it is pointed out in the previoussection, iSIGHT provides a suite of optimizationalgorithms. The modified method of feasible directionsfrom ADS[14] is employed in this study. Since analyticaldesign sensitivities are not available, the gradientinformation has to be obtained through finitedifferencing.

R5 R1 R3

R2R6 R4

AIAA-98-4734


In the example we consider a time-dependent,plastic deforming, isothermal, closed-die forgingprocess. The top and bottom dies are assumed to berigid. The maximum load P normally occurs at the endof the forging stroke as the dies fill out and the materialstarts to move into the flash region. The load changesrapidly with the stroke at this stage of the process.Therefore, it is difficult to accurately compare the loadsat the end of the stroke from different die designs due tothe inherent noise in the load predictions. For thisreason we artificially set P to be the stroke-averagedload in between 98-99% of the final stroke. A goodestimate on the real maximum press load may beobtained by multiplying P with a correction factor.

There are about 1600 quadrilateral elements on theworkpiece, and automatic mesh regeneration is enabledto accommodate the large deformation that is inevitablein the forging process. Four design variables R1 - R4 areused in this application, and the time step is taken to be∆t = 0.1s. Due to repeated remeshing during the forgingsimulation, non-smoothness is introduced in the finiteelement solution. Therefore, we chose a 10%perturbation on the design variables during sensitivityanalysis using finite differencing to smooth out thedesign space. Although the design sensitivities socalculated may not be very accurate locally, theyprovide the optimizer with the right search directions ina global perspective.

100

80

60

40

20

number of simulations

desi

gn v

aria

ble

valu

es

0 5 10 15 20 25 30 35

Figure 6: Initial and final shapes (left), and design variable values versus the number of simulations (right) forisothermal the forging process

100.0

89.5

84.2

78.9

76.3

81.6

86.8

92.1


forg

ing

pres

s lo

ad

94.7

97.4

0 5 10 15 20 25 30 35

Figure 7: Disk volume and press load vs. the number of forging simulations for the isothermal forging process

The initial and final shapes of the disk are shown inFigure 6 (left). The history of design variable valuesagainst forging simulation runs is given in Figure 6(right). The objective (volume) and constraint (press

load) function values versus simulation runs are shownin Figure 7. All numbers have been normalized. Theresults suggest that the optimization is close toconvergence after 15 simulation runs. There are some

initial shape final shape

0 5 10 15 20 25 30 35

100.0

91.6

86.1

80.6

83.3

88.9

94.4


volu

me

of th

e tu

rbin

e di

sk 97.2

AIAA-98-4734


downward spikes in the figures. The smaller ones arethe result of finite difference perturbation, while thelarger ones are due to line search of the optimizer. Sincethe abscissa shows the number of simulation runs asopposed to the number of optimization iterations, theresults of both finite differencing and line search havebeen included. The upper bound of the press load Pub =77.8 is shown as a dashed line in Figure 7 (right). It isapparent that the forging press load far exceeds thislimit initially. As a result of the optimization, the pressload drops from 96.9 of the initial design to 77.8, whichis the upper bound, a 19.7% reduction. The volume,however, has been increased by 12.4% from 82.4 of theinitial minimum-weight shape to 92.6 of the finaloptimized shape. In addition to the single step processdescribed here, a multi-step forging process is presentedin ref. 13.

This work can be extended in several aspects: first,new interprocess communication mechanisms may beintroduced to improve data passing between processes;second, a more comprehensive forging simulationshould be conducted that includes the effect of heat lossduring transport of the billet and positioning of thetools; third, a larger design space may be explored byincorporating more geometric parameters as designvariables; finally, additional constraints, such as thoseon strain rate and temperature, should be considered tomodel more realistic situations.

Heat Treatment Optimization

The purpose of the heat treatment process is todevelop the necessary mechanical properties in theforged part. This is achieved by heating the part tosolution temperature and then cooling it rapidly.During the cooling phase residual stresses areintroduced. In the case of Ni-based superalloys that areconsidered here, a certain minimum cooling rate has tobe maintained to generate the needed creep and tensileproperties. On the other hand, the faster the coolingprocess is, the higher are the resulting residual stresseswhich can lead to excessive part distortions aftermachining to the final shape.

Traditionally, an oil quenching process has beenemployed which ensures fast cooling and thus a highcooling rate, but the oil quenching process introduceshigh residual stresses, and, from a process optimizationpoint of view, offers very little room for improvementas there are very few parameters which can becontrolled. Therefore, fan cooling is gaining largeracceptance where it is possible to control the airflow onindividual sections of the part and thus influence thelocal surface heat transfer coefficients. Obviously, theheat transfer coefficients that can be achieved with fan

cooling are lower than those for oil quenching, so thatfor thick parts it may not be possible to satisfy coolingrate requirements, but for moderately thick parts fancooling offers clear advantages. For very thin parts likeengine seals, where machining distortions due toresidual stresses are especially critical, fan cooling maybe the only process that produces acceptable parts at all.

The challenge here is to formulate a fan coolingoptimization problem without actually having toexecute a combined heat transfer-stress analysis eachtime the optimizer needs a new design point. Anaccurate heat transfer analysis requires small time stepsin the simulation, and a stress analysis, in turn, requiresa fine finite element grid, therefore the combination ofboth is the most computationally expensive analysispossible. In general, though, the stress analysis is muchmore time consuming than the heat transfer analysisalone. Since it is known that spatially uniform coolingreduces residual stresses, the idea is to formulate anobjective function that penalizes non-uniform coolingand at the same time ensures fast cooling at or above thetarget cooling rate. These are obviously two conflictingobjectives since fast cooling always means unevencooling as the heat can only be extracted at the surfaceof the part. Therefore, the objective function for theheat treatment optimization problem is formulated in aquadratic form that penalizes the deviation from thecooling rate target:

( )( ) ( )

objw t t if t t

w t t if t tnodes=

⋅ − <

− ⋅ − ≥

∑& & & &

& & & &

target target

target target

2

21

(2)

W is a user-defined weighting factor between 0 and1 that penalizes under- and over-achievement of thetarget cooling rate differently. A value close to one (butless than one, of course) seems to give the best results.The target cooling rate is also a material-dependentvalue. Design variables are the surface heat transfercoefficients, hi, which can be related back to a certainairflow produced by the fan cooling apparatus.

A total number of up to ten or twelve designvariables seems to be in the range of what can becontrolled by current fan cooling fixtures. Theprocedure developed here gives the user a choice interms of optimization constraints. He can impose ahard constraint on the cooling rate:

cnodes

t t if t t

if t tcr

nodes=

− <

≥

∑

1

0

& & & &

& &target target

target(3)

This constraint has a discontinuity at zero, exactlywhere it is active, and will never assume a value less

AIAA-98-4734


than zero, that is satisfied and not active, caused by the“if” in the constraint formulation. This discontinuityleads to problems with gradient-based optimizers,which will always see a zero constraint gradient for asatisfied or active constraint, therefore in the case ofconstraint satisfaction the constraint value of zero isreplaced with the difference of the target cooling rateand the minimum of all nodal cooling rates:

′ = −c t tcr & &target min (4)

In this fashion at least the sign of the constraint gradientthat the optimizer sees above and below a constraintvalue of zero will be equal. An additional constraintcan be placed on the nodal fraction that fulfills thecooling rate target which has to be equal to 1.0 if thetarget is met everywhere. The two constraints mayseem somewhat redundant, but depending on theoptimization strategy used, one or the other or acombination of both lead to the best convergence.

The heat treatment optimization proceduredescribed above was applied to a generic turbine disk.Figure 8 shows the heat treatment geometry and thedistribution of the nine design variables employed.

h9

h8h7 h6

h5

h4h3h2

h1

Fig. 8: Turbine Disk Geometry and Fan CoolingVariables

In order to cut down on analysis time, theoptimization was started with all heat transfercoefficients linked to only one design variable. Thisproblem was executed for six iterations, using thesequential linear programming technique from ADSinside iSIGHT, until both constraints were active. Thefull convergence history is depicted in figure 9, and theconstraint history is shown in figure 10. Negativeconstraint values indicate a satisfied constraint.

At this point, all nine design variables wereactivated, and the new optimization problem convergedwithin seven more iterations, that is 13 total. For thissegment, the modified method of feasible directions,also from ADS, was chosen as the optimizationtechnique. The deviation function was initially reducedfrom a value of 1.4 to about 0.6 and then further downto under 0.2. These numbers as such have no physicalmeaning, but the significance can be seen in a

comparison of the initial and final cooling ratedistribution (fig. 11 and 12), normalized with respect tothe target value, indicating a much more uniformcooling than at the starting point.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1 2 3 4 5 6 7 8 9 10 11 12 13

Iteration

Co

ol O

bj

Fig. 9: Objective History

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

1 2 3 4 5 6 7 8 9 10 11 12 13

Iteration

Co

nst

rain

t V

alu

e

Ccr

1-Cnf

Fig 10: Constraint History

Fig. 11: Initial Cooling Rate Distribution

AIAA-98-4734


Fig. 12: Final Cooling Rate Distribution

The question still to be answered is what effect thisoptimization procedure, which is based on heat transferanalysis only, has on the residual stresses of the partwhich is what we are ultimately interested in.Therefore, a combined heat transfer-stress analysis wasperformed on both the starting configuration and on adisk with the final heat transfer coefficient distribution.For comparison purposes, an analysis of a typical oil-quenching process was also performed.

Fig. 13 through 15 show the resulting hoop stressesfor the three cases, all normalized with respect to themaximum tensile stress of the oil-quenched part. Thestresses are highest for the oil-quenched disk, closelyfollowed by the non-optimized fan-cooled disk withuniform high fan blowing all around. The residualstresses for the optimized disk, in turn, are considerablylower, almost by one order of magnitude compared tothe oil-quenched part in terms of tensile stresses. Thereductions in compressive stresses are not quite thatlarge, but still by a factor of between six and seven.These results clearly show the advantage of anumerically optimized fan cooling process compared tothe traditional oil-quenching. Ref. 15 describes the heattreatment optimization process in greater detail. Thesefindings were confirmed during multiple runs withdifferent starting points on actual geometries whichare of proprietary nature and cannot be shown here.The formulation of the objective function as a quadraticclearly aids in this behavior.

Fig. 13: Hoop Stress, Oil-Quench Process

Fig. 14: Hoop Stress, Starting Point

Fig. 15: Hoop Stress, Final Configuration

AIAA-98-4734


Thermo-Mechanical Design

The mechanical design and engineering analysisportion of the integrated process is currently laggingbehind the efforts on the processing side. This hasseveral reasons, one being that the development of fullyparametric master models proved to be more time-consuming than anticipated. But a major bottleneck isthe automatic generation of high-fidelity finite elementanalysis models complete with loads and boundaryconditions which update with parametric changes of themodel. Several pilot projects have been ongoing sincethe last year, evaluating different concepts of relatinganalysis-related information to the geometry. Oneapproach is the “tagging” of the geometry, applyingbasically CAE-type information on the geometry on theCAD side. A major hurdle here is reluctance from theside of analysis engineers who rather want to workwithin their CAE tool of choice instead of the CADsystem. Also, the processing of “tags” inside the CAEtool has proven not to be very robust. Anotherapproach is the use of scripts for the CAE tool, wherethe engineer prepares the model once manually and thensaves the session log file for subsequent reuse. Thisapproach demands entity-consistent import of thegeometry from the CAD tool into the CAE system,which again is not robust at the moment. This approachcertainly breaks down in the presence of topologicalchanges. Before the issue of reliable, repeatableautomatic generation of analysis models for complex 3-D-geometries is resolved, any effort to use optimizationon the mechanical design side beyond conceptualstudies is premature.

Outlook

The plan is to complete one full manufacturingprocess exercise by the end of the year. How fast thedevelopments on the mechanical design side will beable to catch up remains to be seen and depends largelyon external factors beyond GE CRD’s control. In orderto reduce analysis times for the forging optimization,the use of approximate models and response surfaceswill be investigated. The machining simulation will beintegrated with the heat treatment optimization package,so that the final machining distortions will be availableautomatically without manual intervention. Once thesystem is in place for the complete process simulationand process optimization, the question of theapplicability of formal MDO algorithms will berevisited.

In parallel, various strategies will be furtherinvestigated on how to capture analysis modelinformation and make it reusable in a robust fashion so

that analysis models for complex geometries will finallyautomatically update without human intervention. Oncethis obstacle has been cleared, 3-D-shape optimizationduring the mechanical design phase can be addressed,probably initially limited to relatively simple featurescomparable in complexity to the 2-D-forging shapeoptimization discussed earlier.

Computer resources continue to be a problem inconjunction with the long analysis times required for thesolution of industrial size problems. A forgingsimulation as it is considered here may take 6 to 7 hourson an SGI workstation. Finite differencing couldpotentially be done in parallel, but there are theproblems of software licensing and maximum numberof processes one user is allowed to run at any giventime. It seems clear that the computer resource issueswill remain a major bottleneck for the application ofMDO to industry problems.

One of the highlights so far in this project has beenthe optimization and integration framework itself,iSIGHT, which has performed very well, although stillunder development. It fits with GE’s paradigm shiftaway from proprietary software development to the useof commercially available CAD and CAE tools, whichrequire a loose and non-intrusive coupling of theindividual analysis modules.

References

[1] A Collaborative Optimization Environment (COE)for MADE, Technical Proposal, General ElectricCorporate Research and Development andOptimum Technologies, Inc., August 1995

[2] iSIGHT Version 3.0 User Manual, EngineousSoftware, Inc., Raleigh, NC, 1997

[3] Wujek, B.; Renaud, J.E.: Design DrivenConcurrent Optimzation in System DesignProblems Using Second Order Sensitivities, 5th

AIAA/NASA/USAF/ISSMO Symposium onMultidisciplinary Optimization, Panama City, FL,September 1994

[4] Braun, R.D.; Kroo, I.M.: Development andApplication of the Collaborative OptimizationArchitecture in a Multidisciplinary DesignEnvironment, Aug. 1995

[5] Engineous User Manual, General ElectricCorporate Research and Development,Schenectady, NY, 1995

[6] Lee, H.; Goel, S.; et al.: Toward Modeling theConcurrent Design of Aircraft Engine Turbines,Presented at the International Gas Turbine and

AIAA-98-4734


Aeroengine Congress and Exposition, Cincinnati,OH, May 1993

[7] Powell, D.J.: Inter-GEN, A Hybrid Approach toEngineering Optimization, General ElectricTechnical Report, Feb. 1991

[8] Design Environment for Forming (DEFORM),Version 5.0, Online Users Manual, ScientificForming Technologies Corporation, Columbus,OH, August 1997

[9] Kumar, V.; German, M.D.; Srivatsa, S.: DesignOptimization of Thermomechanical Processes withApplication to Heat Treatment for Turbine Disks,Presented at the Manufacturing InternationalConference, Atlanta, GA, 1990

[10] Unigraphics V13 Online User Documentation,EDS, Cypress, CA, Oct 1997

[11] Tappeta, R.; Nagendra, S. et al.: Concurrent Sub-Space Optimization (CSSO) MDO Algorithms iniSIGHT; GE CRD Technical Report, January 1998

[12] Conversations with S. Kodiyalam, EngineousSoftware, Raleigh, NC

[13] He, B.; Röhl, P.J. et al.: CAD and CAE Integrationwith Application to the Forging ShapeOptimization of Turbine Disks. To be Presented atthe 39th AIAA/ASME/ASCE/AHS/ASC Structural,Dynamics, and Materials Conference, Long Beach,CA, April 1998

[14] Vanderplaats, G.N.: ADS - A FORTRAN Programfor Automated Design Synthesis, Version 3.00,VMA Engineering, March 1988

[15] Röhl, P.J.; Srivatsa, S.K.: A ComprehensiveApproach to Engine Disk IPPD. Proceedings of38th AIAA/ASME/ASCE/AHS/ASC Structural,Dynamics, and Materials Conference, pages 1250-1257, Kissimmee, Florida, April 7-10, 1997

AIAA-98-4733


BOEING ROTORCRAFT EXPERIENCE WITHROTOR DESIGN AND OPTIMIZATION

Frank Tarzanin*

Darrell K. Young†

The Boeing Company

Philadelphia, Pennsylvania

AbstractThis paper reviews 12 years of progress in applying optimization to the helicopter rotor design problem. This involves multiple

disciplines, multiple objective functions, a large number of design variables and irregular design space. The initial step was todevelop a single interdisciplinary analysis to evaluate the objective function. By understanding the problem, approaching it in-crementally and learning how to adapt optimization techniques, dramatic progress has been made. Numerous optimizationtechniques have been tried, including: gradient-based methods (with finite difference and automatic differentiation), biologicalmodels, surface approximations and direct search. Each of these methods had to be properly adapted to the problem. Initial prog-ress was made using a gradient-based method along with numerous “prodding” techniques to avoid local minima. Thoughsuccessful, it required extensive labor hours. In search of more efficient methods, a scaled down representative problem was de-fined and multiple derivative free optimization (DFO) methods were investigated. All this has led to a hybrid approach that weare currently using in rotor design.

Introduction1

elicopter rotor design is a complex interdiscipli-nary process. Optimization was applied to thisprocess with two major objectives in mind. First,

to define a rotor with improved characteristics (lowerloads, longer life, reduced weight, lower vibration, bet-ter aerodynamic performance) and second, to automatethe rotor design process to reduce labor hours and de-sign cycle time. Achieving these objectives requiresmany steps. As a first step, we chose to focus only onthe lower vibration aspect, which is an ambitious start-ing point with large potential benefits. The plan was toincrementally build upon this base, adding more com-plexity at each step.

Historically, a major problem in the rotorcraft indus-try has been vibration. The primary cause of thisvibration is the hub loads coming from the rotor. Thetransformation of the rotating vibratory hub loads intothe fixed system causes a selective cancellation andaddition. This results in fixed system vibratory hubloads at frequencies that are integer multiples of thenumber of rotor blades times the rotor speed. This fre-quency is represented as NP, where N is the number ofblades and P represents the frequency of rotation. Thefixed system vibratory vertical (Fz), longitudinal (Fx),and lateral (Fy) forces along with the roll (Mx) and

* Manager, Dynamics and Loads† Senior Technical Specialist1 Copyright ©1998 The American Institute of Aeronau-tics and Astronautics Inc. All rights reserved.

pitch (My) moments at the rotor hub excite the fuselagecausing vibration. The resulting vibration annoys andfatigues crew and passengers, cracks structure, and failscomponents and electronics. Collectively, this contrib-utes significantly to operating cost and safety. To keepvibration reasonable, though still at undesirable levels,devices are added to most helicopters. These includeabsorbers in the fixed and rotating system, isolation andactive force generators together with fuselage structuraltuning. All these devices add cost, complexity andweight.

As a result, substantial research has been performedto reduce the inherent vibratory hub loads that causeaircraft vibration. The Ref. 1 research and unpublishedwind tunnel testing showed substantial potential forreducing rotor vibratory hub loads by more than 50 per-cent when the blade tip was swept. However, thenumber of design variables, the interaction between thefive different hub load components (vertical force, in-plane forces and inplane moments), the real designconstraints, and four years of trying convinced us that atrial-and-error, follow-the-logic approach would notwork. Only computer-automated optimization couldefficiently juggle all the variables and find its waythrough the conflicting requirements.

The objectives of this paper are to describe the stepstaken thus far in the development of our rotor designtool. We will describe the various optimization tech-niques tried to date and show how they are being usedto design low vibration rotors (LVR). We will presentour experiences, including lessons learned, as applied to

H

AIAA-98-4733


using the various optimization techniques. The resultsfrom wind tunnel model rotor tests are included to showthe benefits achieved from the design process. We willalso briefly describe recent activities undertaken withvarious researchers applying derivative free optimiza-tion (DFO) techniques to this problem. Finally, wedescribe some of our future plans.

The Rotor ProblemThe helicopter rotor represents the classic aeroelastic

problem. Figure 1 plots angle of attack versus Machnumber for different blade stations. Each loop in thefigure represents the travel of one blade station throughone rotor rotation. The blade encounters transonic flow,stall, reverse flow (the angle of attack exceeds 180 de-grees) and unsteady effects, including dynamic stall(since the blade performs multiple revolutions eachsecond). As the blade rotates, the large changes in dy-namic pressure and angle of attack result in largevariations in lift. This, in turn, results in trailed and shedvortices leaving the blade as shown in Figure 2. Bladesthat follow run into this complex wake, referred to asnon-uniform downwash, resulting in further lift varia-tions. In addition, since the rotor blade is long andslender, there are substantial elastic deformations, in-cluding nonlinear structural dynamics such as radialshortening, Coriolis forces and bending-torsion cou-pling. Therefore the airloads are functions of the aircraftflight condition, the non-uniform downwash and theelastic deflections of the blade. Clearly, there is no hopeof predicting rotor behavior with a loosely coupledanalysis. Figure 3 shows the close coupling required toperform a complete rotor analysis.

Historically, the aerodynamics, flying qualities, dy-namics and acoustics departments develop and maintain

separate simulation codes for performing their tasks.The aerodynamics department is responsible for rotorperformance and aero-acoustics, and developing newtechnology for airfoils, non-uniform downwash predic-tion and blade/vortex interaction. The dynamicsdepartment is responsible for rotor vibratory loads andstability, and developing aeroelastic models (blade cou-pled dynamic response and unsteady aerodynamics).The flying qualities department is responsible for theflight control laws and developing full aircraft trim the-ory. We all are trying to solve the same problem, butwith different emphasis.

Each simulation has to contain most of the problemelements, but not necessarily all or the best. For aero-acoustic predictions, the blades were assumed rigid; forperformance and trim predictions, approximate bladedeflections were used; and for vibratory loads, simpli-fied (quick running) downwash models were used. Inthe late 1980’s the development of code configurationmanagement tools (like DSEE2 and later ClearCase3),increased computer power, and relentless cuts in devel-opment budgets forced a consolidation.

The aerodynamics, acoustics and dynamics depart-ments then combined their best technology into a singleinterdisciplinary rotor code4. Code configuration man-agement tools allowed each department to continue todevelop and enhance their traditional areas of expertiseand be able to utilize a simulation code that had all thebest technology and was superior to any of the previoussimulations. Faster computers and the proliferation ofaffordable workstations lessened the need to simplifyportions of the theory to reduce run time and turn-around. Program options allow less rigorous, quickerrunning versions to be used when needed.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-6 -4 -2 0 2 4 6 8 10 12 14 16 18 20

Instantaneous Angle of Attack (deg)

Lo

cal M

ach

Nu

mb

er

Stall

Blade Tip84% Span

50% Span

Blade Root

Figure 1. The challenge. Figure 2. Trailed and shed vortices.

AIAA-98-4733


- CALCULATE WAKE • FIXED ROTOR RESPONSE • FIXED CONTROLS

- DETERMINE RESPONSE • FIXED AIRLOADS • FIXED CONTROLS

MAIN ITERATION LOOP

- DOWNWASH

- AIRLOADS

- CONTROLS

- AIRLOADS

- FLEXBEAM DEFORMATION

- CONTROL MOTION - END LOADS

- DETERMINE CONTROLS • FIXED ROTOR RESPONSE • FIXED DOWNWASH

NON-LINEAR FLEXBEAMINDUCED

WAKE

LINEAR FUSELAGE RESPONSE

TRIM CHECK

- HUB MOTION

- HUB LOADS

- ROTOR RESPONSE

- NON-LINEAR LOADS

NON-LINEAR BLADE

COUPLINGAIRLOADS ROTOR

RESPONSELO

OP

LOO

P ROTOR RESPONSE

CHECK AIRLOAD CONVERGENCE

TO PROCEED

CHECK RESPONSE

CONVERGENCE

END

AIRLOADS

Figure 3. Rotor analysis.

Though flying qualities was not part of the initialsimulation consolidation, provisions were made to makethe interface with the trim model more robust. By satis-fying rotor trim forces, instead of postulating controlinputs, we are assured that the fuselage force and mo-ment balance is maintained. We plan to link thecombined interdisciplinary rotor analysis with the trimanalysis in the future.

Therefore, our present function evaluator is a single,tightly coupled, interdisciplinary rotor analysis. An it-eration method is used to satisfy compatibility amongall the disciplines.

The Optimization ProblemOur first step was to prove that the optimization proc-

ess worked by defining a rotor with significantlyreduced vibratory hub loads, building a Mach scaledmodel and performing a validation test. Once validated,the second step was to obtain a better design with ac-ceptable risk and cost. Since there are so manyconflicting intangible requirements like manufacturing,total ownership cost, tolerance to variability, etc., weneeded to define the design space so that the designteam could find an acceptable compromise. Performingmultiple point optimizations to define the local designspace as a function of key variables would do this.

For our first attempt, we linked the rotor analysis, (todefine the objective function), with a gradient-basedoptimization code, NPSOL5. This code is what we willrefer to as the gradient optimizer throughout this paper.The rotor blade is typically modeled with 25 structuralelements. There are six design variables for each ele-ment, which are listed in Table 1. Hence, there is a totalof 151 available design variables (there is one extra

design variable for the control system stiffness). Theobjective function (OF) is made up of a weighted linearsum of the five hub load components as follows:

O F W F W F W F W M W Mx y z x y= + + + +1 2 3 4 5(1)

where

F fN D n D C, , ,= ∑ (2)

M mN D n D C, , ,= ∑ (3)

Wi is a coefficient for weighting the hub load compo-nents, so as to account for fuselage response due to eachhub load. F and M are the fixed system hub forces andmoments at N times rotor speed (where N is the numberof blades), in D directions (x, y, z). Equations 2 and 3represent the transformation of the rotating systemblade root forces and moments, (f and m), in directionD, at frequency n, and flight condition C, into the fixedsystem forces and moments (F and M). Due to thistransformation of rotating hub load components into thefixed system, there can be a shift in the frequency ofone times the rotor speed. Therefore, the rotating fre-quency n may be at a frequency of (N-1), N, or (N+1)times the rotor speed.

This objective function formulation results in verycomplex design space. Since each component of theobjective function has a different trend, due to a designvariable change, the design space will have many peaksand valleys. Therefore, finding the lowest valley is ademanding challenge.

AIAA-98-4733


Table 1. Rotor design variables.

Symbol Description

m section masscg section chordwise center of gravityEIβ section flap bending stiffnessEIζ section lag bending stiffnessGJ section torsion stiffnessΛ built in sweep angle between sectionsKz control system stiffness (only one value)

Initial Optimization Using GradientsOne major complication with the optimization proc-

ess is the large demand for computer resources. Sincefinite differences are used to determine derivatives andthere are 151 design variables, an optimization wouldrequire thousands of function evaluations. With runtimes of 20 to 30 minutes for each evaluation (on a HP715/100), and questions of numerical noise, we decidedto use a simple, less costly approach.

The reason the function evaluation is so computerintensive is that the airloads are a function of both theaircraft flight condition and the rotor blade elastic de-flections. The elastic deflection with the largestinfluence on the airloads is the blade elastic twist. It washoped that by fixing (not varying) the design variablesthat influence elastic twist, such as the torsional stiff-ness, chordwise center of gravity, aerodynamic centerand chord sweep, the elastic effect on the airloadswould be minimal and could be ignored. This wouldmean that the airloads could be assumed to be only afunction of the flight condition. Therefore, the airloadscould be prescribed and the rotor blade structural prop-erties optimized to minimize the resulting vibratory hubloads. Bending-torsion coupling would still be ac-counted for, but changes should be minimal, so thatairload changes resulting from this coupling would besmall and not prevent us from finding a good dynamicresponse optimum.

A new, simpler function evaluator was made from therotor blade dynamic response portion of the rotor analy-sis. The airloads were read into the simpler function boxas a prescribed forcing function. The vibratory hubloads were calculated from the new function evaluator.Since the airloads were held fixed at the initial distribu-tion in the optimization, the blade geometryaerodynamic configuration was also fixed. Eliminatingthe torsion degree of freedom from consideration re-duced the number of design variables to the sectionmass, flap stiffness and lag stiffness at 25 blade stations.It turned out that the chord stiffness did not vary withthe optimization process for reasons we do not under-

stand. So effectively, there were only 50 designvariables.

This simplification allowed the function evaluator torun in seconds. However, the optimizer still gave lack-luster results. It would run through a few optimizationiterations and proclaim victory, but usually the reduc-tion in hub loads were less than twenty percent. It wasclear that the gradient-based algorithm was getting stuckin local minima.

The problem is that a gradient optimizer cannot find asolution far from the initial design if the design spaceresembles the Rocky Mountains. There was no mecha-nism for a gradient-based optimizer, which follows asteepest decent, to search on the other side of a responsepeak, (which is perceived as first going up hill).

To resolve this problem, numerous techniques weredeveloped to encourage the optimizer to avoid localminima. These techniques are described in more detailin the following subsections.

Different Starting FrequenciesBlade properties were changed to get different start-

ing frequencies. By making random variations to thephysical properties, new starting designs were found forthe optimizer. These new designs were generated inhope of forcing the optimizer to follow a differentsearch path. This path would either lead to the samelocal minimum, a different local minimum, or to theglobal minimum.

Large Range of Design Variable ValuesBy changing the range between the upper and lower

bounds of the design variables, it is possible to encour-age large changes in the design value. These largechanges would often cause the optimizer to explore anew design space, which resulted in finding a moreglobal minimum. Once a good solution was found, wewould then squeeze the range down until we achieved asolution, which was the best compromise between hubload level and ease of manufacturing the rotor blade.

Apply Constraints IncrementallyThis technique goes hand-in-hand with the large

range of the design variables described above. By al-lowing the upper limit of a constraint, such as the totalrotor weight, to be large at the start, it is possible to getinto another region of the design space. Just as de-scribed above, once a good solution is found, theconstraint would be squeezed to slowly force the solu-tion into the desired design space.

AIAA-98-4733


Adjust Objective Function Weighting ValuesAnother technique that can be used to foster new so-

lutions is varying the relationship of the weightingcoefficients in equation (1). For example, if one of thehub load components is resistant to change, all of theother coefficients can be set to zero and the problemrerun. Once the optimizer has been forced into a newregion, the original weighting coefficients can be usedagain to continue exploring the design space. Anotherapproach is to increase or decrease the importance of agiven hub load over that of the others to encourage fur-ther improvement.

Recalculate Constant AirloadsAs described above, the airloads are a function of

both the aircraft flight condition and the blade elasticdeflections. The initial optimization process did notallow the design variables that influence elastic twist tovary and the airloads were held constant. Our practicewas to verify any design solution obtained from theoptimizer in our full interdisciplinary rotor analysis.While the assumption that the effect of bending-torsioncoupling would not prevent us from finding a good dy-namic response was true, there were times where a stillbetter solution was found by simply updating the air-loads and continuing the optimization process.

CompetitionsThis technique uses the different solutions, which

have been generated by the above prodding techniques,to compete against each other. Individuals were givendifferent starting designs and tried to improve the opti-mum. Weekly meetings would share lessons learned,eliminate the worst and continue refining the best. Thecomparison included items such as the objective func-tion value, how well the design satisfied the constraints,and how realistic the blade section properties were.

Identify Related DesignsOne observation is that as the number of potential de-

signs increases, there will be promising designs withsimilar characteristics. By grouping similar designs to-gether, not all of the competitions will need to beperformed. This is important when time and computerresources are limited, since it is easier to eliminate adesign then to perform the competitions and determinewhich ones to keep.

Optimizer RestartWhen the optimizer terminates, the history of the ob-

jective function is reviewed. Usually this history showsan initial rapid reduction followed by a gradual levelingout. However, some times the objective function would

still be declining when the optimizer would stop. Therewould not be the typical leveling out. When this oc-curred, the optimizer would be restarted and usuallycontinued reducing the objective function. We suspectthat the premature termination of the optimization is dueto contamination of the Jacobian. Since the Jacobian isbuilt from a finite difference process and uses currentand historical data, noise in the numerical gradientscould cause the contamination. Restarting allowed anew Jacobian to be generated and the determination ofclear direction for the process to proceed.

Multiple Flight ConditionsWe wanted a blade design that was robust over the

whole aircraft flight regime, not just a single designcondition. This is important since the rotor must operateover a wide range of airspeeds, altitudes, ambient tem-peratures and gross weights. By performing completeairspeed sweeps at multiple gross weights, we were ableto select up to five critical flight conditions to include inthe objective function. This virtually insured that theoptimum would lower vibration over the whole flightregime. Typical selections included cruse at two grossweights, transition, and the corner of the flight enve-lope.

Initial Wind Tunnel ModelThe procedure described above was developed and

refined by applying it to the design of a Mach-scaled,four-bladed, fully-articulated, ten-foot diameter windtunnel rotor which was fabricated and tested in ourV/STOL wind tunnel6. The wind tunnel test allowed thegathering of steady-state vibratory rotor loads necessaryto validate the low vibration rotor concept. The goalwas to develop a rotor, which would substantially re-duce the fixed system 4P vertical hub load and the fixedsystem 4P roll and pitch hub moments. Accomplishingthis goal required the design and fabrication of two ro-tor blade sets — a reference rotor and a low vibrationrotor. Both rotor blade sets would then be tested back-to-back in the wind tunnel.

Both rotor sets had identical blade radius, chord,twist, and airfoil shape distributions, as well as the sameblade and hub attachment points. The only parametersthat differed were the spanwise and chordwise distribu-tion of the rotor mass and elastic properties. Thereference rotor is a scaled version of the Boeing Model360 experimental rotor, which flew to over 210 knots inlevel flight on an all-composite tandem rotor demon-strator aircraft. This rotor was designed by using thetraditional approach of adjusting the rotor properties toprovide adequate frequency separation from the har-monic aerodynamic forcing. The low vibration rotor

AIAA-98-4733


was designed using the optimization techniques as de-scribed above.

A comparison of the measured normalized 4P hubloads, obtained from dynamically calibrated balances,for the reference rotor and the low vibration rotor(LVR) is shown in Figures 4 and 5 for a level flightcondition corresponding to a nondimensional rotor lift,CT'/σ of 0.07 and a nondimensional rotor propulsiveforce, X , of 0.08. (The prime symbol is used through-out this paper to indicate a deviation from the classicaldefinition of the marked quantities). The forces havebeen normalized by the nominal rotor thrust and themoments have been normalized by twice the nominalrotor thrust times the dimensional flap hinge offset. Thehub loads are plotted versus rotor advanced ratio, µ′,(which is defined as the free stream velocity divided bythe rotor tip speed). Figure 4 shows that a 67 percentreduction was achieved in the 4P vertical hub load inthe low airspeed transition region (µ'≈0.10 or about 39knots), and a 56 percent reduction was achieved in thehigh airspeed region (µ'≈0.43 or about 183 knots) for a3.4 percent increase in total rotor flapping weight.

Figure 5 shows that a 45 percent reduction wasachieved in the measured 4P overturning hub momentin the low airspeed transition region (µ'≈0.10), and a 77percent reduction was achieved in the high airspeedregion (µ'≈0.43). The overturning hub moment refers tothe magnitude of the vector sum of the roll and pitchhub moments.

The initial wind tunnel model design was a success. Ithad meet our goal of proving that the optimization pro-cess worked in defining a rotor with significantlyreduced vibratory hub loads. It also showed us howlabor intensive the optimization process could be. Whilethe gradient-based approach had been successful, it leftus looking for a better way of finding the global mini-mum. We were just getting started on the literaturesearch when a new opportunity came along. We wereasked to apply our optimization techniques to definingan advanced rotor for the CH-47 Chinook.

A Real RotorThe Mach scaled wind tunnel test results were so en-

couraging that funding was found to apply theoptimization to an advanced CH-47 Chinook rotor. Thedevelopment of a full-scale low vibration rotor was un-dertaken to understand and evaluate the rotordesign/optimization process needed to satisfy all the fullscale requirements. This included considerations like

0

0.1

0.2

0.3

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

4P V

erti

cal H

ub

Lo

ad (

ND

)

Reference Rotor

LVR

Figure 4. Measured 4P vertical hub load.

0

0.05

0.1

0.15

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

4P O

vert

urn

ing

Mo

men

t (N

D) Reference Rotor

LVR

Figure 5. Measured 4P overturning hub moment.

blade tie down fittings, track and balance hardware,fatigue life, tooling and manufacturing requirements.

In addition, the optimization problem was reformu-lated to include additional hub loads and constraints.Another wind tunnel test7 using the previously definedLVR showed that the 8P hub loads could be measuredand predicted well enough to warrant design optimiza-tion to reduce these loads as well.

The improved optimization method was applied tothe design of the advanced Chinook rotor8. It involvedworking with the designers to define realistic minimumand maximum limits for each design variable. Iteratingwith manufacturing was required to insure that the finaldesign was buildable. In addition, the same prodding

AIAA-98-4733


techniques, described above, were used with the samegradient-based optimizer.

Since rotor design is a high cycle fatigue problem, thestress group periodically checked the stress/strain lev-els. To insure that the rotor had an infinite life, aconservative strain allowable was adapted. This strainwas not to be exceeded during normal level flight.Whenever the strain was too high, the minimum bladesection stiffness and weight was adjusted to lower thestrain.

As the design/optimization progressed, this iterationbetween the optimum design, blade loads, stress andadjusted minimum stiffness and weight proved fruitless.Each time the optimizer defined a significant vibrationimprovement, the stress proved too high and the designconstraints were adjusted. This process was increasingboth our design time and cost. We either had to proceedwith a less than optimum design or modify the optimi-zation process.

Therefore, the optimization process was modified toinclude a maximum strain constraint. In addition, a re-lationship between the blade section stiffness andweight was also provided as a nonlinear constraint. Thissimulated the design process of strengthening the bladewhen the stress was too large. When additional strengthwas needed, the optimizer automatically added the cor-rect weight. This made a real solution possible. Asshown below, we achieved both lower vibrations, atboth 4P and 8P, with a reduced blade weight.

Another issue was the determination of local designspace. This would allow the design team to perform atradeoff between total rotor weight, vibration and strain.Point optimizations were performed where the totalrotor weight and strain constraints were incrementallydecreased while satisfying all other constraints. Byplotting the optimization results as a function of vibra-tion versus constrained blade weight and allowablestrain, the tradeoff between weight, strain, and vibrationcould be more clearly understood. Using this data wecould choose the best compromise.

Figure 6 shows the 4P blade vibration index versusnondimensional blade weight for the design strain leveland for a 30 percent larger strain level. The 4P vibrationindex is a normalized measure of the calculated 4P ver-tical force, roll moment and pitch moment, times thepilot vertical vibration response to hub loads as meas-ured from an aircraft shake test. Hub loads from boththe forward and aft rotors at 20 and 150 knots wereused.

Two baseline rotors are shown. The reference modelrotor (solid square) has a weighted vibration indexbased on calculated hub loads and is normalized tounity. The full scale Model 360 nondimensional

.2

.4

.6

.8

0

1.0

0.020 0.021 0.022 0.023 0.024 0.025 0.026Blade Weight

Thrust (@C /σ = .07) x Scale FactorT

Weighted4P

VibrationIndex

Full ScaleModel 360

Wind TunnelBaseline

NewLVR

Design Strain Limit

30% Larger Strain Limit

OriginalLVR

Weight

7.5%

3.4%

2.7%

Figure 6. Weighted vibration index versus blade weighttrend for two strain levels.

flapping weight is included for reference to show howclose the scaled weight of the model and full scale ro-tors are. The nondimensional weight of the originalLVR model rotor (previously discussed) is also pro-vided for reference. This rotor is 3.4 percent heavierthan the reference model rotor.

Using this design space definition, a blade weight atthe knee of the curve for the lower strain was selected.This represents a flapping weight that is 7.5 percentlighter than the reference rotor and 11.9 percent lighterthan the original LVR.

Observe that if desired, a new material with a higherstrain allowable could be identified and qualified, or afinite blade fatigue life defined. This would result infurther reduced rotor weight and/or lower vibration, butwith increased cost and development risk.

The final rotor properties were Mach scaled and awind tunnel test was performed in the same manner asthe previous tests. The improved LVR used a hub with acoincident elastomeric bearing. Due to model elastomerbearing size limits, it was not possible to get the modelflap hinge at the same offset as the previously describedreference rotor. Therefore, to compare with the refer-ence rotor hub loads the improved LVR measured hubmoments are scaled to account for this difference.

Figures 7 and 8 show the measured normalized 4Phub loads, obtained from dynamically calibrated bal-ances, for the reference rotor, the LVR, and theimproved LVR at the same flight condition. Comparedto the reference, the improved LVR shows that the 4Pvertical hub load is 74 percent lower in transition and69 percent lower in cruise for a 7.5 percent decrease intotal rotor flapping weight. The 4P overturning hubmoment is 88 percent lower in transition and 55 percent

AIAA-98-4733


0

0.1

0.2

0.3

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

4P V

erti

cal H

ub

Lo

ad (

ND

)

Reference Rotor

LVR

ILVR

Figure 7. Measured 4P hub load for the reference rotor,LVR and ILVR.

0

0.05

0.1

0.15

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

4P O

vert

urn

ing

Mo

men

t (N

D) Reference Rotor

LVR

ILVR

Figure 8. Measured 4P overturning hub moment for thereference rotor, LVR and ILVR.

lower in cruise. The 8P hub loads (shown in Ref. 8) hada 1 to 68 percent reduction in vertical hub load and a 53to 79 percent reduction in overturning hub moment

Solve the Whole ProblemWe achieved our objective of proving the optimiza-

tion process works with a simplified function evaluator.A rotor that significantly reduced vibratory hub loadswas defined and validated in a wind tunnel test.

It was now time to improve the process. This in-volved several enhancements. First, the fullinterdisciplinary rotor analysis code was used as thefunction box, allowing us to investigate the full aero-elastic optimization problem. Now the additional design

variables that cause elastic blade twist, which leads tochanges in the airloads, could be exploited by usingoptimization. This increased the number of design vari-ables from 50 to 151.

The second enhancement reduced optimization turn-around time. A typical function evaluation with aprescribed set of airloads (not changing due to bladeresponse) was a few seconds. When the full interdisci-plinary rotor analysis was used (with compatibilitybetween the blade response, airloads and rotor wake),the time increased to about 20 or 30 minutes. Hence,nearly all of our processing time would be spent evalu-ating the finite differences. To perform a singleoptimization using the full theory would require monthsof run time. Two developments that helped overcomethis computer time problem were the continuing work-station speed increase and the use of parallel processingto evaluate each gradient independently9.

The third enhancement focused on non-gradientbased optimization methods. As pointed out above, sev-eral prodding techniques were needed to encourage agradient-based optimization method find a global mini-mum. This was very time consuming and laborintensive. It was hoped that non-gradient methodswould prevent getting stuck in local minima by provid-ing a diverse set of potential optimum solutions. Thesepotential solutions would be found by exploring thewhole design space, instead of being limited to the localspace of the initial design, like gradient methods. Thebest non-gradient designs would then be refined usinggradient methods. This approach is equivalent to flying-over the Rocky Mountains to identify the most promis-ing valleys, then sending in explorers to search for thebottom of each valley.

To help us explore the many nongradient-based opti-mization methods, a simple design problem wascreated. This problem was then given to various re-searchers so that they could apply their nongradient-based methods to the same rotor problem. They werealso asked to use our enhanced function box evaluator.

The problem represents a three-bladed helicopter ro-tor with advanced planform geometry including bladetip sweep. Early optimization efforts had shown that itwas more difficult to reduce vibration for a three-bladedrotor than for a four-bladed rotor. The rotor was discre-tized into a model consisting of 13 bays of which the 10outboard bays had airloads applied. Normally 25 baysare used. This problem had 56 design variables whichrepresented the level of section mass, stiffness (in flap,chord, and torsion), and chordwise center of gravityposition at different blade stations along the span of theblade. Further CPU run time reductions were obtainedby prescribing the rotor induced non-uniform down-

AIAA-98-4733


wash and using only two flight conditions in the func-tion evaluation. This simplified model was chosen sinceit captured the main effects of the vibration problem andstill had a rather short function evaluation CPU time ofa few minutes per airspeed.

The objective function to be minimized was the linearcombination of the weighted fixed system 3P and 6Pthree hub forces (Fx, Fy, Fz) and two hub moments(Mx, My). The 3P loads were weighted as being twiceas important as the 6P loads and the two airspeeds wereweighted equally. The only constraint limited the totalrotor weight to be less than or equal to 1.685 times thenominal weight.

The methods explored were:1) design of experiments (DOE) with response

surfaces by Boeing, Seattle10,11

2) evolutionary programming (EP) by Boeing,Philadelphia9

3) parallel direct search (PDS) by Boeing Seattle,IBM, and Rice University11,12

4) analytical derivatives using ADIFOR by NASA,Langley13

5) derivative free optimization (DFO) by IBM11,14

6) genetic optimization (GA) with a neural net byRensselaer Polytechnic Institute (RPI)

Table 2 compares the results obtained from each ofthe methods along with the results from our gradient-based method using NPSOL. Please note that the resultspresented here were obtained prior to the end of 1997and that more recent results may be shown at this con-ference by the individual researchers. Also, we will notprovide detail of how each researcher obtained theirresults. That too is left for the papers they will presentat this conference.

Note that the Table 2 results are not global minimumand need to be refined with gradient methods (exceptfor NPSOL and ADIFOR, which are gradient methods).Even though they are not minima, three of the deriva-tive free methods; EP, PDS, and GA have objectivefunction values lower than the best gradient result.

Table 2. Comparison of the resulting designs.Design Nondimensional

Total WeightObjective Function

ValueBaseline 1.000 1.000NPSOL 1.452 0.559DOE 1.680 0.644EP 1.635 0.487PDS 1.289 0.501ADIFOR 1.323 0.564GA 1.685 0.512

Our experience, to date, has been that a nongradient-based method by itself is not the fastest way to reach aglobal minim. Because the function to be evaluated iscomputation intensive and many function evaluationsare needed, a combination of methods is required. Byautomating a combined process, labor costs can begreatly reduced. For now, we have selected a hybridapproach that uses our EP method and our gradient-based method. We have chosen these two for the fol-lowing reasons. First, we have both codes in house andhave some, albeit limited, experience in using them.Second, the other methods are still under investigation.It is possible that multiple methods may be needed.Third, Table 2 shows that the EP method gave the bestresults.

The advantage of using a hybrid method is that thenongradient-based method provides a diverse set ofsolutions, which explore the whole design space withouthaving to use prodding techniques. These diverse solu-tions are also automatically generated by the processitself and do not require labor intensive human inter-vention.

Recent Design ActivityRecently (last quarter of 1997), we were asked to de-

fine a replacement rotor for an existing helicopter. Thenew rotor would have a 12 percent larger blade chordand a 67 percent increase in blade twist but the rotorvibration could not be any higher than the existing rotorvibration. Historically, when a rotor blade has its chordincreased, the section airloads increase, thereby in-creasing the rotor loads. In addition, increasing theblade twist also causes increased hub loads. A conven-tional preliminary design had been performed prior toour involvement, and the predicted vibration was sub-stantially higher than the existing aircraft.

Using the hybrid method described above along withthe complete interdisciplinary rotor analysis, the evolu-tionary programming method defined a promising rotor.It satisfied the vibration requirement, but was heavy anddid not satisfy all the constraints. This was expected,since the initial goal was to find potential candidates,not the final design. The gradient-based method wasutilized to improve this design. The weight was system-atically reduced and the strain constraint applied. Adramatic improvement was made while reducing thetotal rotor weight by 7 percent. Figures 9 to 11 showpreliminary results of the normalized hub loads for aLVR compared to the existing production rotor.

AIAA-98-4733


0

0.02

0.04

0.06

0.08

0.1

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

NP

Ver

tica

l Hu

b L

oad

(N

D) Production Rotor

LVR

Figure 9. NP vertical hub load reduction.

0

0.02

0.04

0.06

0.08

0.1

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

NP

Inp

lan

e H

ub

Lo

ad (

ND

)

Production Rotor

LVR

Figure 10. NP inplane hub load reduction.

0

0.02

0.04

0.06

0.08

0.1

0 0.1 0.2 0.3 0.4

Advance Ratio, µ'

NP

Ove

rtu

rnin

g M

om

ent

(ND

)

Production Rotor

LVR

Figure 11. NP overturning hub moment reduction.

ConclusionsThe rotor design problem involves a large number of

design variables, interdisciplinary considerations andcomplex design space. The function evaluator is a sin-gle, tightly coupled, interdisciplinary, computationintensive code. Steady progress has been made towardsdeveloping an effective optimization-based rotor designprocess. However, the process requires excessive com-puter resources, long calendar time and is too laborintensive.

To date derivative free optimization (DFO) hasshown the greatest promise in improving the rotor de-sign process. By using these methods to explore thewhole deign space, we are able to get many variedstarting designs for refinement with our gradient-basedmethod, and save substantial labor hours previouslyspent avoiding local minima. This hybrid approach hasincreased our confidence that a global optimum can bereached. This approach also lends itself to paralleliza-tion and we have been able to make excellent use of idleworkstations.

Future PlansOur major objectives are to define a rotor with im-

proved characteristics (lower loads, longer life, reducedweight, lower vibration, better aerodynamic perform-ance) and to automate the rotor design process to reducelabor hours and design cycle time. Some of the im-provements described below are only notional. As weget closer to implementation, our vision will becomemore focused, allowing better definition of what wewant to achieve.

First, we want to add rotor aerodynamic performanceto the objective function. To accomplish this, more de-sign variables and constraints must be added to theproblem formulation.

Next, we want to continue investigating DFO meth-ods. Which method is most robust (gives the best resultsin the least calendar time, uses less computer resourcesand fewest labor hours)? Are approximate methodsmost efficient, or are errors too large to give meaningfulresults? Will only using “main effects” allow substantialreductions in the number of design variables or willvariable sensitivity be impossible to evaluate over thewhole design space? How should the optimization con-trol parameters be set to perform the most efficientsearches? These and many other questions need to beanswered.

Another improvement is the development of a methodfor classifying the many promising designs that resultfrom a DFO optimization. By identifying similar de-signs, only the best, unique (unrelated) need to be

AIAA-98-4733


refined with the gradient method, eliminating duplicateeffort.

Data mining is another potential source of efficiency.By adding all the previously evaluated designs in a non-dimensional database, a resource can be developed forfuture DFO activity. Future rotor design requirementscan have different emphasis on performance, loads orvibration, with different constraints. This will require anew design optimization problem. The database can besearched to rapidly define favorable designs to start theDFO process. Another application is to use the databasefor building response surface approximations. As moredesigns are investigated, the database will grow and sowill the efficiencies.

The whole optimization/design process needs furtherautomation. This may use natural language to set up theoptimization, run hands off until the requested task iscomplete, automatically display local design space forselected parameters so intelligent tradeoffs can bemade, provide status data to monitor progress, and useparametric design variable ranges and constraints forinitial optimization.

We also want to improve the parallel nature of ourcodes. Currently we are doing most of our computationon a network of UNIX workstations. We need to im-prove the robustness of our controller so that if onenode “crashes” (as they inevitably do), the process cancontinue with the remaining nodes. In addition, we wantthe controller to automatically search out idle computersso we can take advantage of this resource, on a nonin-terference basis.

Automated optimization and design are critical for thefuture of manufacturing in developed nations. Marketforces are requiring us to design, build, and get to mar-ket faster. Reality is pushing us to reduce design cost bydoing more with less. It can be done!

AcknowledgmentsThe authors would like to express their gratitude to

Joel Hirsh for his helpful insights and knowledge ofoptimization techniques. His implementation of the par-allel computing has been greatly valued. We would alsolike to thank the many researchers, too numerous tomention by name, who have worked with us in our in-vestigation of derivative free optimization methods.

References1. Tarzanin, F. J. and Young, D. K., “Blade TipSweep Effect on Hub Vibratory Loads,” NASA CR177425, Sep 1987.2. Domain Software Engineering Environment(DSEE) Command Reference, Atria Software, Inc.,Natick, Mass., 1986.

3. ClearCase User’s Manuel, Atria Software, Inc.,Natick, Mass., 1995.4. Shultz, L. A., Panda, B., Tarzanin, F. J., Derham,R. C., Oh, B. K., and Dadone, L., “InterdisciplinaryAnalysis For Advanced Rotors — Approach, Capabili-ties And Status,” Presented at the AHS AeromechanicsSpecialists Conference, San Francisco, CA, Jan 1994.5. Healy, M. J., “User’s Guide for the SOL/NPSOLNonlinear Programming Library — Boeing Version,”Engineering Technology Applications Library ReportETA-LR-41, Boeing Computer Services, Jun 1987.6. Young, D. K. and Tarzanin, F. J., “Structural Op-timization And Mach Scale Test Validation of a LowVibration Rotor,” Journal of American Helicopter So-ciety, Vol. 38,(3), Jul 1993.7. Staley, J. A., Mathew, M. B., and Tarzanin, F. J.,“Wind Tunnel Modeling of High Order Rotor Vibra-tion,” Presented at the AHS 49th Annual Forum, St.Louis, Missouri, May 1993.8. Tarzanin, F. J., “An Improved Low Vibration Ro-tor,” Presented at the AHS Aeromechanics SpecialistsConference, Bridgeport, CT, Oct 1995.9. Hirsh, J. E. and Young, D. K., “Evolutionary Pro-gramming Strategies with Self-Adaptation Applied tothe Design of Rotorcraft using Parallel Processing,” 7thAnnual Conference on Evolutionary Programming, SanDiego, CA, March 25-27, 1998, Springer-Verlag.10. Booker, A. J., “DACE - Design and Analysis ofComputer Experiments,” Presented at the 7th AIAA/USAF/NASA /ISSOMO Multidisciplinary Analysis andOptimization Symposium, St. Louis, Missouri, Sep 2-3,1998.11. Booker, A. J., Dennis Jr., J. E., Frank, P. D., Sera-fini, D. B., Torczon, V., and Trosset, M. W., “ARigorous Framework for Optimization of ExpensiveFunctions by Surrogates,” Technical ReportSSGTECH-98-095, Boeing Shared Services Group,Applied Research and Technology, March 1998.12. Dennis Jr., J. E. and Serafini, D. B., “Model Man-agement,” Presented at the 7th AIAA/USAF/NASA/ISSOMO Multidisciplinary Analysis and OptimizationSymposium, St. Louis, Missouri, Sep 2-3, 1998.13. Walsh, J. L., Young, K. C., Tarzanin, F. J., Hirsh,J. E., and Young, D. K., “Optimization Issues withComplex Rotorcraft Comprehensive Analysis,” Pre-sented at the 7th AIAA/USAF/NASA/ISSOMOMultidisciplinary Analysis and Optimization Sympo-sium, St. Louis, Missouri, Sep 2-3, 1998.14. Conn, A. and Mints, K., “Derivative Free Optimi-zation,” Presented at the 7th AIAA/USAF/NASA/ISSOMO Multidisciplinary Analysis and OptimizationSymposium, St. Louis, Missouri, Sep 2-3, 1998.

1


AIAA-98-4736THE CHALLENGE AND PROMISE OF

BLENDED-WING-BODY OPTIMIZATION

Abstract

Multidisciplinary design optimization (MDO) is animportant part of the Blended-Wing-Body (BWB)aircraft design process. It is a promising technology,but faces many challenges in routine application toaircraft advanced design. This paper describes currentapproaches, recent results, and future challenges forMDO as reflected in our experience with BWB designover the past four years. Current efforts have employedthe Wing Multidisciplinary Optimization Design(WingMOD) code, targeting broad optimizations withlarge sets of design variables and constraints. Theseefforts have shown substantial payoffs stemming fromthe natural ability of MDO to handle the geometriccomplexity and the integrated design philosophy of theBWB. Challenges to MDO have been identified in thebreadth and depth of the analysis desired to captureaerodynamic, stability, and control issues for thisconfiguration. Future efforts include incorporatinghigher-fidelity codes while maintaining the breadth ofscope, possibly with methods such as response surfacesand collaborative optimization.

Introduction

The Blended-Wing-Body (BWB) is a revolutionaryconcept for commercial aircraft1-2. It requires a designapproach that departs from the conventionaldecomposition of the airplane into distinct pieces andinstead integrates wing, fuselage, engines, and tail toachieve a substantial improvement in performance.This provides an arena rich in opportunities formultidisicplinary design optimization (MDO). The highlevel of integration breaks the normal design process;instead of satisfying specific requirements with adistinct airframe part, an array of requirements must besatisfied with an integrated airframe. This changes thedesign philosophy and requires developing experiencein the new way of thinking. MDO presents a solution

* Senior Engineer/Scientist, Member AIAA† Associate Professor, Member AIAA

Copyright © 1998 by Sean Wakayama. Published bythe American Institute of Aeronautics and Astronautics,Inc. with permission.

for these new design challenges. This paper describessome of the early application of MDO in thedevelopment of Boeing’s BWB concept, focusing onthe aerodynamic and structural optimization of theblended-wing planform and highlighting theopportunities for an expanded role of MDO incontinuing design work.

Current State: WingMOD

MDO in the BWB program has been undertaken usingseveral codes. Early conceptual and cabin layoutoptimization was carried out at Stanford Universityusing both gradient-based and genetic algorithms;however, most of the current MDO work has been donewith the Boeing Company’s Wing MultidisciplinaryOptimization Design (WingMOD) code. This code wasoriginally developed for conventional wing and taildesign, but has been adapted for use on the BWB.While this paper focuses on the use of WingMOD forBWB design, the basic conclusions regarding MDO inaircraft advanced design are more generally applicable.

Basic WingMOD Analysis

As described in References 3 and 4, WingMODoptimizes aircraft wings and horizontal tails subject to awide array of practical constraints. It performs wingplanform, thickness, and twist optimization, with designvariables including overall span plus chord, sweep,thickness, and twist at several stations along the span ofthe wing. It also optimizes skin thicknesses, fueldistribution, spar locations, and control surfacedeflections. During optimization, WingMOD enforcesconstraints on range, trim, structural design, maximumlift, stability, control power, and balance.

WingMOD handles structural design and maximum liftconstraints at a higher fidelity level than the traditionalconceptual design process. It also incorporatesstability, control, and balance considerations directly inthe aircraft optimization, where the traditionalconceptual design process handles these constraintsoutside the sizing loop. By performing detailedoptimization while attending wide-ranging constraintsearly in the design process, WingMOD identifies waysto trade and maximize interdisciplinary advantages,generating well-rounded configurations that are usuallyachieved at great cost with traditional design processes.

Sean Wakayama* Ilan Kroo†The Boeing Company Stanford University

Long Beach, CA 90807 Stanford, CA 94305

2


To provide this capability, WingMOD employsanalyses that have higher fidelity than those forconceptual design, but are faster than those generallyassociated with preliminary design. The basicWingMOD method models an aircraft wing and tailwith a simple vortex-lattice code and monocoque beamanalysis, coupled to give static aeroelastic loads. Themodel is trimmed at several flight conditions to obtainload and induced drag data. Profile and compressibilitydrag are evaluated at stations across the span of thewing using empirical data with lift coefficientsevaluated from the vortex lattice code. Structuralweight is calculated from the maximum elastic loadsencountered through a range of flight conditions,including maneuver, vertical gust, and lateral gust. Thestructure is sized based on bending strength andbuckling stability considerations. Maximum lift isevaluated using a critical section method that declaresthe wing to be at its maximum useable lift when anysection reaches its maximum lift coefficient, which iscalculated from empirical data. For trim, section zero-lift pitching moment is modified for trailing-edgedeflections using empirical relations.

WingMOD fits within an advanced design process assketched in Figure 1. The process begins withconfiguration and cycles through the disciplines, endingwith a sized baseline after performance analysis. Fromthe baseline configuration, WingMOD generates anoptimized design. The airplane is analyzed in moredetail than in the process developing the baseline. Thisincludes explicit modeling of control surface deflectionsfor trim and explicit calculation of span loading forweight and drag assessment. The optimized design canbe cycled through an optional computational fluiddynamics (CFD) analysis to verify the aerodynamicpredictions in WingMOD and to generate a true outermold line. For faster cycle time with lower fidelity, theCFD analysis could be skipped. Either way, theoptimized design is passed from configuration throughperformance analysis to validate the weight andperformance estimates.

Genie Optimization Framework

Optimization services for WingMOD are provided bythe Genie framework. Genie, a GENeric Interface forEngineering, was originally developed at StanfordUniversity as a shell for performing generic engineeringoptimization problems. The idea behind itsdevelopment was to build a single interface that waspowerful enough to be used for most engineeringproblems yet simple enough to be linked with anyanalysis code.

The version of Genie used in WingMOD was modifiedat Boeing under NASA contracts to handle therequirements of several aircraft design optimizationtasks. Efforts were made to develop features, which theoriginal software lacked, that were needed on variousoptimization projects. Since most problems for Genieat Boeing could be cast as a single, integrated analysis,little was done to make it an integration tool withdistributed computing capability; however, there are noobstacles to developing that capability. In its presentform, Genie enables easy linkage between the analysisand optimizer, allows automated data calculation,provides data output in useful formats, providesinformation to facilitate scaling design variables andconstraints, provides a selection of optimizers, allowsflexible definition of optimization problems, and allowsfor the development of graphical user interfaces.

Enabling easy integration of new analyses wasimportant in getting Genie to be used on more than oneproject. Linking an analysis to Genie involves writing atrivial analysis interface and communicating design datathrough simple data interface commands. The analysisinterface takes commands from the command interfaceor the optimizer and simply calls the analysis with noarguments. The data interface provides simplefunctions that the analysis uses to get and putinformation from and to the database. Since these aresoftware subroutine calls, programming is required tolink an analysis to Genie. This may seem less attractivethan communicating through files; however, theprogramming is very simple and pays for itself in fasterdata transfer between analysis and framework. For anall new analysis, data interface calls can replacetraditional input-output, saving programming time.

Genie had automated optimization and calculationcapabilities early in its development. Optimizationscould be set up and run as background jobs on Unixplatforms using a simple command language. To allowbetter visualization of the design space, the commandlanguage was expanded to allow calculations oroptimizations at points in multiple dimensions to mapobjectives and constraints through the design space.While of limited use for wing design, this feature is veryimportant for airplane sizing applications.

A complementary development was the capability tooutput results in formats for special graphic programsthat generate multi-dimensional sizing thumbprints.More important for wing design problems, outputcapabilities were added to generate data summaries thatcould be rapidly inserted in spreadsheet programs tocreate detailed graphical reports that illuminate dozensof characteristics across the wing span.

3


For detailed wing design problems, design variable andconstraint scaling is extremely important for achievingtimely, converged optimizations. Poor scaling can slowdown or prevent convergence. The difficulty inselecting proper scaling comes from having a mix ofvery different variables and constraints that relate toeach other in often non-intuitive ways. Very little issaid about how to determine proper scales for all thevariables in an optimization problem, and too oftenproper scaling is the result of a lot of experience by trialand error. There is a systematic approach to design

variable scaling4, which Genie facilitates through theNon-Linear Optimizer (NLOpt). NLOpt is based onsequential-quadratic programming and was written foruse with Genie. At the end of each optimization,NLOpt provides information that can be used toimprove scaling for subsequent optimizations. Thisfeature has been essential to enabling optimizations inover one hundred design variables.

A motivation of using an optimization framework is theopportunity to make several optimizers available to the

Propulsion

Configuration Drawing

Engine Deck

Lift and Drag Polars,Buffet Boundaries,Baseline Weights,Thrust and Fuel Burn Tables

Sized Weights, Engine Thrust,Fuel Burn, Block Time

Configuration

Aerodynamics

Weights

Performance

Weights

Economics

Weights Iteration

WingMODBaselineDesign

OptimizedDesign

GeometryCabin Shape

DragBuffet

Maximum LiftStructures

AeroelasticityLoadsWeightBalance

TrimControl Power

Stability

0

0.05

0.1

0.15

0.2

0.25

0 50 100 150 200

Spanwise Coordinate (ft)

true

begin cruisefwd landing stall

0

0.05

0.1

0.15

0.2

0.25

0 50 100 150 200


true

begin cruisefwd landing stall

-10.00

-8.00

-6.00

-4.00

-2.00

0.00

2.00

4.000 50 100 150 200


begin cruiseend cruiseferry cruise

-10.00

-8.00

-6.00

-4.00

-2.00

0.00

2.00

4.00

0 50 100 150 200



0.000.05

0.10

0.150.20

0.25

0.30

0.350.40

0.45

0 50 100 150 200



0.000.05

0.10

0.150.20

0.25

0.30

0.350.40

0.45

0 50 100 150 200



Sized Weights,Maximum Lift Data,Compressibility Drag Data

Outer Mold Line

CFD

Figure 1. WingMOD design process

4


analysis. Genie provides access to the efficient NPSOLoptimizer as well as the robust NLOpt. The switchingbetween optimizers occurs within Genie where theanalysis programmer does not need to worry about it.The optimizer is connected to a goal function interface,which acts like an ordinary function with arguments tothe optimizer; however, the goal function interfaceworks with the command and data interfaces totranscribe the abstract optimizer variables into physicalvariables. This way, the command interface can set anydatabase variable to be a design variable, objective, orconstraint, providing great flexibility in setting upoptimization problems. The complex programming toprovide this capability is invested in the frameworkwhile connections between analysis and framework arekept simple. This offers a large payoff for the low costof linking an analysis with the framework.

Graphical user interfaces (GUI’s) provide a similarmotivation for using optimization frameworks. Whileoptimizations are run as Unix command linebackground jobs, Genie does have Macintosh and X-Window GUI’s, which overlay the command interface.Investing in a GUI for a framework like Genie isattractive because that benefits every analysis connectedto the framework. The challenge is then to create ageneric GUI that can perform as well as application-specific GUI’s for a range of analyses.

The combination of optimization framework and winganalysis make the WingMOD code. As described tohere, WingMOD had been applied to design of acomposite wing for a stretched MD-905 and for studieson the MD-XX. Application to the BWB would requiresubstantial changes.

Challenges of the BWB

Radically different from conventional aircraftconfigurations, the BWB presents special designchallenges. The integrated nature of the configurationis one challenge for which MDO offers a promisingsolution. Where the design of conventional aircraft canbe divided between different disciplines, no discipline

can work independently on the BWB. Whereconfiguration can set the fuselage and aerodynamics canset the wing on a conventional aircraft, the twodisciplines are forced to work together in defining alow-drag wing that adequately encloses the payload onthe BWB. In that task, the large number of geometricdegrees of freedom coupled with a number of geometricand aerodynamic considerations present a substantialMDO problem. Adding consideration of weight,balance, stability, and control issues turns this into anMDO challenge.

Further increasing the challenge, the BWB has uniquedesign features that require higher fidelity modelingthan might be acceptable for conventional designs. Toenclose the payload within the wing, the BWB has verythick airfoil sections over its body. Attaining low drag,transonically, with these airfoils is an aerodynamicchallenge. In this region, the wing structure doubles aspressure vessel for the cabin, presenting flat panels thatmust support pressure loads over large spans dictated bythe cabin arrangement. Designing and analyzing thesepanels and assessing a weight for them is a substantialchallenge for structures and weights disciplines. Toreduce drag, the design is tail-less, but this createsinteresting challenges for stability and control: first, tobalance the airplane and provide sufficient controlpower, and second, to ensure that control deflections fortrim do not adversely affect the spanload and hence thedrag. A final challenge lies in the aft-mounted enginesand the difficulties with propulsion and airframeintegration. Before undertaking a credible MDO efforton the BWB, some of these issues had to be addressedwith new analysis methods.

Aerodynamic Method Improvements

In aerodynamics, access to rapid Navier Stokessolutions has provided tremendous insight andconfidence in the aerodynamic understanding of theBWB. The turn-around time for these solutions hasbeen adequate for wing design in the cruise condition,allowing substantial progress in the aerodynamic designof the BWB. Unfortunately, these methods are notdirectly used by WingMOD. To touch on disciplinessuch as loads, low-speed aerodynamics, stability andcontrol, WingMOD evaluates 20 flight conditions ineach analysis. To explore a broad range of designchanges, optimizations include over 100 designvariables. With 20 flight conditions per analysis, 100analyses per gradient calculation, and a minimal 100major iterations of the optimization, we end up with200,000 aerodynamic calculations per optimization.This strongly discourages any attempt to include a high-fidelity aerodynamic analysis directly withinWingMOD.

file data

command path

data path

GenieBatch File

Project File

Plot File

Command Interface

Analysis

Database

Optimizer

Data Interface Get(), Put()

Analysis Interface

Goal Function Interface

Figure 2. Genie optimization framework.

5


The difficulty that severely delayed credible applicationof WingMOD on the BWB was the finding that theoriginal WingMOD aerodynamics model was missingimportant characteristics that were captured in NavierStokes codes. Because the flow around the center-bodyis three-dimensional, the center-body pressurescorrespond to the flow over a thinner effective 2-Dsection. 3-D relief is felt because the neighboringairfoils around the center-body are not as thick. Thisallows the thick sections that are needed to enclose thepayload. In return, the outboard wing feels increasedvelocities because of the thick center-body and thepressures on its airfoils correspond to effectively thickersections. These effects were modeled as described inReference 6. An example of this effect is shown inFigure 3. Without this model, WingMOD could notproduce aerodynamically feasible designs.

Figure 3. Baseline effective t/c distributions.

This example highlights a few of the obstacles to use ofMDO in industry. First there is the reluctance to back-off on fidelity. Second is the breadth of criteria thatshould be considered in developing an optimal design.Coupled with the first obstacle, this either leads toprohibitively long optimization times or a substantialreduction in scope of the optimization problem. Thirdis a lack of intermediate-fidelity codes that canadequately substitute for high-fidelity codes at afraction of the computing cost. The WingMODapproach tackles this third obstacle but continues tomeet resistance on the issue of fidelity.

Structural Method Improvements

In structures and weights, a new method was employedto model the BWB center-body. The center-body isessentially wing structure, but it is pressurized and hasvery large rib spacing to accommodate the cabin.Structural equations were introduced in WingMOD toanalyze wing skin panels as beam-columns with appliedlateral pressure loads. This differs from the basicWingMOD buckling analysis, which looks only atbuckling stability. Lateral pressure loads andcompressive column loads from global bendingmoments are applied to the skin panels, generatingnonlinear loads. Skin panels are modeled as sandwich

structure with composite face sheets. While the coredepth is set externally by manufacturability or damagetolerance constraints, the face sheet thicknesses aresized directly in the optimization to meet stressallowables. Panel stresses are evaluated at designrunning loads that are set in the optimization and areconstrained to exceed actual running loads calculatedthrough a wide array of structural design conditions.

Stability and Control Improvements

In the area of stability and control, the BWB forced theinclusion of new concerns in the WingMODoptimization, including scheduling control surfacedeflections and observing center-of-gravity issues.Scheduling control surface deflections is importantbecause the airplane is trimmed with control surfacesdistributed along the wing, which will impact thespanload and have first-order impacts on drag andweight. Center-of-gravity (CG) and balance issues areimportant because they indirectly affect the spanload bydefining the trim points for the airplane.

To enable optimization of control surface deflectionswhile emulating a realizable control law structure,WingMOD was modified to accept five deflectionschedules: high-speed trim, high-speed control, low-speed trim, low-speed control, and maneuver loadalleviation. These gear the control surfaces of allelements in the WingMOD model to pilot trim control,pilot maneuver control, and load factor. Duringoptimization, control settings are set to trim the airplaneand control surface gearing is selected to optimizeperformance. The high-speed trim gearing targetsminimum trimmed cruise drag. The high-speed controland maneuver load alleviation gearings seek reducedcritical loads. The low-speed gearings provide controlauthority over a range of conditions while preventingcontrol surfaces from saturating or wing sections fromstalling.

To assess center-of-gravity issues, WingMOD wasmodified to track the longitudinal position of structure,fuel, payload, and general discrete masses. The array ofconditions analyzed in WingMOD includes conditionsthat set both forward and aft CG limits. Duringplanform optimization, the limits are matched to theactual longitudinal balance. The range for theperformance cruise mission is based on trimmed dragevaluated at the calculated CG. This encouragesplanforms that minimize CG range.

Propulsion Airframe Integration

Propulsion-airframe integration is an intimidatingchallenge for the BWB. This has been attacked throughCFD analysis and inverse design, with initial resultsshowing promise for solving the design problem albeitthrough a lengthy process. With the fine detail required

6


for this work, there is little hope of incorporating thisdirectly in a WingMOD optimization, although newapproaches to course-grained distributed design arebeing investigated to accomplish this kind of integrateddesign capability2.

Designing with WingMOD

With this brief description of the fundamentalmethodology of WingMOD and the design challengesof the BWB, we next summarize an example of thework that has been accomplished with WingMOD onthe BWB. This example will hint at the detail andcomplexity that is needed to address an industrialaircraft design problem with MDO. This exampleshows the substantial gains that might be achieved onnovel concepts, such as the BWB, where tight designintegration and lack of design experience make theapplication of MDO not just a nicety, but a necessity.

Critics may argue that the problem addressed in thisexample is not broad enough or that the analysismethods are not deep enough to satisfy the concerns ofindustry. More is definitely desired in both breadth anddepth, and much work remains to be done to achievethese improvements; however, WingMODoptimizations are providing answers that are useful toindustry now. While the BWB program has yet to studyan MDO-based design in detail, the directions taken byWingMOD in seeking optimal designs have provokedthought, discussions, and conventional studies that haveled to improved designs. MDO has gained acceptancein the BWB program as a tool to find ways to improvethe design.

This example uses a notional BWB developed underTask 18 of the Advanced Subsonic Technology (AST)program. The baseline airplane was configured andsized conventionally. The airplane mission was to carry855 passengers 7,500 nmi at Mach 0.85, although lessambitious BWB configurations are currently understudy. Further details of the optimization are given inReference 6.

Design Conditions

To touch on most of the critical issues affecting theBWB, 20 design conditions were examined, asdescribed in Reference 6. The BWB is highly sensitiveto CG location because that governs the deflection ofthe control surfaces, the spanload, and ultimately thedrag and weight. Where we can usually identify acritical CG location for each condition on aconventional airplane, the influence of control surfacedeflections on the spanload makes this difficult orimpossible on the BWB; hence, several conditions areexamined at both CG locations. This is one way theBWB stretches the breadth of any MDO effort. Evenwith this breadth, more conditions are desirable, with

the first additions likely to be used for analyzing yawcontrol constraints.

Design Variables

Design variables are listed in Table 1. The details arediscussed in Reference 6. The design variables coverboth external geometry and interior arrangement of themajor structural components. The boundaries of thecabin can be optimized as well as the distribution offuel. Schedules for deflection of control surfaces andstructural sizing can be handled. Optimizing thesequantities results in a 134 variable problem.

This number of variables is admittedly small relative tosome optimization problems (e.g. detailed structuralsizing and trajectories through collocation); however,the extent of the geometric degrees-of-freedom makethis an ambitious MDO problem. One obstacle to theuse of high-fidelity codes in MDO has been the abilityto automatically handle major geometry changes. FEMmodels and, to a lesser extent, CFD models would offerresistance to the planform changes examined in thisexample. The simpler models in WingMOD allow verybroad variations in geometry to be explored. This isimportant for the BWB because there is too littleexperience with the design to substantially narrow thedesign space.

To those unfamiliar with the use of formal optimizationin aircraft conceptual design, 134 design variables isquite a lot. It is more than a human would be able tosort out using conventional trade studies and exceedsthe capability of most current advanced design codes.

Name Numbermission takeoff weight 2chord 9sweep 7t/c 8incidence 7payload chordwise extent 10spar location 7fuel distribution 6nose tank fuel 3CG limits 2CG location 3trim deflection schedule 8control deflection schedule 8trim angle of attack 16trim or control deflection 16trim load factors 2design running load 13center-body skin thickness 7total 134

Table 1. Design Variables

7


This is particularly important for the BWB becauseexisting tools that size thrust and wing area do notproperly handle geometric changes to the BWB oraccount for important BWB constraints.

Constraints

Constraints are listed in Table 2. The constraints coverperformance, stability, control, balance, structuraldesign, buffet, and maximum lift. They also includegeometric constraints that force the wing to wraparound a fixed payload. The details are left toReference 6, but this table should indicate the breadthof constraints that are necessary to undertake anindustrial MDO problem. There are a large number ofconstraints, 705, but only 90 are active. The constraint-based sequential quadratic programming algorithm usedin WingMOD handles large numbers of constraints veryeasily, so the approach taken is to include all theconstraints that could possibly drive the design and tolet the optimizer determine the ones that do. When theactive constraints are compared against the 134 designvariables, there are 44 unconstrained degrees offreedom. This is a large dimension to explore thatwould take a prohibitively long time to navigate withconventional advanced-design methods.

Optimization Results

When the optimization was carried out, the designmoved from the baseline configuration sketched inFigure 4 to the optimized configuration sketched inFigure 5. Additional human design input may be usedto simplify the design from the optimizer, smoothingfeatures that add much design complexity for smallperformance gains. This leads to a final design such asthat shown in Figure 1. Alternately, the design could bere-optimized with fewer design variables after initialoptimizations reveal the most important planformbreaks.

The most substantial design changes were tighterpackaging of the payload and the thinning of airfoils inthe kink of the wing. By changing planform andthickness, the optimized design achieved a much tighterfit of the payload between the spars. The payloadextent is indicated by the shading in the figures. Thisreduced the area of pressurized skin for a substantialreduction in weight. Thinning of the kink airfoilsections relieved compressibility drag penalties andallowed the optimized design to load the kink region fora better spanload and lower drag. This is described inmore detail in Reference 6.

The final performance results are shown in Table 3.

Name Number NumberCritical

range 2 2L/D 1 0static margin 6 1payload weight 1 1payload height 10 10payload chordwise extent 20 4spar location 12 0minimum chord 3 3fuel volume 2 0fuel distribution 6 0nose tank fuel 3 1CG location 3 0CG limits 5 4control surface deflection 25 1trim load factor 17 17trim pitching moment 16 16center-body stress 8 7running load bounds 13 0running load 322 15maximum lift 184 2buffet 23 5buffet character 23 1total 705 90

Table 2. Constraints

Figure 4. Baseline Configuration.

Figure 5. Unmodified Optimized Configuration.

8


Operating empty weight is reduced by better packagingof the payload. L/D is increased, largely because ofbetter span loadings. The baseline airplane has aWingMOD-optimized span load that balances weight,drag, and control considerations. Had the baseline spanload been aerodynamics-optimized, the optimizeddesign would show little improvement or evendegradation in L/D, but it would show more substantialempty weight reduction. In optimizing fromaerodynamics-defined wings, MDO almost always findsways to improve the other disciplines at a small expenseto aerodynamics. This can make it difficult for MDO-based designs to gain acceptance from aerodynamics,especially because the aerodynamic penalties can becaptured with high-fidelity early in the design processwhile the projected gains in other disciplines can takemonths to substantiate.

performance figure % change from baselinetakeoff weight -6.9operating empty weight -5.0fuel burn -12.0gross area +0.8average L/D +7.5

Table 3. Optimization Results

The combination of weight and drag reduction results insubstantial reductions in fuel burn and takeoff weight.Wing area, which is a primary design variable forconventional sizing methods, is virtually unchanged,meaning that improvements were made through muchfiner manipulation of the geometry. This shows afundamental advantage of multidisciplinaryoptimization over conventional sizing processes. Inaddition, the design was accomplished in a short time,with overnight optimization runs and a few tries toperfect the optimization problem. This contrasts withthe months of study that would be required to optimizethe design conventionally. The design improvementsand speed that MDO offers show great promise foradvancing BWB design.

The Promise of MDO

The basic conclusion of this exercise is that the designcapabilities of an MDO process can lead to substantialimprovement in the design of a novel configurationsuch as the BWB. There are some less-visibleadvantages that come from MDO codes that aredescribed below.

Design Cycle Time

In design studies using conventional methods, thefollowing observations could be made. Theconventional advanced design process uses 3 to 6 weeksfor a BWB planform change to cycle through

configuration, weights, aerodynamics, and performanceanalysis. Optimizing an aircraft could take severalcycles (months) to optimize the aircraft. The cycle timelimits the number of design variations that can beexplored. Even worse, this cycle time only coversperformance analysis; additional time is required forbalance, aeroelastics, stability and control.

In the course of this study, the advantages of theWingMOD approach could be seen. It still takes 3 to 6weeks to model and calibrate a baseline design inWingMOD. This is comparable to the cycle time for aplanform analysis using conventional methods;however, only a single run is needed to optimize theaircraft, reducing months of cycle time to an overnightjob. In addition, the optimization handles many moredesign variations than could be explored by theconventional methods. Finally, the optimization dealswith balance, aeroelastics, stability and control issuesthat the conventional approach leave for later analysisand revision.

First-Cut Information

To perform multidisciplinary optimization, couplingaerodynamic loads and structural design is almost amust. From there, it is natural to make that anaeroelastic calculation. Doing this for MDO adds theadvantage that the resulting system is highly-automated,fast, and robust. An unexpected benefit is that an MDOcode, like WingMOD, with grandiose expectations ofplanform optimization becomes amazingly useful formundane tasks such as providing a first estimate ofloads, an initial sizing for skin thicknesses, andaeroelastic stability data. While there are industrialprocesses in place to do all this, they are expensive,time-consuming, and have a chicken-or-the-eggproblem: how do you generate loads when you needskin thicknesses to capture the aeroelastic effects, butyou need loads to figure out what the skin thicknessesneed to be? The fidelity of those processes justifiestheir expense, and we would never use WingMOD tocertify an airplane, but WingMOD is perfect for gettingthe first cut at the loads and structural sizing from whichthe detailed processes can start.

Individual Versus Total Good

Pushing for overall airplane improvement overindividual discipline improvement can be a difficultpractice to incorporate in a large design team, but it isespecially important for a revolutionary concept such asthe BWB. Traditionally, aerodynamics has taken thelead in defining wing shape. A compelling reason forthis is the speed of aerodynamic processes: a wing-onlyplanform change can be put through CFD in as little asa few days. The other disciplines are not so lucky: afinite element model can take six months. So while

9


aerodynamics can push for a particular design with hardfacts, the other disciplines can offer only qualitativeobjections, or the design cycle must drag on for hardnumbers from the other disciplines. For conventionalaircraft, this is not very critical: the wing turns outheavier and the tail turns out bigger than they ought tobe, but the airplane will still work. For the BWB, thiscould be disastrous: the wing shape that maximizes L/Dis unlikely to lead to a balanced airplane with thecontrol authority to rotate for takeoff. The WingMODapproach looks at all the design drivers it can to offer adesign that is the best compromise between thedisciplines. Analyses provide hard numbers, albeitapproximate, for each discipline, making it difficult forany one discipline to dominate. It is difficult to acceptthat WingMOD designs inevitably come in with lowerL/D than aerodynamics group knows they can achieve,while offering benefits in other disciplines that cannotbe immediately verified. Even within aerodynamics,WingMOD will compromise cruise performance toenable meeting low-speed lift and control requirements.

Future Directions

While WingMOD optimization has made promisingfirst steps toward solving the BWB design problemthrough MDO, much more is desired. Problem areasspecific to the BWB are identified below.

Increased Breadth

While the breadth of conditions examined byWingMOD is relatively well accepted, there areinstances where more is desired. An example ismodeling engine-out lateral control. From experiencewith the BWB-17 Flight Control Testbed2, this coulddrive the sizing of the outboard wing and wingletchords, which affect the effectiveness of the ruddersneeded to control this condition.

Higher-Fidelity Codes

The WingMOD aerodynamics module certainly leavessomething to be desired for analyzing the BWB;however, the speed of this analysis is required to coverthe breadth of flight conditions that are essential toperforming any multidisciplinary planformoptimization. Because of their speed, higher-fidelitypanel methods are the most likely next-step toimproving the WingMOD aerodynamic analysis.Incorporating a true CFD analysis promises the benefitof capturing all the important aerodynamic effects andthe ability to directly handle the propulsion-airframeintegration problem; however, direct inclusion of CFDat this time would likely bring a WingMOD-breadthoptimization to a screeching halt.

The inclusion of finite element methods (FEM) is alower priority than CFD. This is because the span time

for generating adequately detailed FEM models is toolong for them to be used actively in the conventionaldesign process. Design work on the BWB uses weightestimates from parametric equations that may becalibrated to but are really independent from FEMresults. The intermediate-fidelity structural analysis inWingMOD is already better than parametric weightequations, so the optimization cannot be faulted withmissing something the standard approach would catch.At this stage, FEM work is very important forcalibrating weights codes and verifying that there are noshow-stoppers in the design, but it works too slowly tosubstantially impact planform trade studies. If FEManalysis had a span time equivalent to that for CFDanalysis, then it would play a stronger role in the earlydefinition of an aircraft, and there would be a greaterimpetus to include it in advanced-design MDO.

A tantalizing prospect for increasing the fidelity of aWingMOD-type optimization is incorporation of adetailed mission analysis code. This could bring highquality to the performance figures at littlecomputational cost. It could also eliminate many stand-in constraints, for example takeoff speed targets insteadof a true field length constraint. The issues here includejudiciously selecting a minimum number ofaerodynamic analyses to provide the data required forthe mission analysis and generating noise-free numbersfrom the mission analysis.

Propulsion-airframe integration is especially importantfor the BWB because of the potential for eitherimproved performance or problems with high distortionassociated with boundary layer ingestion. Because ofthe complex, viscous, transonic flow in this region,simple models are ineffective and one is forced to relyon rather time-consuming CFD simulations for reliableguidance. The simple framework on which WingMODis based is not well-suited to the incorporation of suchmethods and future work is clearly required in this area.

Optimization Framework Improvements

While the example optimization presented in this paperis large, many other parameters must be input to runWingMOD and this presents an often bewilderinglysteep learning curve. Improvements in the way theframework handles large numbers of variables wouldhelp divide the problem to be more tractable to the user.

Applying techniques for decomposition through theoptimization framework would be ideal. That wouldprovide additional capability while allowing sub-problem analyses to remain unchanged and unburdenedby the complexity of the overall optimization problem.A candidate project likely to help BWB optimizationstudies would build collaborative optimizationcapability into the Genie framework.

10


The Challenge of MDO

The promise of MDO has been suggested by our recentexperience with BWB design; however, it has alsohighlighted some of the generic problems andchallenges in industrial acceptance of MDO.

Problem Formulation

As with single discipline optimization, correct andefficient problem formulation is critical to obtaininguseful results from MDO. Because of the subtleinteractions and interdisciplinary feedback that may beless well known to disciplinary experts, it is often moredifficult to anticipate the weaknesses of analyses or theill-posedness of a particular problem. Experience withboth conventional and BWB design suggests that theproblem formulation (selection of design variables,objectives, constraints, and bounds) evolves as thedesign is developed. It is naïve to expect that a realisticlarge-scale MDO problem can be fully-understood by adesign team from the outset. While automaticaerodynamic optimization with a specified planformand a restricted set of design conditions can bereasonably well formulated a priori, themultidisciplinary aircraft design problem is morechallenging and calls for a qualitatively differentapproach. One must structure the problem in such away that changes in design variables and constraints canbe made along the way. Individuals and truly integratedteams must routinely meet to evaluate the results andrefine the analysis requirements or problem definition.The potential for impractical designs that reduce thecredibility of the process is great without such plannedintervention.

Breadth Versus Depth

Multidisciplinary optimization results are oftencriticized for being so limited in scope or fidelity as tobe merely academic exercises—and such criticism isoften well justified.

Based on the notion that a chain is only as strong as itsweakest link, low-fidelity models covering manydisciplines are sometimes omitted, leaving a two or(rarely) three discipline MDO problem that usessophisticated disciplinary models. A chain with missinglinks is worse than one with weak links. A classicexample is that aerodynamic and structural optimizationwithout consideration of maximum lift leads to wingswith absurdly small tip chords3. The BWB designproblem illustrates the large number of disciplines toyield reasonable results.

On the other hand, the BWB represents an example of adesign for which 2-D section analysis superimposed ona simple 3-D model fails to reveal some of thefundamental opportunities available in the BWB design

space. The use of too-simple analyses might lead one toconclude that the advantages of the concept wereinsufficient to warrant the development of improvedanalyses or further consideration.

This is one of the most fundamental dilemmas in MDOthat will not be solved by advances in optimizationtheory or AI. Practical MDO will always require goodengineering judgement to match the scope of theparticular problem to appropriate analyses.Approximate models are often very adequate,depending on the actual sensitivities of activeconstraints and objectives to the particular choices fordesign variables. Rapidly increasing computationalcapabilities including parallel systems and efficientalgorithms will change the selection of appropriatemodels, but will not reduce the importance of this step.As more sophisticated analyses become feasible, theimportance and difficulty of problem formulation andintegration will only increase.

Optimization Analysis Requirements

The breadth versus depth problem would be alleviatedif high-fidelity analyses ran faster. Intriguing optionsfor increasing CFD optimization speed are automaticdifferentiation and adjoint formulations, the latterpromising sensitivity information for little more than afunction evaluation, although the present probleminvolves a large number of constraints that reduce theattractiveness of an adjoint approach. The bestimprovements for FEM lie in automating the modelgeneration process.

Beyond speed, analyses must be robust and smooth tobe used with optimization. The robustness of automaticgrid generation through large planform variations is aproblem. The smoothness of CFD and FEM results isalso an issue.

Speed, robustness, and smoothness are also an issuewith mission analysis codes. While several missionanalysis codes exist that admirably fill the requirementsof engineering analysis, the requirements foroptimization motivate the creation of new codes that arebuilt for optimization from the ground up. Such codescould use techniques, such as collocation, that makesense for optimization but were not important for theengineering needs the existing codes were written for.

Integration

In the development of WingMOD little attention wasgiven to allowing for integration of existing codes oroptimization decomposition techniques: the lack orunavailability of fast, intermediate-fidelity codes madeit more expedient to develop an all-new, tightly-coupledanalysis, which would not benefit from decomposition.

11


As more complex aerodynamic, structural, and dynamicanalyses are included in BWB optimization, the basictightly-integrated framework on which WingMOD isbased begins to become unwieldy. Several researchprograms are currently underway to address suchproblems, although applications as complex as theBWB planform design problem have not beensatisfactorily demonstrated to date. This wouldconstitute an excellent test for the industrial applicationsof various concepts for decomposed analysis anddistributed design, Reference 7.

The best near-term possibilities for bringing CFD intoBWB MDO may be in the use of response surfaces andcollaborative optimization. Collaborative optimizationwould isolate the CFD analysis in its own sub-space.Response surface techniques could map and capture theCFD analysis sub-space, which could then be includedin a collaborative optimization formulation withWingMOD capturing the non-aerodynamic disciplines.Alternately, response surfaces could simply captureaerodynamic data from specific CFD runs to be laidover WingMOD aerodynamic results. A tight couplingof aerodynamics and structures for aeroelastic loadscalculation combined with a loosely-coupled, higher-fidelity aerodynamic performance code may solve someof the problems that involve both high dimensionalitycoupling and the need for very accurate aerodynamicsolutions.

Although various techniques for loosely couplingmultidisciplinary design problems have been proposed,(e.g. concurrent subspace optimization andcollaborative optimization8-9), few have seen applicationin industry projects. We attribute this primarily to thefact that these techniques are still the subject of activeresearch and have not matured to the point that they areeasily implemented as an option in a commercialsoftware package. The availability of such technologymay reduce the need for an individual who understandsboth the particular design problem and the details of theoptimization framework and theory. Although progressin this area continues, Reference 10, we do not expectthat such a system is imminent.

Validation

There are very few examples of MDO-derived designsbeing validated to the point of being real, useableconfigurations. Advanced-design level optimizationneeds to be validated with high-fidelity analysis; high-fidelity optimization needs to be validated throughbroad analysis checks. Reference 5 describes the use ofWingMOD to develop a conventional aircraft wingconfiguration and the subsequent CFD validation.Achieving acceptance for MDO in industry will requiremore examples of validated optimized designs. No

validation of an optimized BWB design has been done,but WingMOD designs are close to being assessed withCFD. Passing the challenge of validation will be mostimportant to bringing MDO to the forefront of BWBdesign.

Conclusions

The BWB is a revolutionary concept that benefits fromMDO and yet illustrates the many challenges to its usein industry. Current efforts with the WingMOD codehave been stretched in depth, particularly to captureunusual aerodynamic characteristics, and in breadth, tocapture stability and control issues. Introducing high-fidelity analysis would be highly desirable, aprerequisite for handling propulsion-airframeintegration, yet the breadth of the BWB design problemalmost prohibits the direct substitution of moresophisticated codes for the current simpler models.

Much progress has been made with the advanced-designlevel WingMOD code. Successful optimization hasbeen made with a large, comprehensive set of designvariables and constraints. Attacking this broad problemhas offered substantial payoffs because of the youth ofthe BWB concept: current BWB configurations are notas finely evolved as conventional transports. Thesuccess in handling this broad design problem has partlybeen facilitated through capabilities provided by theGenie optimization framework.

MDO offers much promise for improving the BWB.Optimization studies have shown potential forsubstantial reductions in takeoff weight. This comesfrom the ability of MDO to handle many more degreesof freedom and track more interactions acrossdisciplines than conventional advanced-designprocesses. The BWB can benefit greatly from MDObecause of the complexity of its geometry and theintegrated nature of its design. In addition, the innateautomation required for optimization offers significantreductions in design cycle time while handlingconsiderations beyond the scope of the existingprocesses, including control surface deflections,balance, control, and aeroelastic effectiveness.

Achieving the promise will involve more work.Increased breadth of analysis and optimizationframework improvements will evolve naturally,although it would be desirable to accelerate thosedevelopments. Incorporating higher-fidelity codeswhile maintaining the breadth of scope will be a largechallenge, offering opportunities to test methods such asresponse surfaces and collaborative optimization.While current BWB work demonstrates the potential forMDO in aircraft advanced design, it remains to verifythe predicted advantages of these optimized designsusing more refined analysis codes.

12


References

[1] Liebeck, R. H., Page, M. A., Rawdon, B. K.,“Blended-Wing-Body Subsonic CommercialTransport,” AIAA Paper 98-0438, Jan. 1998.

[2] “Blended-Wing-Body Technology Study,” FinalReport, NASA Contract NAS1-20275, BoeingReport CRAD-9405-TR-3780, Oct. 1997.

[3] Wakayama, S., Kroo, I., “Subsonic Wing PlanformDesign Using Multidisciplinary Optimization,”Journal of Aircraft, Vol. 32, No. 4, Jul.-Aug. 1995,pp.746-753.

[4] Wakayama, S., Lifting Surface Design UsingMultidisciplinary Optimization, Ph.D. Thesis,Stanford University, Dec. 1994.

[5] Wakayama, S., Page, M., Liebeck, R.,“Multidisciplinary Optimization on an AdvancedComposite Wing,” AIAA Paper 96-4003, Sep.1996.

[6] Wakayama, S., “Multidisciplinary DesignOptimization of the Blended-Wing-Body,” AIAAPaper 98-4938, Sep. 1998.

[7] Kroo, I., “Multidisciplinary OptimizationApplications in Aircraft Preliminary Design—Status and Directions,” AIAA Paper 97-1408, Apr.1997.

[8] Kroo, I., “Decomposition and CollaborativeOptimization for Large-Scale Aerospace DesignPrograms,” in Multidisciplinary DesignOptimization: State of the Art, N. Alexandrov andM. Y. Hussaini, editors, SIAM, 1996.

[9] Braun, R. D., and Kroo, I. M., “Development andApplication of the Collaborative OptimizationArchitecture in a Multidisciplinary DesignEnvironment”, in Multidisciplinary DesignOptimization: State of the Art, N. Alexandrov andM. Y. Hussaini, editors, SIAM, 1996.

[10] Sobieski, I. P., Manning, V. M., Kroo, I. M.,“Response Surface Estimation and Refinement inCollaborative Optimization,” AIAA Paper 98-4753, Sep. 1998.

AIAA-98-4701

Copyright © 1998 by the American Instituteof Aeronautics and Astronautics, Inc.All rights reserved. 1


A DESCRIPTION OF THE F/A-18E/F DESIGN AND DESIGN PROCESS

James A. Young*, Ronald D. Anderson**, andRudolph N. Yurkovich+

The Boeing CompanySt. Louis, MO

* Director of Engineering, F/A-18 Program** Director, Phantom Works, Assoc. Fellow AIAA+ Fellow, Assoc. Fellow AIAA

Abstract

This paper describes the design and the design processused to develop the F/A-18E/F aircraft. It is presentedhere to document the state-of-the art of the designprocess for development of a modern high performancefighter aircraft. It is intended that this information willprovide a background for researchers developingMultidisciplinary Design Optimization (MDO)processes for aircraft design. The design process itselfwas an advance for the F/A-18E/F in that it marked thefirst application of the Integrated Product Development(IPD) design process to an Engineering ManufacturingDevelopment (EMD) program at the McDonnellDouglas Corporation. Since the F/A-18E/F's flight testprogram is well under way, results are available bywhich to judge the success of this design and the designprocess. Finally, some conclusions andrecommendations for additional work to improve thedesign process are made.

Introduction

In 1990 the MDO Technical Committee (TC) wasformed as a technical committee of the AIAA. One ofthe tasks that this committee undertook was to definethe state-of-the-art as it existed at that time and theresults of this study were published as Reference 1.Since 1990 other documents have also presented state-of-the-art approaches with Reference 2 being anexcellent example. The references have done anexcellent job of documenting theoretical developments.However, the AIAA MDO TC felt that more wasrequired to transfer the MDO message from thetheoreticians to the aircraft designer and for thetheoreticians to have a better perspective on what isrequired to design a new aircraft. It was determined thata series of papers by industry documenting the currentdesign process as used on current design programswould be an appropriate step in making this happen.This paper, which addresses the F/A-18E/F, is one ofa series of papers in response to that action item.

The F/A-18E/F, shown in Figure 1, represents the nextstep in the evolution of the F/A-18 aircraft. In addition,

its development represents a next step in the evolutionof the aircraft design process. The E/F was designedusing the Integrated Product Team (IPT) approach andthis represents a significant advance from the designprocess used in the development of the original aircraft.This paper presents a description of the aircraft designas well as a description of the design process.

GP81128001.cvs

Figure 1. F/A-18E/F Super Hornet

In MDO an objective function subject to a set ofconstraints is defined and a mathematical process isused to minimize this objective function withoutviolating the constraints. Sensitivity derivatives areusually computed as part of the optimization process.Reference 3 provides a good description of themathematical process.

If the above definition of MDO is applied in a strictsense, then MDO was not used to design the F/A-18E/F.However, the F/A-18E/F was designed using aMultidisciplinary Design Process. Based on resultsobtained from the flight test program, the aircraft is avery successful one. Thus, the current design processmust also be regarded as successful.

The F/A-18E/F was designed to meet a specific set ofrequirements rather than by optimizing a specificobjective function. From the perspective of MDO, theserequirements can be viewed as constraints whichimplies that the F/A-18E/F is a feasible design.


Description of the Aircraft

Aircraft Missions - The F/A-18E/F is a multi-missionaircraft designed for the US Navy. The concept of amulti-mission aircraft is significant for the MDOprocess in that there are multiple requirements that theaircraft must meet, and this complicates the definition ofan objective function. For a single mission aircraft, theformulation of the objective function is a simpler task.The F/A-18E/F was designed to perform both air-to-ground and air-to-air missions. These missions weredefined as requirements and the goal was to develop adesign that satisfied them. A description of the MDOprocess as it applies to a multi-mission aircraft wasinitially presented in Reference 4.

Figure 2 illustrates the multi-mission concept startingwith maritime air superiority on the left and proceedingto all weather attack on the right. These missionextremes are significant in that historically they havebeen performed by dedicated aircraft. The F-14Dperforms the air superiority mission and the A-6Fperforms the all weather attack mission. While the F/A-18C/D has some capability to perform these missions, ithas not been optimized for them. For fleet defense theF-14 with its Phoenix missile system is superior to theF/A-18C/D. However, as a multi-mission aircraft, theF/A-18C/D still has significant capability in this area.Similar arguments can be made for the ground attackmissions.

GP81128002.cvs

MaritimeAir

Superior-ity

AirCombatFighter

FighterEscort Recce

CloseAir

Support

Day/NightAttack

AllWeatherAttack

F/A-18A/B/C/D

F-14D/NATF

A-6F/A-12

AirDefenseSuppres-

sion

F/A-18E/F

Figure 2. Hornet Spans the Mission Spec

As the F-14D’s and the A-6F’s are retired form thefleet, they will be replaced by F/A-18E/F’s. Thus, theoriginal mission spectrum of the F/A-18C/D has beenexpanded even further for the F/A-18E/F as shown inFigure 2.

Each of these missions has a specific set ofrequirements that the aircraft must meet. An MDOapproach to meeting these requirements was not takenbecause MDO design techniques were not available atthe time the F/A-18E/F was designed. However, forfuture aircraft design this approach may offer significantimprovements if appropriate tools can be developed.

History of the Configuration - The F/A-18E/F is aderivative of the F/A-18C/D aircraft, which was

originally derived from the Air Force lightweight fightercompetition. Consequently, there is a great deal ofhistory behind this configuration with the general shapeof the aircraft being defined by the original YF-17.Figure 3 shows the planform view of these three aircraftand the heritage of the E/F aircraft is obvious. Table 1summarizes some of the basic geometry data. Theoriginal YF-17 had a wingspan of 35 ft and a wing areaof 350 sq. ft. For the F/A-18A the correspondingnumbers are 37.5 ft and 400 sq. ft. While the basicaerodynamic concept of the YF-17 and the F/A-18Awere essentially the same, the interior of the F/A-18Awas completely redesigned. Most of the requiredchanges were a result of transforming what was a light-weight fighter for the Air Force to a ship-board multi-role aircraft for the US Navy.

GP81128003.cvs

YF-17 F/A-18A F/A-18E

Figure 3. Comparison of Aircraft Planforms

Dimensions YF-17 F/A-18A F/A-18 E

Span(without missiles)

35.0ft 37.6ft 42.9ft

Length 56.0ft 56.0ft 60.2ft

Height 14.5ft 15.3ft 15.8ft

Tail span 22.2ft 21.6ft 23.3ft

Wheel track 06.9ft 10.2ft 10.45ft

Wing area 350sq ft 400sq ft 500sq ft

Weights

Empty 17,000lbapprox.

21,830lb 30,600lb

Fighterconfiguration

23,000lb 34,700lb 47,900lb

Maximum 51,900lb 66,000lb

Table 1. Comparison of Specifications


At the time that the F/A-18A was under goingpreliminary design, the lightweight fighter competitionwas still ongoing. This provided a constraint on theoriginal F/A-18A that resulted in the F/A-18A stillhaving essentially the same size and shape as the YF-17. For the MDO process, this is significant because itimplies a constraint that would not be present if theF/A-18A were a totally new aircraft.

Similar observations can be made for the F/A-18Erelative to the F/A-18A. The F/A-18E configurationgrew relative to the F/A-18A configuration. The span ofthe F/A-18E is 44.7 ft and the wing area is 500 sq. ft.However, the general shape of the aircraft has beenmaintained. A comparison of the F/A-18E Super Hornetto the original Hornet is shown in Figure 4. In additionto the Super Hornet being a larger aircraft with a newinlet, changes in the Leading Edge Extension (LEX)and the addition of a wing leading-edge snag areapparent.

GP81128004.cvs

Figure 4. Super Hornet Compared to OriginalHornet in Flight

Hornet 2000 Study - The evolution of this newconfiguration had its origins in the Hornet 2000 study,Reference 5, which was conducted in 1988 by a jointteam composed of the US Navy and McDonnellDouglas. Over the life of the F/A-18A/B and F/A-18C/D aircraft many changes were incorporated thatresulted in an increase in weight and the internal spacebeing used for new and additional avionics equipment.Because of this growth, reductions in range and otherperformance metrics occurred. In addition, changes tomeet the increased threat that the aircraft was to facewere required. It was anticipated that the capabilities ofthe threat would continue to increase. Quoting from theHornet 2000 study;

“Major advances in threat capability have occurredsince the F/A-18 was designed in the mid 70s. Theoriginal design goal for the Hornet was to havesuperiority over FISHBED and FLOGGER class airthreats and to penetrate battlefields with SA-2, SA-3,SA-6, and SA-7 class surface-to-air threats. That threathas changed rapidly in character and capability,primarily as a result of successful Soviet efforts intechnology transfer. The Soviets have demonstrated anability to implement rapidly technologies developeddomestically and acquired through legal and covertmeans. Through this aggressive program ofmodernization, the ability of the threat to confront theCarrier Battle Group has increased significantly.”

Since 1988, a great deal has happened to change thenature of the threat. However, while the need to dealwith the Soviet threat may have diminished, new threatshave emerged. The need to deal with these threatsformed a significant requirement for an advancedHornet.

In addition to recognizing the need for a new aircraft,the Hornet 2000 Study identified planned improvementsfor the F/A-18C/D aircraft through 1995. Theseimprovements were in three major areas: avionics,propulsion, and equipment. The avionics upgrades wereto improve the F/A-18 weapon system capabilities inthe areas of situational awareness, air superiority, air-to-surface attack and survivability. The propulsion upgradeconsisted of replacing the baseline engines with theEnhanced Performance Engine (EPE). This engineoffered significant performance improvements at higherspeeds and could be incorporated without airframechanges. The equipment growth consisted of installingan On-Board Oxygen Generating System (OBOGS)increasing the aircraft cooling capacity by an ECSupgrade, and adding a bay in the left hand LEX to allowinstallation of additional avionics.

In summary, because of the changing nature of thethreat and because the basic aircraft, even with the EPE,had just about reached the limits of its capabilities, anew aircraft was required. The Hornet 2000 Studyproduced a set of requirements and an aircraftconfiguration that addressed them. This study lookedbeyond the 1990s to determine the requirements for theaircraft such that it could continue to meet the threat.The goal of the study was to identify high valueupgrades and develop a phased incorporation plan toensure continued F/A-18 survivability andeffectiveness.

This new aircraft configuration, however, had aconstraint that required as much commonality aspossible with the original aircraft. Even with thisconstraint, early in the design process, several


alternative configurations were investigated and asample of these configurations is shown in Figure 5.While this study was specifically directed to upgradingthe F/A-18, it is believed that it is representative of thetype of trade study that would be conducted by industryand therefore should be relevant to the development ofMDO to the design process. Seven potentialconfigurations to meet the US Navy missions needs in1995 and beyond were investigated. Theseconfigurations spanned the range from minimumchanges through Block Upgrades to major conceptchanges that reflected canard-wing arrangementspopular at the time. The configurations were built fromthe same baseline and took advantage of plannedupgrades. They also shared common requirements foran updated weapon system, survivability improvementsand increased thrust.

GP81128005.cvs

• Increased Fuel• Growth II Engine• Active Array Radar• INEWS

• FY88 Baseline Plus – FY90 and FY92 Avionics Upgrades – Weapon System Upgrade – Survivability Enhancement – Enhanced Performance Engine

• Raised Dorsal

• Stiffened Wing

II

• Fuselage Plugs• Cranked Arrow Wing With Canards

I

IV

• Fuselage Plugs

• Larger Wing With Increased Chord

• Raised Dorsal

• Larger Wing With Increased Chord

• Fuselage Plugs

• Larger Wing With Increased Chord and Span

• Raised Dorsal

• Larger Wing With Increased Chord and Span

III

IIIAIIIB

IIIC

Figure 5. Configuration Options

Configuration I minimized the impact to the airframe.Weapon system updates were achieved within theexisting space/volume. Pilot situational awareness wasimproved and workload decreased by upgrading thecockpit to display integrated weapon systeminformation. Advanced air-to-air missile capability wasprovided along with the capability to carry air-to-airmissiles on the out-board pylons.

The remaining configurations incorporated changes tothe airframe. Common elements include increased fuel,

new Growth II engines, and an electronically scannedactive array radar.

Configuration II expanded mission flexibility withadditional internal fuel in a raised dorsal. Aconfiguration of this type was successfully used on theA-4M. Performance improvements were achieved withthe higher thrust engines that required enlarged inlets.External stores carriage speeds were increased with astiffened wing. Target detection range was more thandoubled by adding the active array radar. Adding newelectronic warfare equipment for passive missiledetection and laser warning enhanced survivability.

Configuration III incorporated the upgrades ofConfiguration II while replacing the stiffened wing withan enlarged wing for enhanced carrier suitability,maneuverability, and mission performance. Additionalgrowth space was also provided. Configuration IIIAenhanced the transonic/supersonic flight regime byutilizing a fuselage plug rather than the raised dorsal forincreased fuel. Configuration IIIB optimized the wingarea growth of Configuration III with an increased wingspan for improvements in mission radius and carriersuitability performance. Configuration IIIC combinedthe fuselage of Configuration IIIA and the wing ofConfiguration IIIB for enhanced transonic/supersonicflight and improved mission and carrier suitabilityperformance.

Configuration IV added fuselage plugs similar toConfigurations IIIA and IIIC. However, theaerodynamic configuration was completely new andwas targeted at potential co-development by the USNand an international customer. The wing was a crankedarrow wing and the stabilator was replaced by a canard.The vertical tails were also of a new design. Thisconfiguration shared the fuselage and all of the internalcomponents of Configurations IIIA and IIIC includingone of the major cost contributors, its avionics suite.

A detailed discussion of the features and benefits ofeach of these configurations is beyond the scope of thispaper. Each presents new operational benefits and, ingeneral, as additional benefits are added so is additionalcost. For the future, MDO could be used to determinewhich configuration best meets the new requirementsfor an improved strike fighter. At the time the study wasconducted, MDO techniques to aid in this decision didnot exist.

The new engine, which was assumed for ConfigurationsII through IV, fostered a significant multidisciplinarydesign integration activity. At the time of the Hornet2000 study, this new engine was designated the F404Growth II engine. The Growth II engine was to be anupgraded version of the F404-GE-400 engine that


would have significant performance improvementsthroughout the flight envelope. It was to provideapproximately a 25 percent increase in sea level, staticinstalled thrust. At up-and-away conditions the installedthrust increase was estimated to be up to 40 percentover the current engine. The improved performance wasto be achieved through incorporation of enginecomponents that featured advanced aerodynamics andmaterials. The engine also featured increased engineairflow and higher operating temperature capabilitieswithout a reduction in the current hot section life. Whilea growth inlet was required for optimization of theGrowth II performance, the engines fit within thecurrent F/A-18 engine bay. The engine that is installedin the F/A-18E/F has been designated as the F414-GE-400 engine and is an advanced derivative of theHornet’s current F404 engine family.

Configurations I through IV were evaluated against thefollowing set of criteria: carrier suitability, strikemission, fighter mission, maneuverability, fire controlsystem, survivability, growth potential, weapon systemeffectiveness and cost, both recurring and non-recurring. This evaluation was summarized in a stop-light format as shown in Figure 6 where G-green-indicates good, Y-yellow indicates marginal, and R-redindicated serious concern. It should be noted that if costis considered, the conclusion as to whichconfiguration is optimum is difficult to formulate.Clearly, all of the configurations represent some degreeof improvement, but at some cost. All of the newconfigurations cost more than the baseline and allrequire some investment. The cheapest solution is to donothing. On the other hand, as described earlier in thediscussion of today’s threat, to do nothing would put theaircraft in a situation where it would not be able tocompete.

The Hornet 2000 Study identified four major studypaths, with seven configurations for the HornetUpgrade. The first path, Configuration I, was attractivefrom a cost standpoint but had degraded aerodynamicperformance and little remaining growth potential. Thesecond path, Configuration II had impressive weaponsystem improvement but suffered from carriersuitability shortcomings. The third path made up ofConfigurations III, IIIA, IIIB, and IIIC, had significantperformance, carrier suitability and weapon system

G

Y

Y

R

R

Y

Y

R

1.00–

Y

Y

Y

R

G

G

R

Y

1.141.00

GP81128006.cvs

CarrierSuitability

Strike Mission

Fighter Mission

Maneuverability

Fire ControlSystem

Survivability

Growth Potential

Weapon SystemEffectiveness

Non-DimensionalCost Range REC NR

Y

G

G

R

G

G

R

Y

1.302.22

Y

G

G

R

G

G

Y

G

1.372.88

Y

G

G

R

G

G

Y

G

1.362.77

Capability

G

G

G

R

G

G

G

G

1.402.89

G

G

G

Y

G

G

G

G

1.392.78

Y

G

G

Y

G

G

G

G

1.463.44

FY88 I II III IIIA IIIB IIIC IV

Configuration

G Good Y Marginal R Serious Concern

Figure 6. Hornet “2000” ConfigurationEvaluation vs. 2000 Threat

improvements. The fourth path, Configuration IV, hadControl Configured Vehicle (CCV) potential with thecanard-cranked arrow wing arrangement.

The Hornet Upgrade Configurations IIIB and IIICoffered the greatest increase in weapon systemcapability, carrier suitability and performance. Theyincluded a larger wing, more fuel, growth engine, 10percent growth inlet, active antenna, upgraded crewstation, integrated CNI avionics and an integratedelectronic warfare system.

The final conclusion was that Configuration IIIC wasthe best path for upgrade since it was considered tohave the best carrier suitability performance.

This discussion of the process that led to what wasdetermined to be the best configuration providesvaluable insight into the design process for MDO codedevelopers. As stated elsewhere, an objective functionthat could be used to determine the optimumconfiguration would prove very difficult to formulate inthis case. In fact, typical parameters that have beensuggested as objective functions such as minimumweight or minimum cost were not the finaldiscriminators of the selected configuration. In the finalanalysis, the configuration that was selected was the onethat best satisfied the requirements within the constraintof retaining major F/A-18A configurationcharacteristics.

The F/A-18E/F Program - The F/A-18E/F program,which has its origins in the Hornet 2000 program, wasawarded to McDonnell Douglas on May 12, 1992. Thecost of this program for the development phase was$5.803 billion in 1992 dollars. This cost number can beregarded as another constraint on the design.


The F/A-18E/F rolled out on September 19, 1995 andits first flight was November 29, 1995. The aircraft as itappeared at roll-out is shown in Figure 7. Ten aircraftwere built to support the flight and ground testprograms, seven flight test articles and three ground testarticles. The flight test program began at Naval AirWarfare Center Patuxent River, Maryland on February14, 1996 and is ongoing. However, it is estimated thatthe EMD portion of the program is now 90 percentcomplete. As of January 31, 1998 1,463 flightsrepresenting 2,239.4 flight hours, had been flown.

GP81128007.cvs

Figure 7. F/A-18 Super Hornet as it Appearedat the Roll-out Ceremony

While the Hornet 2000 study defined the basic shapeand size of the E/F, the details of the design still wereto be worked out.

Basic Changes - The primary changes developedduring the study and the subsequent refinements aresummarized here:

1) The area of the wing was increased by 25% to 500square feet. This change was made to increase the rangeand payload of the aircraft.

2) A snag in the leading edge of the wing wasincorporated. This design feature was part of theoriginal F/A-18A design but was removed due toexcessive loads on the leading edge flaps. It wasreintroduced here to improve carrier landing handlingqualities.

3) The LEX was enlarged and reshaped for better highangle of attack performance. Initially the LEX wasbasically an enlargement of the LEX used on the C/Daircraft. However, during wind tunnel testing the high-angle-of-attack characteristics of the E/F aircraft withthat LEX were not as good as those of the C/D aircraft.The new LEX shape restored the excellent high-angle-of-attack characteristics that were pioneered on the F/A-18A aircraft.

4) The wing thickness-to-chord (t/c) ratio wasincreased. The C/D aircraft has a t/c of 5 percent at thewing root and a linear reduction from there to 3.5percent at the wing fold. It is constant, 3.5 percent,from the fold to the tip. The E/F has a t/c at the wingroot of 6.2 percent, tapering to 5.5 percent at the wingfold and further tapering to 4.3 percent at the wing tip.The increased t/c provides an increase in torsionalstiffness with no increase in structural weight. It alsoallowed increased fuel carriage in the wing. However,the penalty is an increase in supersonic drag. Theincrease in torsional stiffness completely eliminateslimit cycle oscillations when the aircraft is carryingexternal stores as has been verified by the flight testprogram.

5) A third wing station was added. This significantlyenhanced self escort capability and gave the aircraftadditional load carrying capability of 2,300 pounds.These new wing stations can be used for either air-to-airor air-to-ground weapons.

6) The inlets were enlarged for the increased airflowrequired by the F-414 engines and reshaped forimproved radar signature. This reshaped inlet is clearlyvisible in Figure 8.

GP81128008.cvs

Figure 8. F/A-18E/F Super Hornet ReshapedEngine Inlet

There are additional changes below the skin. Theseinclude substantially new structure, new mechanicalsystems, and modified cockpit displays. The avionics,however, are ninety percent common between the twoaircraft. The reasons behind these design changes canbe related to the design requirements described in thenext section.

Description of the Design Process

Integrated Product Development (IPD) - During the1980s McDonnell Douglas ran several pilot programs totest what was then an innovative concept for aircraftdesign called Integrated Product Development and thisprocess played a significant role in the design of theF/A-18E/F. IPD is the process of defining, designing,developing, producing, and supporting a product, using


a multidiscipline team approach. Note that theindividuals on the team need not be multidiscipline butrather that the team has the required disciplines toperform its job. IPD, also known as concurrentengineering, pertains to the concept, analysis, anddesign stages of a product and provides the basis forbringing the optimum new product or product versionto production in the shortest time. The word optimum asused here may not imply the same thing as one wouldobtain from a formal mathematical process. IPDencompasses the product life-cycle from initial conceptthrough production and support. IPD also includesIntegrated Product Definition plus product upgradesand process improvement for the life of the product.Integrated Product Definition is a subset of IntegratedProduct Development.

IPD requires a shift from serial to concurrent processstructures. Traditionally, each discipline completed itstasks and passed the results on to the next disciplineresulting in a sequential, or serialized, developmentprocess which generated rework because the delivereditem did not fulfill the down stream customer’srequirements, was incomplete, or was changed afterrelease. Several iterations may be required to get theproduct delivered, corrected, and completed. Theprocesses involved in the definition of a product haveserial tasks. The IPD process strives to take the serialprocesses and perform as many of them concurrently aspossible. Concurrent performance of sequential tasksrequires redesign of those tasks to accommodate thenew processes.

The IPD approach to product development has sixdefinition phases that are shown in Figure 9. The firstfour phases are referred to as configuration synthesisand the last two are referred to as product/process

GP81128009.cvs

High Level Requirements Definition

Initial Concepts

Configuration Baselines

Conceptual Layout

Assembly Layout

Build/Buy/Support Packages

ConfigurationSynthesis

Product/ProcessDevelopment

Hornet2000and

Pre EMD

EMD

Figure 9. The Six Phases of IntegratedProduct Development

development. At the end of configuration synthesis aconceptual layout of the aircraft is available and at the

end of the product development phase, the build to /buy to packages are defined. For the F/A-18E/F, theHornet 2000 study corresponds to the configurationsynthesis portion of the IPD process.

The six phases of product definition are executed duringthe DoD Acquisition Phases as shown in Figure 10.Each acquisition phase will satisfy certain milestonerequirements before contracts are let for subsequentphases. Configuration synthesis, consisting of high levelrequirements, initial concepts, and configurationbaseline definition phases, is executed during theconcept exploration and development acquisition phase.The conceptual layout definition phase of configurationsynthesis will occur during the Demonstration andValidation (DEM/VAL) acquisition phase. Theassembly layout and build-to and support-to-packagedefinition phases, for product and process development,are accomplished during the EMD acquisition phase.

GP81128010.cvs

Initi

al F

light

Des

ign

Mod

s

Low

Rat

e

Pre

-Pro

duct

ion

Bui

ld

Bui

ld-T

o, B

uy-T

o,S

uppo

rt-T

o

Ass

embl

y La

yout

Con

cept

ual L

ayou

ts

Con

figur

atio

n B

asel

ine

Initi

al C

once

pts

Hig

h Le

vel R

eqs

Build Evaluation/Production PhasesIPD Phases

Phase 3Production

Phase 2Engineering andManufacturing

Development Phase

Phase 0Concept Exploration

Pha

se 0

Dem

/Val

Program Phases

Figure 10. IPD Phases Related to the MajorDoD Acquisition Phases

The F/A-18E/F program hierarchical team structurefollowed the Work Breakdown Structure (WBS),segmenting the work into discrete elements forestimating and budget allocation, tracking, andperformance as shown in Figure 11. Budgets wereallocated to each product center and team, making iteasier for the team leader to manage the assignedwork and maintain control of budget and schedule. Eachlevel could then be assigned the responsibility,authority, and accountability for their product.

The E/F program was managed under the Cost ScheduleControl System or C/SCS. This system works with theWBS defined above along with a detailed schedule andcost for each task. Metrics in the form of a CostPerformance Index (CPI) and a Schedule PerformanceIndex (SPI) are two of the tools that were


SystemEngineering/

ProjectManagement

SystemTest and

Evaluation

AirVehicle

TechnologyAvionicsPropulsion

Secondary Power

Hydraulic

IntegratedLogisticsSupport

GP81128012.cvs

Airframe

Level III

Level I

Weapon SystemWBS/ProductTeam Level

Armament/WeaponDelivery

Structures

Flight Control Fuel System

Level IV

Forward FuselageWing Horizontal Tail Vertical Tail

Level V

Inner Wing Outer Wing LE Devices

Level VI

Product Teams Align With WBS

Level II

Figure 11. Work Breakdown Structure asImplemented on the F/A-18E/F

used to ensure that the E/F Program remained onschedule and within budget. These two indices providethe following information. The CPI is a measure of thework accomplished versus what it cost to accomplish it.This is an indication of the cost efficiency with whichwork has been accomplished. The SPI is a measure ofthe work accomplished versus what was scheduled to beaccomplished. This is a measure of the scheduleefficiency with which the work has been accomplished.These indices, along, with others were applied to thetasks defined through the WBS. Results were reportedto the program managers so that they always knewwhere they stood relative to cost and schedule. It shouldbe noted that the E/F program has basically remained oncost and on schedule since contract award in 1992.

Design Requirements - While the F/A-18C/D hasperformed well and demonstrated that the concept of amulti-mission aircraft is valid, usage also showedseveral areas where the aircraft could be improved.During the advanced design process, a number ofrequirements were investigated using standard tradestudies and a final set of requirements was formulatedand an enlarged aircraft that met these requirements wasdefined. These requirements were formulated relative tothe C/D aircraft. In addition to the requirements definedbelow, if a requirement were not specifically identified,it was implicitly assumed that the E/F would be as goodas or better than the C/D aircraft.

These requirements covered five areas where increasedcapability was desired. These were:

1) Increased Bring Back - The maximum weight ofordnance and fuel with which the aircraft can land onthe carrier has been increased from 5,500 lb to 9,000 lb.

2) Increased Payload - The aircraft store stations havebeen increased from 9 to 11 and can be used for eitherair-to-air or air-to-ground weapons.

3) Increased Range - The maximum range of theaircraft has been extended up to 40 percent dependingon the mission.

4) Increased Survivability - The ability to avoiddamage from hostile forces was improved by up to 8times depending on the threat.

5) Growth - Space for new hardware as well aselectrical power and cooling capability have beenincreased by up to 65 percent.

These requirements were quantified and in effectbecame constraints that the design had to satisfy. Inaddition to the requirements described above a set ofTechnical Performance Measurements (TPMs) weredefined which were allocated as appropriate to the IPDteams and were tracked for the aircraft. These TPMswere: weight empty, reliability, maintainability,survivability, signature, average unit airframe cost,growth in terms of internal volume, electrical power,and cooling, and built-in test which was tracked as falsealarm rate and fault detection and fault isolation. Inaddition each team had requirements for cost, schedule,and risk.

Each of the TPMs was tracked in terms of its currentvalue relative to a design-to value and a specificationvalue. Figure 12 shows this tracking process as afunction of time for empty weight. The chart shows thatas of May 98 the actual weight was 666 lbs. above thedesign-to weight. However, this weight was over 384lbs. below the spec value. Thus, while weight was notbeing minimized as an objective function, its value wasbeing closely tracked to ensure that its upper limit wasnot exceeded. In addition the weight was being keptbelow the spec value in anticipation that changes mightbe required as a result of EMD testing. Similar trackingwas carried out for all of the TPMs.

If the strict definition of MDO is used, MDO was notused to design the F/A-18E/F. However amultidiscipline process that produced a design thatsatisfies all of the constraints was used. As an example,the technology disciplines of aerodynamics, flightcontrol flying qualities, structural loads and dynamics,and materials and structural development were linked


GP81128013.cvs

31.0

30.5

30.0

29.5

29.0JA S OND J FMA MJ J A S OND J FMA MJ J A S ON DJ F MA

1997 19981996

Pou

nds

(1,0

00)

Good

T&E Allowance

EMD Margin

Design-To..........29,514 lb

Out of Tolerance

Current Status....30,180 lb

Variance................. 666 lb

Design-To......0%Estimated...... 2%Calculated... 23%Actual.......... 75%Total.......... 100%

Current Status Basis

1999

Figure 12. F/A-18E/F Empty Weight

through a common database and analysis tools as shownin Figure 13. Each of these disciplines is driven by aspecific set of requirements and each is responsible fora given set of products that taken together define theairplane.

For example, the structural loads and dynamics group isresponsible for design loads, the dynamic environment,and aeroelastic stability. In order to accomplish thiseach discipline must communicate with the otherdisciplines. One tool used to accomplish this was theuse of a common database.

Taken as a whole the interactions among thesedisciplines produce a balanced set of requirements.

GP81128014.cvs

LoadsDatabase

StructuralLoads andDynamics

ControlLaws

FlightControl Flying

Qualities

Common Database and Analysis Tools

Interactions Produce Balanced Requirements

• Mission Performance• Carrier Suit Perform• Weapon Separation Requirements

• Flying Qual Criteria• Control Laws• FCC Hardware/ Software

• Design Loads• Dynamic Environment• Aeroelastic Stability Requirements

• Design Allowables• Composite Allowables• PPV/Full Scale Test Requirements

AeroDatabase

MaterialsDatabase

Requirements

Teams

AerodynamicsMaterials

and StructuralDevelopment

AIR 4.3.2

AIR 4.3.3

AIR 4.3.4

Figure 13. Airframe Technology Key Productsand Requirements

The interactions among the disciplines can be viewedfrom the standpoint of common tools as well ascommon data. An example of this is shown in Figure14. In this case a tool referred to as MODSDF which isa six degree-of-freedom simulation code is being usedto determine critical design loads. For this tool to work,input is required from several sources. These inputs canbe in the form of criteria, such as Mil Specs or datasuch as mass properties. One of the ingredients is pastexperience.

GP81128015.cvs

Criteria MODSDF LoadCritical

Load/StructureCriteria

LoadsWT Tests

StructuralStiffness FEMS

Aero WTTests

DetailedSpec

Load AeroCoeff's A/C

Aero

InterfaceSubroutinesWing, Pylon,

etc

DesignTeamInput

PastExperience

MilSpecs

MassProperties

ControlLaws

PropulsionRequirements

Figure 14. Flight Technology RequirementsDevelopment

An example of how this process can be used to improvethe design is shown in Figure 15. In this case the trailingedge flap was used as a maneuver load alleviationdevice and its effectiveness was determined using theMODSDF code. As the aircraft pulls load factor thetrailing edge flap is scheduled down by the flightcontrol system as a function of load factor. Theresult is a modification of the lift distribution with lesslift on the outer panel of the wing and more on the innerpanel. This reduces wing bending moment, whichresults in a reduction in wing weight. This process is anexample of a multidisciplinary approach to design thatproduces a better aircraft than would be possible if eachdiscipline simply worked alone.

One final point about the design process needs to bemade and that is the importance of the aerodynamicdatabase. Figure 14 shows that two of the drivers for theMODSDF code are the aerodynamic wind tunnel testsand the loads wind tunnel tests. The generation of thisdata is one of the key ingredients in the design process.During the period from the start of EMD in 1992 to firstflight in 1996, approximately 18,000 hours of windtunnel occupancy time was accumulated with more thanhalf of this being used by aerodynamics. In addition tothe MODSDF simulation tool, pilot in the loopsimulation is also extremely important and over


Figure 15. Load Alleviation of Wing-Fold andWing-Root

1000 hours of pilot in the loop simulation wereaccumulated by first flight. Both of these simulationtools require a detailed aerodynamic database.

Results

The F/A-18E/F has completed the majority (90%) of itsflight test program and the results to date have beenoutstanding. The program is on schedule, on cost, andthe aircraft is below the specification weight. Theaircraft has met all of its requirements and will providethe Navy with an aircraft that will meet its needs wellinto the 21st century. While these results validate thedesign process that was used for the F/A-18E/F aircraft,it is always possible to improve. What follows is adiscussion of a series of questions from the sessionorganizers concerning what is needed in the MDOprocess. This discussion is based on the experiencefrom F/A-18E/F program and other experience of theauthors.

Barriers, Obstacles, etc. - The major obstacle to MDOis the inability to analytically determine the designvariables and their sensitivities. Meaningful design doesnot occur until the wind tunnel data base has beendetermined. While Computational Fluid Dynamics(CFD) may ultimately replace the wind tunnel, until thishappens the aerodynamic model cannot be coupled withthe other disciplines. Organizational barriers can exist.However, the F/A-18E/F program showed that the

transition to an IPD organization is possible. Surely, thetransition to an MDO based organization is possibleonce the benefits are demonstrated.

Design Problem and Design Goal - The goal is todesign an aircraft that satisfies the requirements. AnMDO code should aid in making the design feasible asrapidly as possible. Once a feasible design has beenfound, the next most important thing is to determine therobustness of the design.

State of Software Integration Tools - Tools such asdatabase management, simulation, distributedcomputing, etc. have all contributed to the integration ofthe design process. The F/A-18E/F uses a commondatabase for aerodynamics, control dynamics, loads,and structures.

MDO Simulation for non-linear Loads - Thesignificant challenge here is the generation of the non-linear aerodynamic database. Once this data base isgenerated, the simulation and the control law design canproceed. Once these elements are in place, loadscalculations can proceed.

Barriers to the Use of Disciplinary Analysis in MDO- While several issues were identified, fidelity of themodels is the most significant. It makes no sense tooptimize a design based on low fidelity data.

Loosely Coupled versus Tightly Coupled Approach -There is no inherent reason why a tightly coupledapproach could not be used. However, it is difficult tosee how a tightly coupled approach could contain all ofthe constraints that are present in the loosely coupledone that makes use of all current detailed design tools.A tightly coupled code run by an expert could serve as acheck on the more detailed loosely coupled approach.However, this could also create conflict if the twomethods don’t agree.

Use of Sensitivity Derivatives - The use of sensitivityderivatives will become wide spread only after thedesign community becomes familiar with them. Atpresent the concept of dollars-per-pound is well understood by all designers but it is not clear that allsensitivity derivatives in general are in this category. Onthe other hand, a trade study where two variables arecompared directly can usually be understood by anyone.

Automatic Differentiation - The trend in industry istoward off-the-shelf software when possible. Extendingthis to automatic differentiation might imply that thesoftware vendors should assume the lead here.

Single Most Important Obstacle to MDO - Theaerodynamic model matures first and the other modelsdepend on this one. An accurate aerodynamic model is


based on wind tunnel data that may not produce thesensitivity derivatives needed for MDO.

Use of MDO Based on Decomposition - Since MDOtools as such did not play a major role in the F/A-18E/Fdesign, there is no reason to single out any particularmethod. However, before any method will be used in aproduction aircraft design environment, it will have toprove itself.

Top MD Development - Rapid CFD for air loads.

Summary/Conclusions

The design process for a modern high performanceaircraft is a complex process that involves theintegration of analyses, tests, databases, and finally thepeople who make the process happen. For the F/A-18E/F aircraft these ingredients have come together toproduce a superior product. While the process did notmake use of mathematical optimization in a formalsense, the final product does indeed satisfy all of thedesign requirements that would be represented in theform of constraints in the MDO process. In fact, sincethe F/A-18E/F is a multi-role aircraft, the formulation ofa single objective function would be difficult if notimpossible. The following observations can be made:

1. Formal MDO was not used as part of the F/A-18E/Fdesign process.

2. For a multi-mission aircraft, the formulation of anobjective function is difficult if not impossible todefine.

3. The aircraft is designed by its requirements. This isanother way of saying that the aircraft is designed tomeet a set of constraints.

4. The design process involves more than a coupling ofmathematical tools. The people who operate these toolsare an essential ingredient.

5. The IPD design process contributed to the success ofthe F/A-18E/F program.

6. The design process is serial in that an aerodynamicdatabase is required to design the flight control system.Both the aerodynamic database and the flight controlsystem are required to define loads. Loads are requiredto define structure. Flex-to-rigid ratios are defined afterthe structure is sized. These ratios are used to correctthe aerodynamic database. And the whole process isiterated. All of this can be done once the moldline of theaircraft is defined.

7. The aerodynamic database is the key. This databaseis very non-linear. For the F/A-18E/F, the aerodynamicdatabase was established by wind tunnel testing. In the

future CFD may have the capability to generate thisdatabase.

8. For a multi-mission aircraft, MDO tools that rapidlygenerate a feasible design, one that satisfies therequirements, would be valuable. Once the design isfeasible, these tools should allow for rapid “what if”studies. The manufacturer and his customer shouldmake the ultimate decision for what is best to meet therequirements.

References

1. “Current State of the Art on Multidisciplinary DesignOptimization,” An AIAA White Paper, Approved bythe AIAA Technical Activities Committee, September1991.

2. Sobieszczanski-Sobieski, J. And Haftka, R. T.,“Multidisciplinary Aerospace Design Optimization:Survey of Recent Developments,” AIAA Paper No. 96-0711, January 1996.

3. Venkayya, V. B., “Introduction: HistoricalPerspective and Future Directions,” published inStructural Optimization: Status and Promise, edited byManohar P. Kamat, Progress In Astronautics andAeronautics, published by the AIAA, 1993.

4. Yurkovich, R. N., “MDO from the Perspective of aFighter Aircraft Manufacturer,” published in“Multidisciplinary Aircraft Design,” Proceedings ofIndustry-University Workshop, Virginia PolytechnicInstitute and State University, compiled by R. T.Haftka, et. al., pp. 49-80, May 1993.

5. Anon. “Hornet for 2000 Final Report,” TheMcDonnell Douglas Corporation MDC Report B0833,29 February 1988.

A Summary of Industry MDO Applications and Needs

byJoseph P. Giesing The Boeing Company Jean-Francois M. Barthelemy NASA Langley Research Center

AIAA 98-4737

7th AIAA/USAF/NASA/ISSMOSymposium on Multidisciplinary Analysis and Optimization

Sept. 2-4, 1998 St. Louis MO.

1


OUTLINE

• Introduction

• 10 Invited Papers Synopsis Process

• Development of MDO Categories (Taxonomy)-Process of Extracting Salient Points from Invited Papers

• Discussion of Categories- Challenges and Issues- Needs ( in Industry)

• Conclusions- Satisfying MDO Development Needs- Concluding Remarks

2

INTRODUCTION


• Last AIAA MDO Technical Committee White Paper 1991- Technology Push, Providing Benefits of MDO

• Current White Paper Meant to be a Technology Pull from Industry- Industry Needs in the Area of MDO- Provide MDO Developers Help in Planning and Direction

• White Paper Process- 10 Invited Papers From Industry- Plus a Summary Paper - Summary Paper to be Reviewed by MDO TC and Invited Authors- 10 Papers Plus Summary Will be Put on MDO TC Web Site

3


SYNOPSIS OF 10 INVITED PAPERS

• Summary Paper Presents Short Synopsis of Each Invited Paper- Basic Design Problem Summarized- Several Highlights of the Main Points

FrontierFidelityLevel

Level of MDOTradeStudies

Full MDO

Limited Optimization/Iteration

F/A-18 E/F

ConceptualDesign

IntermediateFidelity

High FidelityCFD, FEM F-16

Agile Falcon

GM Auto

BWB

GE Engine

Rotocraft

Lrg A/C

Incr

easin

g Diff

iculty

f-22

SpaceTelescope

4

ConceptualDesign


High FidelityCFD, FEM

f-22



Full MDO


F/A-18 E/F

ConceptualDesign



Agile Falcon

GM Auto

BWB

GE EngineRotocraft

Lrg A/C

Incr

easin

g

Difficu

lty

f-22

SpaceTelescope


DEVELOPMENT OF CATEGORIES

Questions Asked by an Industrial Designer

1) What are my design objectives and critical constraints2) What are my disciplinary analysis capabilities/limitations/Automation level3) How do I get critical high fidelity elements into my design in an efficient manner?4) What design process steps are needed to meet my design objective most efficiently and to know that I have reached my objectives and satisfied my constraints? 5) What MDO or design formulation do I need or what formulations are available to me?6) What kind of approximation analyses are required?7) How do I overcome Optimization problems (scaling, smoothness, robustness, effic.)?8) How do I feed data among disciplinary analyses and the MDO process?9) How do I overcome computing and data handling issues10) What is the easiest way to visualize my design space?11) How robust is my design and how do I check it? 12) Are there commercial systems that can effectively help me?13) How do I make it all happen at my plant?

5


Design Formulations &Solutions

• Design Problem Objectives• Design Problem Decomposition,Organization• Optimization Procedures and Issues

Analysis Capabilities & Approximations

• Breadth vs.. Depth Requirements• Effective Incl. of High Fidelity Analyses/Test• Approximation & Correction Processes• Parametric Geometric Modeling• Analysis and Sensitivity Capability

MDO Elements

Information Management & Processing

• MDO Framework and Architecture• Data Bases and Data Flow & Standards• Computing Requirements• Design Space Visualization

Management & Cultural Implementation

• Organizational Structure• MDO Operation in IPD Teams• Acceptance, Validation,Cost &, Benefits• Training

FINAL CATEGORIES (MDO TAXONOMY)

6


One L

iner

Sal

ient P

oint

s

One Liner Salient Points

InvitedPapers

Design Formulations (1) Design Prob. Obj. (2) Decomp., Organiz. (3) Opt. Proc. & Issues

Analysis & Approx. (4) Breadth/Depth. (5) Approximations. (6) High Fidelity (7) Parametric Models (8) Analysis/Sensit.

Information Mngt. (9) MDO Frameworks (10) Data Bases Stds. (11) Computing Req. (12) Des. Spc. Visual..

Management & Culture (13) Organization (14) IPD Teams & MDO (15) Acceptance & Benefits (16) Training.


One Liner Salient PointsOne Liner Salient Points


MDO ELEMENTS (TAXONOMY)

PROCESS OF EXTRACTING SALIENT POINTS FROM INVITED PAPERS AND PLACING THEM INTO THE MDO ELEMENT CATEGORIES

7


(1) Design Problem Statement

SpaceTelescope

ImprovedFeasible Optimal Pareto

F/A-18 E/F

F-16Agile Falcon

GM Auto RotocraftGE Engine

Lrg A/C BWB

f-22

Frontier

Distribution of Design Problems for 10 Papers

• Industry Design Objective Priority Order- Feasible and Viable Design- Robust Design- Improved Design- Optimal Design

DISCUSSION OF CATEGORIES

8


Challenges and Issues

(1) Design Problem Statement

Needs

• Flexible Framework - Reconfigurable to Multiple User Needs

• Continued Development of Objective Functions for Industrial Applications

• Each Design Problem Unique• Design Problem May Not be Known A-Priori

9

DOC/DOCo =A +B OEWA/OEWAo + C FWT/FWTo + D PLWT/PLWTo+ E Swt/Swto +F T/To


Comparison of Three Objective Functions; Area DOC, MTOGW, Weight DOC

0

0.1

0.2

0.3

0.4

0.5

0.6

Const OEW FuelWeight

PayloadWeight

WettedArea

EngineSize

DO

C/D

OC

o (

or F

/Fo)

Area DOC

MTOGW

Wt. DOC

SIMPLIFIED COST RELATED OBJECTIVE FUNCTION FOR MDO

A B C D E F

10


(2) Design Problem Decomposition and Organization


Needs• Loosely Coupled Systems

- Include Legacy Codes- Global-Local (Multi-Level) Decomposition - Easy to Understand Processes

• Decomposition Processes that Converge to High Fidelity Results Without High Fidelity Analyses being Called Directly by Optimizer

- Update Processes- Approximation Processes- Other

• Decomposition Processes Tailored & Adapted to Needs and Deficiencies of Analysis Processes

• Sophisticated Decomposition Processes (e.g.. CO , CSSO )- Not Fully Mature - Not Fully Understood by Industry

• High Fidelity Analysis Processes Difficult or Impossible to Include in MDO- Non Automated & Very Long Computing Time

11

UpdateApproximation Surfaceor Intermed. Level Analysis

Update #1

Update #2

Update #3

ITERATION

OBJECTIVEFUNCTION F


Notional Update Process

(2) Design Problem Decomposition and Organization

Trust Region

High FidelityAnalysis

12


(3) Optimization Procedures and Issues


• Lack of Experience in Optimization in Industry• Optimization Robustness

- Smoothness Requirements- Scaling, Convergence Issues

• Efficiency• Continuous, Discrete & Hybrid Optimization• Local Minima

Needs

• Self-Smoothing or Noise Insensitive Opt. Processes• Self-Scaling Opt. Processes• Robust Processes for Finding Global Minimum• Rapid/Efficient Optimization Processes

13


(4) Breadth and Depth Requirements


• Identify and Include All Critical Constraints to Avoid Academic Design• Identify and Include All Critical Physical Mechanisms to take Advantage of Available Design Opportunities• Fidelity Requirements for Each Discipline not Known/Quantified

Needs

• Process for IPD Team to Identify All Critical Aspects of Design as it Progresses• Process for Identifying the Fidelity Requirements of Various Disciplines

- Possible Use of MDO Process Itself (Sensitivities) to Est. Req.• Process for Identifying Critical Physical Mechanisms

14

Effect of Objectives and Constraintson Optimal Design


Minimum total drag, fixed weight, low speed lift constraints, fuel inertia relief, and static aeroelasticity.

EMinimum total drag, fixed weight, low speed lift constraints, and fuel inertia relief.

D

Minimum total drag at fixed weight with low speed lift constraints.

C

Minimum total drag at fixed weight.BMinimum induced drag at fixed weight.A

15


(5) Effective Inclusion of High Fidelity Analyses/Test


• High Fidelity Process Deficient in- Automation ( Many Manual Steps)- Robustness (Model has to be Iterated and Re-worked)- Efficiency ( Requires Many Hours on the Computer)

Needs

• Advances in Disciplinary State-of-Art - Robustness- Efficiency

• Advances in Disciplinary Automation and Parametric Modeling• Advances in Decomposition or Approximations to Make Up for Deficiencies in High Fidelity Analyses

16

Semispan Fraction

C

* c/

cL

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

TLNS3D N-S, C = 0.4654, αα = 10.0°°Ltot Woodward Linear, C = 0.4690, αα = 11.5°°Ltot

Nonlinear (Reduced Bending Moment)

Linear

Semispan Fraction

C

* c/

cL

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

TLNS3D N-S, C = 0.2803, αα = 4.5°°Ltot Woodward Linear, C = 0.2795, αα = 4.7°°Ltot

Nonlinear (Increased Bending Moment)

Linear


17


(6) Approximation and Correction Processes


• Generating Data for Response Surfaces (Curse of Dimensionality)• Isolating Physical Mechanisms • Intermediate Level Analyses Not Simulating All Critical Physical Mechanisms

Needs

• Response Surface and Other Generic Approximation Software• Addition of Missing Critical Mechanisms in Intermediate Level Analyses• Advanced Correction Procedures for Intermediate Level Analyses

- Separate Correction of Each Physical Mechanism - Using Intermediate Analyses as Interpolation/Extrapolation Process

• Reduced Order Approximations for Use in Optimization- Parameter Identification

18



High FidelityAnalysis orSub-Optimiz

GenericApproximations -Response Surf. -Neural Nets -Taylor Series

IntermediateFidelity Analysis - Phy. Mech. A - Phy. Mech. B - Phy. Mech. C


Correction Process


Reduced OrderRepresentations

Optimizer

Optimizer

Optimizer

Rational Function Approx.

A=ΣΣ a /(S + b )i i

19

Aerodynamic Span Loading on Deformed HSCT Configuration at 15 Degrees Angle of Attack

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 10 20 30 40 50 60 70

Span Station (Inches)

Correction Procedure

Linear

Direct CFD Solution

Subsonic, High αα Case (M=0.5, AOA=15o)Sectional Lift Distribution



Correction Process

20


(7) Parametric Geometric Modeling


• Large Common Models Expensive• Existing CAD Software Not Robust Enough for Topology Optimization• Morphing (Rubberizing) Does Not Always Produce Adequate Layout

Needs

• Automatic Modeling Tool Kit (e.g.. CFD Meshes, Finite Element Models)• Parametric Layout Techniques for Changing Topology• Grid-Mesh Mapping (e.g.. aerodynamic forces on structural nodes and surface deflection of FEM on CFD mesh)• Software for Robust Processes (Commercial or Otherwise)

21

c1 c1

c2c2

c3 c3


(7) Parametric Geometric Modeling

Parametric Model with Topological Changes Can Not be Rubberized

22


(8) Analysis and Sensitivity Capability


• Lack of Automation of High Fidelity Codes and Sub-Optimization Processes• Lack of Cost Models for Use in MDO• Checking of Analysis Model and Data At Each Step• Large Computer Run Times for High Fidelity Codes

Needs

• Robust Automated Disciplinary Analyses Modules (Preferably Commercial)- CFD, FEM, Nonlinear Loads, Aeroservoelastic- Global/Local Structural Sizing- Efficient Aerodynamic Optimization- Other

• Interactive Analysis Data Monitoring and Checking Tools• Simplified and Detailed Manufacturing and Maintenance Cost and Constraint Models

23


(9) MDO Frameworks and Architecture


• Commercial OTS MDO Frameworks are Not Yet Industrial Strength- Problem and Model Size Limited- Distributed Computing Not Robust

Needs

• Commercial MDO Framework that is:- Mature, Robust, Efficient, and Industrial Strength - Flexible and User Friendly- Able to Include Legacy Codes

24


(10) Data Bases, Data Flow and Standards


• No Universal Standard for Data• Hugh Amounts of Data are Used in Industrial Design• Multiple Platforms and Locations Need to Interact with Design Data

Needs

• Standardized, Industrial Strength Data Base - Handle Large Amounts of Data (terabites)- Multisite and Heterogeneous Accessible ( Internet?)- Efficient and User Friendly

25


(11) Computing Requirements


• High Fidelity Analyses Require Massive Computing Power- CFD Single Analysis 10 hrs on C-90 Type Super Computer- CFD Design (20 design variables) 300 hrs C-90- Solution 200 (structural sizing optimization) 50 hours for 9 Iterations on a High End Work Station- F-22 Used 10 Terabites of Storage for Structural Design

Needs

• Improved Networking Systems for Work Stations and Other Computers• New Methods that Work Efficiently on Massively Parallized Computers (e.g. CFD, Structural Sizing Optimization)• Generic Algorithms Designed Especially for Massively Parallized Computers (e.g. Matrix Manipulation, Eiganvalue Analysis)

26


(12) Design Space Visualization


• Can Not Physically Visualize More than 3-Dimensions• Designers May Be More Interested in Seeing the Design Space Than Finding the Optimum Design Point - e.g. How Flat is Design Space?• IPD Team Needs to Understand the Design Space to Make Design Decisions

Needs

• Creative Design Space “Depiction” Techniques Needed- Visualize Multi-Dimensional Design Space Directly- Generate a New Breed of Data that Will Impart the Needed

Design Space Information Without Direct Multi-Dimensional Representations

*Reduced Dimensions (Modal or Related Approach)27


(13) Organizational Structure


• Who is In Charge of MDO?• Currently Advanced Design Group Responsible for Own Technology

- Small Interaction with Discipline Groups• Each Discipline Responsible for Methods and Data Integrity/Quality

- How Do These Groups Interact with MDO?- How Do Disciplines “Buy In” to the MDO Process & Results?

• MDO Solutions Tend to Compromise Performance of Each Discipline for the Benefit of the Whole System

Needs

• Industry Needs an MDO Team - One Part of Team Provides Coordination (Advanced Design?)- Each Discipline Maintains Technical Autonomy - Each Discipline Is Part of and “Buy’s In” to the Process- Each Discipline Maintains Responsibility for Integrity and Quality of Data and Technology- All Disciplines and MDO Coordination Agree on Interfaces

28


(14) MDO Operation in IPD Teams


• Not All Design Issues are Incorporated into MDO • IPD Team Unfamiliar with MDO Tool and How to Interact with It• Defining MDO Problem is Evolutionary as Design Progresses

Needs

• Understanding the MDO Process and Accepting It As a Tool• Experience and Training of IPD Teams

- Interaction and Direction of MDO Process- Setting and Changing Constraints and Groundrules for MDO Process- Interpreting Results

29

mission analysis

weights, compressibility drag,

max lift datareasonable

design?yes

no

check design, modify analyses or constraints

prospective design, cruise spanload

acceptable aerodynamics?

recalibrate buffet constraint

preliminary loads, sizing

validate weights, advanced design

validate performance

yes

no

detailed sizing and loads

FLO22, CFL3D

Finite Element

CASES

CWEP

WingMOD

CASES

WingMOD

Final Design

initial design 3 months


(14) MDO Operation in IPD Teams

IPD Team Members- Structures- Aerodynamics- Manufacturing- Stability and Control

IPD Direction

30


(15) Acceptance, Cost and Benefits and (16)Training


• Lack of Understanding of MDO and Its Place in Industry Environment• Lack of Training and Education in MDO Techniques• Cost of Developing a New MDO System • Lack of Documented Practical Benefits• Fear of Loosing Tried and True Processes

Needs

• Series of Full Industrial Validation Test Cases- Benefit Over Current Practices- Industrial Strength Validation Cases, Preferably on Actual Vehicle

• MDO Process Plan and Cost

31


SATISFYING MDO DEVELOPMENT NEEDS

• Industry •Government Labs.• University •Commercial Software

GovernmentContracts &R&T Base

IndustryIR&D

University

Commer. SoftwareInvest.

Financial Resources

IndustrialStrength

MDO

Technical Challenges and Needs

Financial Leveraging

TechnologyDevelopment& Verification

TechnologyTransfer &Applications

TechnologyCommercialization

TEAM• University• Gov. Labs• Com. Soft.• Industry

Technology Development Community

32


CONCLUSIONS

• Broad Cross-Section of MDO Applications and Experience in Industry Represented

• Wide Range of Design Objectives Encountered

• Modified Taxonomy of MDO Elements Suggested

• Practical MDO Challenges and Issues Delineated

• Industry MDO Development Needs Presented

• Teaming of University, Government Labs, Commercial Software Developers, and Industry Suggested

- Produce Industrial Strength MDO Processes for the Future

33

1

A SUMMARY OF INDUSTRY MDO APPLICATIONS AND NEEDS

Joseph P. Giesing, The Boeing Company, Long Beach, CA1

Jean-François M. Barthelemy, NASA/Langley Research Center, VA2

1 Boeing Technical Fellow, Associate Fellow, AIAA2 Manager, Aircraft Morphing, Airframe System Program Office, Senior Member, AIAA.

Copyright © 1998 by the American Institute of Aeronautics and Astronautics, Inc. No copyright is asserted in the United States under Title 17, U.S. Code.The U.S. Government has a royalty-free license to exercise all rights under the copyright claimed herein for Government purposes. All other rights arereserved by the copyright owner.

Abstract

The AIAA MDO Technical Committee has sponsored aseries of 10 invited papers dealing with industry (andrelated) design processes, experiences, and needs. Thispaper presents a summary of these papers with emphasison the needs of industry in the area of MDO. Togetherthe 10 invited papers and this summary paper comprise anAIAA MDO Technical Committee “White Paper” on thissubject. This summary paper contains; 1) a short synopsisof each paper and the industrial design it describes, 2) asorting of all of the salient points of each of the papersinto MDO categories plus a discussion of each category,and 3), a summary of industrial needs distilled from thepapers. It is hoped that this summary paper will provide atechnology “pull” to the MDO technology developmentcommunity by presenting the industrial viewpoint ondesign and by reflecting industrial MDO priorities andneeds.

1. Introduction

Upon the establishment of the Multidisciplinary DesignOptimization Technical Committee (MDO/TC), a WhitePaper was prepared to assess the State of the Art in theMDO technical area1. Jointly written by foundingmembers of the TC, the paper provided a brief history ofaerospace design and made the case for integrating all thedisciplines in the design process. The White Paper thenreviewed recent developments, addressing in turn thehuman interface aspects of design, its computationalaspects and its optimization aspects. The discussioncontinued with an approach to transitioning the designenvironment to Concurrent Engineering and a discussionof how MDO can support that transition. The WhitePaper finally concluded by stating that MDO provides ahuman-centered environment 1) for the design of complexsystems, where conflicting technical and economic

requirements must be rationally balanced, 2) thatcompresses the design cycle by enabling a concurrentengineering process where all the disciplines areconsidered early in the design process, while thereremains much design freedom and key trades can beeffected for an overall system optimum, 3) that is adaptiveas various analysis/simulation capabilities can be insertedas the design progresses and the team of designers tailortheir tool to the need of the moment, and 4) that contains anumber of generic tools that permit the integration, of thevarious analysis capabilities, together with their sensitivityanalyses and that supports a number of decision-makingproblem formulations.

Since the publication of the first White Paper, much workhas been devoted to MDO as attested in the proceedingsof the successive AIAA MA&O Symposia, for example.A number of detailed surveys have been written (seeSobieski and Haftka2, for example), updating the researchcommunity to the latest developments in MDO in general,and in some subareas of MDO as well. The MDO/TC istaking the occasion of the current (7th) MA&Osymposium to add to the constant dialogue between MDOusers and MDO researchers. It invited designers fromvarious organizations to contribute a technical paperdescribing a recent design exercise in which they havebeen involved and to take that opportunity to offer someinsight into their application of formal MDO methodologyto their problem. In particular, the users were asked toaddress whether they had used MDO, whether it helped ordid not help, and what developments they needed toimprove their process. This paper is a draft synopsis ofthe lessons gleaned from the various contributions. Thepaper will be reviewed and edited by the MDO/TC and itwill be posted on the Web, together with the individualcontributions, at the same site as the 1991 White Paper.

2

It is hoped that this paper will provide some insight intowhat are the MDO developments most critical to MDOusers (industry, or others). Because this paper is directlybased on the inputs of only ten different design exercises,it cannot be presented as a consensus opinion on whatMDO should be for the engineering design processhowever it is felt that a very good representation andcross-section of industrial applications, challenges andneeds are given and that the conclusions of the datacontained here will be helpful to the MDO technologydevelopment community for prioritizing future MDOdevelopment.

For the purpose of this paper, we use the followingdefinition for MDO: A methodology for the design ofcomplex engineering systems and subsystems thatcoherently exploits the synergism of mutually interactingphenomena. One can argue that ever since systems havebeen designed, multiple conflicting requirements have hadto be taken into account and therefore multidisciplinaryprocess have always been used. This point is not debatedhere, however the key word in the definition ismethodology. MDO provides a collection of tools andmethods that permit the trade-off between differentdisciplines involved in the design process. MDO is notdesign but enables it.

Ideally the MDO-based environment of the future will becentered on the IPD design team. To facilitate its use theMDO process will be interactive and will permit thedesign team to formulate its design problem in real time asthe design issues become clear. Specifically, the MDOprocess should be flexible enough so that the problemformulation, applied constraints, and the level ofsimulation can all be specified by the design team. Tofacilitate technical communication, the design team maywish to create and update a single parametric model of thesystem being designed and reshaped it (automatically) inthe course of the design. It could be used to automaticallygenerate consistent computational models forsimultaneous use in various disciplines. An environmentthat offers visibility to the process, permitting the team tomonitor progress or track changes in the problemsdependent or independent variables will be beneficial. Allalong, the process control would remain squarely in thehands of the design team. The environment could bedistributed to reflect the nature of today’s design projects.Specifically design exercises can be distributed over many

different groups, many sites, often even in differentcountries. In addition to providing a challenge to themanagement of the process, its distribution also mayprovide additional resources as it could open up a networkof computing nodes that could be harnessed to carry outthe process. The ideal environment would automaticallyroute the computational process to the mostsuitable/available resources. Since very large amounts ofdata will be generated, they could be stored in adistributed fashion as well for convenience and efficiency,but the environment would make the data readily availableto all design team members in a transparent fashion.

The paper is written from the perspective of the user ofMDO, and begins with a brief summary of the paperscontributed to the sessions by the designer teams. Then,the challenges and issues addressed by the different papersare identified and categorized, forming a taxonomy ofMDO, as perceived by the designers. The paperconcludes with an assessment of industry needs and somerecommendations for MDO development. Note thatSobieski made an earlier attempt at developing ataxonomy for MDO3 ; his efforts could be seen as a‘Technology Push’ approach at defining the needs fromMDO, being developed from an distinguished experiencein government research. The new taxonomy offered inthis paper is coming largely from the other ‘ApplicationPull’ perspective. It is expected that the combination ofboth perspectives will prove thought provoking andhelpful to the planning and development of MDOtechnology.

2. MDO Applications, A Synopsis

A short synopsis of each paper is presented in this section.The basic design problem encountered in each paper issummarized along with highlights of a few of the mainpoints made. Figure 2-1 gives a general overview ofwhere each paper lies with reference to fidelity level and“MDO level.” MDO level is loosely defined as follows.Trade studies indicate that point designs were generatedand graded relative to each other without formaloptimization. “Limited Optimizations/Iterations”indicates a disciplinary sub-optimization or one withlimited disciplinary interaction. Full MDO indicatesvehicle level optimization with most critical disciplinesinvolved.

3

Figure 2-1: Distribution of Design Process Fidelity and Level of MDO

The Challenge and Promise of Blended Wing BodyOptimizationWakayama and Kroo4 describe the application of theWingMOD MDO process to the minimization of theBWB Take-Off-Gross-Weight. The process is fullymultidisciplinary and includes design variables forplanform shape/size, mission, aerodynamic, structuralsizing/topology, fuel/payload, and trim schedule (134 inall). WingMOD uses a close-coupled approach usingintermediate fidelity disciplinary analyses for high aspectratio wing aircraft. An optimization framework (Genie)makes calls to all of the analysis routines, using finitedifferences to compute sensitivities. The aerodynamicanalyses include the vortex lattice method and quasi two-dimensional compressibility corrections. The structuralsizing and constraints are based on aeroelastic loads anddeflection analysis, simplified buckling, and stressanalysis of simple beams. The weight is based on thestructural analysis corrected by some statistical data. Awide breadth of practical constraints are considered (705in all) along with 20 design flight conditions that covermost of the critical design considerations.

One of the main points of the paper is that all criticalconstraints and disciplines (breadth) must be included toproduce a realistic/practical configuration and that allcritical physical mechanisms should be included, to somelevel of fidelity (depth), to reach the highest potentialbenefit of integrated design. The main need of the processis inclusion of CFD (mainly for propulsion/airframeintegration) into the process without rendering it

intractable. Indeed, this close-coupled system makesmany (thousands) of calls to the analysis routines.

Issues in Industrial Multidisciplinary OptimizationBennett et al5describes the application of the GM IVDA(Integrated Vehicle Design Analysis) system to themaximization of automobile fuel efficiency. The systemis composed of both commercial (ODYSSEY,NASTRAN, LPM, DYNA3D, CAL3D, ADAMS, ) andGM codes (aerodynamics, solar load, fuel economy, andothers). The user can configure the process within IVDAto produce an optimization sequence which was done forseveral examples in an ad hoc manner. The examplesdescribed included one global design variable (vehiclelength), and suboptimizations are performed in the localdisciplines (structural member cross-section design). Thelocal designs and analyses feed a results database whichis then fit with approximations. For instance theaerodynamic drag data was a neural net fit to test data.The optimizer then uses these approximations to re-designthe vehicle.

The authors make the point that, in an industrial designenvironment, the design process does not necessarily fit aparticular MD algorithm, rather, the implementation needsto be reconfigurable, on the fly. This introduces the ideaof a toolbox of MD tools and off-the-shelf specializedtools that can be interfaced automatically, with theprovision that “... a menu of appropriate actions shouldbe generated to guide the user through the process.”



Full MDO


F/A-18 E/F

ConceptualDesign



Agile Falcon

GM Auto

BWB

GE Engine

Rotocraft

Lrg A/C

Incr

easin

g Diff

iculty

f-22

SpaceTelescope

4

Boeing Rotorcraft Experience with Rotor Design andOptimizationTarzanin and Young6 describe an exercise of optimizationto reduce helicopter blade hub dynamic forces. Theobjective function is a weighted sum of hub forces andmoments. The optimization process is tightly coupled anduses an analysis simulation system maintained by severaldisciplines. Two levels of fidelity are available in thissimulation; an approximate analysis level that requires 1minute per function call, and a high fidelity level thatrequires 30 minutes per call. The authors make the pointthat the complexity of the detailed analysis led them tofully integrate their high fidelity codes, thereby obviatingthe need for any decomposition method. Optimizationcan proceed by interfacing a single optimizer with theintegrated high-fidelity analysis. Practical verification ofthe benefits of this MDO approach was obtained withwind tunnel tests.

The design space encountered in this class of problems ischaracterized by many local minima and the paperdescribes several techniques for arriving at the globaloptimum and overcoming non convergence. Some ofthese techniques involve probing unexplored portions ofthe design space by: 1) employing multiple starting points,2) initially employing loose constraints and graduallytightening them down to the required value, 3) allowingthe constants in the objective function to take largeexcursions and then adjust back to the proper value, 4)updating aeroelastic loads at various times during theoptimization.

The F-22 Structural/Aeroelastic Design Process withMDO ElementsRadovcich and Layton7 describe a process for the detailedstructural design of the F-22 aircraft after theconfiguration has been fixed. The focus of the effort is theminimization of weight while satisfying all of the detailedstress safety margins, flutter margins, and fatigue liferequirements. This involves modifying active controls toalleviate loads and includes filtering control laws toeliminate unfavorable interactions resulting in flutter.Design considerations include, detailed part geometry,materials, external loads, elastic-to-rigid ratios, stiffness,mass, and flight control laws.

A single high-fidelity air vehicle FEM is a keyrequirement for the success of this effort. This FEM isused for stress, loads, flutter, allowables, internal loads,and checking of aeroservoelastic affects. This FEM is themain feature in a tri-company coordination effort, and itpayed for itself many times over in providing astraightforward process and in facilitating communication.The only restriction on the FEM is that it not overload theConvex 10 terabyte storage capacity. The design processconsists of cycling all of the necessary analyses anddesign steps. Some of the disciplines are iterated severaltimes within the global cycle. In addition, because of

differences in discipline cycle time, several disciplines areat different stages, being 1, 2 or even 3 cycles behind thecurrent global cycle. In the time allotted, four globalcycles are carried out, however, the inconsistenciesbetween the discipline stages do not seem to affectconvergence greatly.

The Role of MDO within Aerospace Design and ProgressTowards an MDO Capability Through EuropeanCollaborationBartholomew8 presents three European MDO projects; 1)the GARTEUR regional transport aircraft structuraloptimization, 2) the EU IMT project where the A3XXtransport aircraft direct operating cost (approximation) isminimized and, 3) the ESPRIT Frontier project where aPareto front is identified for a multiple objective problem,and where trade-offs between the different objectives areidentified.

In addition to the examples, a discussion of MDO ingeneral and Europe in particular is presented. The MDOprocess of choice is loosely coupled, and multilevel. Atthe lower level, it uses a detailed design process normallyused by engineers. An integrated software system isneeded that has a flexible user interface, provides forchecking all along the way, and uses standardized productdata formats (STEP).

MDO Technology Needs in Aeroelastic Structural DesignHoenlinger et al9 present explicit answers to the questionsposed by the organizers of this session. The highlights oftheir paper are two tables, and accompanying discussions,that provide a wealth of information on past experiencewith structural sizing/optimization and expert opinions onwhat is needed in MDO. The industrial applications rangein time from 1985 to the present and cover the ACA,X-31, Ranger 2000, Stealth Demonstrator, and the “MDOAircraft” (A3XX). The history of the development of theLAGRANGE aeroelastic structural optimization softwareis sketched, ending with the decision not to extend thissystem to the controls discipline as it is thought that amore general architecture is warranted and that it is betterto include LAGRANGE itself in a more generalarchitecture (e.g., iSIGHT). The existence andapplication of a rapid parametric FEM model generatorfor high aspect ratio wings is also discussed.

Like several other contributors, the authors points at thefact that there are serious organizational aspects inintroducing MDO in an industrial environment. (“..nocoordinating position for MDO is present in typicalindustrial hierarchies.”)

A Collaborative Optimization Environment for TurbineEngine DevelopmentRohl et al10 describe the development of an MDO processfor the design a jet engine rotor disc; they show that asignificant part of the challenge to performing MDO is to

5

be able to do MDA (Multidisciplinary Design Analysis).The first order of business is feasibility (fatigue life anddistortion tolerance). The second consideration isminimum weight, both for the finished part and for thebillet (cost). The components of the process are:mechanical design, thermal cycling/loads, forgingoptimization, heat treatment optimization, machiningsimulation and life prediction. The mechanical design tomeet the mission requires material properties, residualstress, and life prediction which are not known ahead oftime and are determined in the forging, heat treatment, andmachining simulations and suboptimizations, and the lifeprediction analysis. Forging is a minimum billet weightoptimization (using DEFORM) with constraints on theforging requirements. Heat treatment has conflictingobjectives for its suboptimization; i.e. maximum materialproperties, with minimum residual stresses and requiresvery high fidelity meshes. The authors point to the factthat the complex analysis capability resulting from theintegration of the individual simulations required is not assmooth as desired, and that large step size finitedifferences are required to obtain robust derivatives.

The MDO process was initially implemented in iSIGHTand both the CSSO and CO decompositions, were tried.These proved impractical due to the nature of the problemand the requirement for high fidelity. A modifiedsequential process is suggested but this work is still inprogress. Currently most of the emphasis is on thedisciplinary tools and automation of these high-fidelitysimulations. Specifically, two “tool kits,” the ProductModeling Kit (PMTK), and the Discrete AnalysisModeling Kit (DMTK) are being developed underDARPA contract.

Multidiscipline Design as Applied to SpaceLillie et al11 describes a systems engineering process forthe feasible and affordable design of the NGST (NextGeneration Space Telescope). The final product is abaseline design and the associated technologydevelopment necessary to implement the design. FiveIPD Teams are used to design the telescope; 1) OpticalTelescope Assembly (telescope structure), 2) ScienceModule (instruments), 3) Spacecraft Systems (power,propulsion, vibration and thermal control), 4) OperationsSystems (ground systems, data handling, operations), and5) Systems Engineering (Integration of systems andrequirements). Requirements related to targets,observations, aperture, quality, imaging spectral bands,stare time, agility, pointing stability, imaging field ofview, coverage, field of regard, lifetime, and cost makethis a very challenging design for feasibility. The processis one of multidisciplinary integration. An example is therequirement for minimum contamination of the telescopeoptics from the propulsion system.

The design is presented as a series of mostly discretedecisions, few of the variables used are continuous.

Usually a short list of available options exists for eachchoice. The importance of each of the requirements isclassified as; 1) required, or 2) highly desired, or 3)desired, and 4) goal. The design decision is made basedon the ability of the option to meet the requirement, theimportance of the requirement, and the performanceimpact of the choice. Currently this TRW team isassembling a full structural, thermal, opticalmultidisciplinary simulation (not reported in the paper).Their objective is to “optimize” the design using thesimulation. The issues with the simulation involveinterfacing various systems together, converting andtransmitting data among the three disciplines anddeveloping a common model.

Multidisciplinary Design Practices from the F-16 AgileFalconLove12 describes the process for determining the “best”design for a more “agile” F-16 aircraft at reasonableincremental cost. “Best” is not formally defined butinvolves ranking of discrete designs on the basis ofmaneuverability, controllability, weight, and producibility.The design is carried out in two steps, and the wingplanform shape is selected in the first step, its twist andcamber distributions in the second. A baseline wasavailable for the new “agile” design and variations aredeveloped about this baseline. Specifically, wing span,sweep, and area variations are analyzed and tested using 6discrete design points. No one configuration providedsuperior performance. A new baseline was derived fromthe aerodynamic, weight, and system interface studiesperformed using a qualitative process. Further designrefinements/studies are performed about the new baselinewhich consider variations in basic camber and twistdistributions of the wing to enhance agility. Aeroelastictailoring is used to optimize the new baseline, as well as awash-in and a wash-out wings (i.e., wings that twist up ordown, with increased aerodynamic loads). A rankingtable that considered maneuverability, controllability,weight, and producibility was used to select the best of thethree cases.

The author makes the point that “... the approach toachieve integration would probably be the same today(1998) as in 1988-89. The differences in the overallprocess would be in the tool selection... and the amount ofdata generated.”

A Description of the F/A-18 E/F Design and DesignProcessYoung et al13 describe the re-design process of the F-18 tomeet multiple missions not originally intended for theoriginal aircraft. Some of the increased requirementsinvolved: carrier suitability (landing weight), strikemission (payload), fighter mission (range), increasedsurvivability, maneuverability, growth potential, andothers. The objective is to reach a feasible design atacceptable cost and a “Stop-Light” (red, yellow, green)

6

process was used to grade each requirement. Sevendiscrete configurations are analyzed and graded by an IPDTeam. Only two configurations had no red stop signs. Ofthese two, one had slightly better grading and lower costthan the other and this one is selected. Some of the designchanges include: a 25% wing area increase, a snag in theleading edge, an enlarged leading edge extension (LEX), athickness-to-chord ratio increase, enlarged inlets, and anadded third weapons carrying station. The authors put alot of emphasis on the building of an aerodynamicdatabase made of a combination of CFD results and windtunnel data which will prove critical to good aeroelasticoptimization.This paper also describes the IPD Team function andprocess, the Cost/Schedule Control System (C/SCS)accounting system, a Technical PerformanceMeasurements (TPM) tracking system, and finally asection answering questions on, barriers to MDO andfuture needs.

3. Industrial Challenges and Issues

Selection of CategoriesMany of the issues, needs, conclusions and salient pointsgleaned from the 10 papers are summarized, categorizedand discussed here. The categories used here wereinspired by a classification of “MDO ConceptualElements” (MDO Taxonomy) given by Sobieski3 butmodified to reflect industrial needs, prospectives, andpriorities. One such modification is the addition of ageneral classification dealing with “Management andCultural Implementation” issues in the industrialenvironment. The industrial influence on Sobieski’sTaxonomy was derived, in part, by a series of hypotheticalquestions (Figure 3-1) that an industrial designer mightask before designing an MDO system to solve hisparticular problem. These questions range from “What ismy design objective?” to “How do I make it happen at myplant?”

Figure 3-1: Concerns of an Industrial Designer Prior to Setting Up an MDO Process

The final categories or MDO elements selected for thispaper are shown in Figure 3-2. There are four generalcategories which include design formulation issues(prompted by questions 1, 4, 5, and 7 in Figure 3-1),analysis capabilities (related to questions 2, 3, 6, and 11),

information management (see questions 8, 9, 10, and 12)and management and culture constraints (question 13).Each general category contains several sub-categories ofits own.

Questions Asked by an Industrial Designer

1) What are my design objectives and critical constraints2) What are my disciplinary analysis capabilities/limitations/Automation level3) How do I get critical high fidelity elements into my design in an efficient manner?4) What design process steps are needed to meet my design objective most efficiently and to know that I have reached my objectives and satisfied my constraints? 5) What MDO or design formulation do I need or what formulations are available to me?6) What kind of approximation analyses are required?7) How do I overcome Optimization problems (scaling, smoothness, robustness, effic.)?8) How do I feed data among disciplinary analyses and the MDO process?9) How do I overcome computing and data handling issues10) What is the easiest way to visualize my design space?11) How robust is my design and how do I check it? 12) Are there commercial systems that can effectively help me?13) How do I make it all happen at my plant?

7

Design Formulations &Solutions

• Design Problem Objectives• Design Problem Decomposition,Organization• Optimization Procedures and Issues

Analysis Capabilities & Approximations

• Breadth vs.. Depth Requirements• Effective Incl. of High Fidelity Analyses/Test• Approximation & Correction Processes• Parametric Geometric Modeling• Analysis and Sensitivity Capability

MDO Elements

Information Management & Processing

• MDO Framework and Architecture• Data Bases and Data Flow & Standards• Computing Requirements• Design Space Visualization

Management & Cultural Implementation

• Organizational Structure• MDO Operation in IPD Teams• Acceptance, Validation,Cost &, Benefits• Training

Figure 3-2: MDO Elements Grouped by Categories (MDO Taxonomy)

Each of the salient points from the 10 papers have beensummarized into short one-line sentences. An initial isplaced at the end of each of these sentences to identify theauthor from which they came. These points (one-liners)were sorted and placed in the categories given in Figure3-2. The results of this sorting is given in Appendix I. Alegend at the beginning of the Appendix gives the keyrelating the initials to the paper authors.

Discussion of CategoriesA general discussion of the challenges and issuesassociated with each of the categories (shown in Figure 3-2) is presented here. The basis of these discussions arethe sorted one-line salient points presented in Appendix I.The content of the discussion is mostly taken mostly fromthe pertinent items listed in each category, howeveradditional interpretations, generalizations and theexperience of the current author are also sometimesincluded.

Design Problem ObjectivesThe range of industrial design objectives sampled in the10 papers is illustrated in Figure 3-3. The scale is animaginary continuum of problem statements that rangesfrom making a design satisfy all requirements (i.e.,feasible), to finding the optimum design for severalobjective functions. Intermediate possibilities areimproving a feasible design and finding a single-objective

optimum. Most of the papers included in this series arelumped in the “Feasible” and “Optimal” categories.However, even though many of the design problems arecast as optimization problems it is probably true that thereal goal of the effort is an improved design. Forexample, in the helicopter rotor design problem discussedby Tarzanin et al

6, the optimized design was tested tosee if it presented an improvement over earlier designs,not to see if the improvement matched the predictedoptimum. Young, Anderson, and Yurkovitch13 show thatanother implicit goal of the effort is probably designrobustness since point designs can be sensitive tounknown problem parameters and are not always ofpractical value. Bartholomew8 discusses a pareto-optimization approach; a parameterized series ofoptimizations carried out to effect trade-offs betweendifferent conflicting objectives.

The authors describe a situation where, generally, theproblem statement is not known a priori. Rather, it isdefined in an interactive fashion in the course of thedesign exercise. As an initial statement is adopted, aparticular design emerges that may be lacking in one wayor another. At that point, the problem statement ismodified to address the shortcomings of the initial design.This process is continued, until a satisfactory design isobtained.

8

Figure 3-3: Range of Design Objectives

Design Problem Decomposition and OrganizationThe consensus appears to be that loosely coupled systemsthat can work with legacy analysis codes hold the mostpotential for future advances (see, for example Bennett etal

5, Bartholomew8, Hoenlinger et al9). Such a system

also affords flexibility and can be reconfigured as theproblem formulation evolves, as indicated in the previoussection. This implies a need for an off- the-shelf modularsoftware framework that facilitates the integration of thedifferent analysis codes. In some instances, multilevelprocesses are used, rather than “all-at-once” systems forseveral applications since it seems inefficient to bringevery disciplinary design variable and constraint up to theglobal level. This is commonly the case in structuraloptimization with detailed structural element models,where both local and global constraints are employed andwhere local variables are optimized. (See Bennett et al

5,

Rohl et al10 for examples). One of the advantages of

decomposed procedures is that they can be used for multisite operations (Hoenlinger et al

9).

Wakayama and Kroo4 and Tarzanin et al 6 pointed out,however, that currently some of the more successfulapproaches use close-coupled, all-at-once procedures,however, their success depends, in part, on the fact thatautomated, fast-running analysis codes (intermediatefidelity level) are used.

As indicated by Rohl et al10, and also Hoenlinger et al

9,industry seems to feel that the more sophisticated MDOdecomposition processes (e.g., CO, CSSO) are not yetfully proven or sufficiently matured. Rohl et al

10 indicatethat, in some cases these approaches are not even suitedfor some of the applications to which they were applied.In other cases, as indicated by Bennett et al

5, it may bethat the more complicated approaches are not easy tounderstand or follow and thus simpler processes areselected. Additionally, it seems obvious from the various

inputs that decomposition process flexibility is an absoluterequirement and that the optimization process must bereconfigurable and tailorable to the specific problemencountered and to possible variations that might emergein the problem formulation.

Optimization Procedures and IssuesThe contributed papers state few requirements on thecomponent optimization capabilities, althoughBartholomew8 points to the lack of robustness of off-the-shelf optimization software. In general, industrypractitioners need more experience in the art and scienceof applying optimization algorithms and interpreting theirresults. The typical engineering optimization problem isnon-linear and non-convex, therefore, a great amount ofexperience is needed to reliably operate the optimizationalgorithms. Help in coping with lack of smoothness orscaling requirements, in overcoming slow convergenceand local minima problems could significantly reduce theturnaround of typical optimization exercises. Wakayamaand Kroo 4 point at the need for more robust and efficientindustrial-strength, commercial-grade software to solvelarge scale problems.

Hybrid schemes that can handle discrete and continuousdesign variables can prove also be very helpful in anindustrial environment according to Bartholomew8. Also,

Rohl et al 10 point out that interdigitation, a procedure bywhich a combination of different algorithms is used to getto the global optimum of the problem. Tarzanin et al6encountered local minima and suggested various processto avoid them including a hybrid evolutionary processwith NPSOL.Breadth and Depth RequirementsAs detailed by Wakayama and Kroo4, industrial designprocesses must possess sufficient breadth. Specifically allof the critical constraints must be considered, otherwisethe design will not be practical or feasible. This implies,among other things, that multiple flight conditions must be

SpaceTelescope

ImprovedFeasible Optimal Pareto

F/A-18 E/F

F-16Agile Falcon

GM Auto RotocraftGE Engine

Lrg A/C BWB

f-22

Frontier

9

verified, whether for demonstrating performance, flyingqualities or for verifying stress/stability constraints. It hasalso been pointed out that all of the critical physicalmechanisms should be included, to take advantage of allthe available design opportunities.

Some authors contend that the highest fidelity models areneeded throughout the optimization process, othersindicate that various level of accuracy are adequate. TheMDO process itself can be used to help determine thefidelity levels required by performing accuracy sensitivitystudies on the various critical physical mechanisms in thevarious disciplines.

Effective Inclusion of High Fidelity Analyses/TestBartholomew8 has a defined set of analysis fidelity levelsas follows:- Level 1: empirical equations,- Level 2: intermediate level models (e.g., beam theory,

panel aero, etc.)- Level 3: state-of-the-art, high fidelity models(e.g., CFD,

FEA)and has observed that industry MDO is moving towardLevel 3 since disciplinary experts usually insist on usingthe latest, best, and highest fidelity information. If theycannot then they do not feel comfortable with the results.(They may even be uncomfortable with the best analyses/tests results since they are never fully assured that the realworld is being faithfully simulated.)

Therefore, effective inclusion of high fidelity data into thedesign optimization process is necessary, especially fordesigns at the preliminary and detailed design levels. Thismay be the most formidable challenge facing industryMDO users and methods developers. Such high fidelityprocesses are usually neither automated nor robust andmany times require hours (even days) of computer time.Allowing an optimizer the opportunity to call suchroutines as often as it needs to, even if these routines werefully automated, is impractical, so various approximationmethods need to be incorporated (Wakayama and Kroo4,and Tarzanin et. al. 6 ).

Approximation and Correction ProcessesOne class of approximations methods include genericlocal approximations like Taylor series or variations aswell as generic global approximations like responsesurfaces and neural nets, etc. These provide smooth,simple, explicit analytical expressions that can begenerated automatically and that can be called by theoptimizer as many times as needed without unduecomputational burden. Alternately, these approximationscan be created concurrently off-line by disciplinaryexperts who can be responsible for their validity. Thechallenge in producing these approximations is the trade-off between the amount of data needed to create them, andthe control of their accuracy in the design variable space.

For approximations in this class, the number of designvariables that are strongly coupled still remains small,otherwise, the curse of dimensionality sets in and theapproximations become unduly expensive. Also, it iscritical to augment them locally to increase their fidelity incertain critical design regions.

Another approximation class uses Level 1 or 2 fidelitydisciplinary codes that have been corrected using highfidelity codes, or experimental results (see, for exampleChang et al, 14, Baker et al15, 16). In essence, the lowerfidelity codes can be used as a “smart”interpolator/extrapolator. The challenge, as underlined byWakayama and Kroo4, is to make sure that all of thecritical physical mechanisms are represented to somedegree/level so that the high fidelity code information canbe effectively utilized.

A third class of approximations that can be considered foruse in MDO are “Reduced Order” methods17. Theseprocesses extract the essence of the high fidelitynumerical results and expresses them in a relative simpleanalytic form.

Parametric Geometric ModelingBennett et al

5 and Honlinger et al9, Radovcich and

Layton7 highlight the need for a shareable commonvehicle description to facilitate communication amongdisciplines and among various companies and sites.Radovcich and Layton7 report that a single high-fidelitymodel was used for most of the detail structural sizing anddesign of the F22 and that this model paid for itself manytimes over in communication and facilitated analysis anddesign iteration. They also pointed out that sometimessmall changes in structural FEM grids can causesignificant changes in internal loads and design, thus it isimportant to have a high-fidelity model.

Automation is one of the essential requirements for MDOand many authors make the point that parametric andfeature-based models facilitate automatic model changes(See, for example Hoenlinger et al

9, Wakayama and

Kroo4, Love12). Morphing (rubberizing) is one approachat parameterization, but it does not always produce amanufacturable, or even reasonable structural layout.Hoenlinger et al

9 indicates, that, in such cases moresophisticated processes are called for which may requirefitting continuous processes to discrete layouts.

The resulting unified and parameterized geometrydescriptions must be compatible with existing CADsoftware, however, as indicated by Rohl et al 10 additionaldevelopment work is required since the parametricfeatures of CAD available now are not robust enough fortopology optimization. The work on the Technical Data

10

Modeller and Browser (TDMB) reported byBartholomew8 appears to be a response to this need.

Analysis and Sensitivity CapabilitySeveral examples of this were encountered in thecontributed papers where middle-level fidelity analysiscodes are directly interfaced with the optimizers. (See forexample Wakayama and Kroo4). This was only possiblebecause of the relatively low computational cost of theindividual simulations.Some papers made use of off-the-shelf single-disciplinehigh-fidelity optimization codes that were eitherautomated (see Tarzanin et al

6) or partially automated

(Hoelinger et al9, Bartholomew8). In each instance, the

detailed analysis is interfaced with the optimizer throughapproximations of different kinds. Several systems suchas STARS, LAGRANGE, NASTRAN Sol. 200, andothers are available to automate and facilitate structuralsizing but much additional work is yet to be done to fullyintegrate local panel design (as-built weight, compositemanufacturability, cost, and mass balancing). Automated,robust, and efficient CFD analysis, optimization design isalso needed but is still in the future.

Industry prefers, in general, to utilize off-the-shelf (OTS)detailed analysis capability when ever possible. Rohl et al10 give a good example of such an application to thedesign of jet engines which is based on UG, PRO-E, I-DEAS, PATRAN, ANSYS, ABAQUS, NASTRAN, andDEFORM.

It must be emphasized that the drive towards inclusion ofall the disciplines relevant to a complete design problemstatement still requires major developments in differentdisciplines. While these developments are mostly outsideof the field of MDO itself they deserve reference here.

Satisfactory structural optimization requires detailedaerodynamic loads. A large number of critical flightconditions occur in the transonic regime or at high-angleof attack. While this information is now derived fromwind tunnel experiments, significant reduction in designcycle can be achieved by deriving it computationally.Young et al

13 detailed the need for a comprehensiveaerodynamic database, and, together with Hoenlinger etal

9 highlight the need for a methodology for nonlinearaerodynamic loads calculation and identification ofcritical load cases.

The next step in disciplinary integration for MDO is tobring controls into a full aeroservoelastic formulation ofthe design problem. Methods are required that enablederiving controls metric and constraints early in the designprocess, at a time when very little is known of the aircraftconfiguration (see, for example Hoenlinger et al

9,

Radovcich and Layton7, Love12).

Finally, central to a successful application of MDO aredetailed first-principle-based cost models that includedevelopment, manufacturing, acquisition, operations anddisposal. (See Love12, for example.) Other Analysiscapability needed are:- nonlinear aerodynamic loads (Hoenlinger et al

9)- wing load alleviation and aeroservoelastic integ. into

str. sizing opt. (Ref.7, 12).- intermed. level fidelity codes (which incl. critical

physical mechanisms) (Ref. 4)- robust reduced order processes

MDO Frameworks & ArchitectureCommercial off-the-shelf (OTS) software for MDOframeworks are desired by industry and some areavailable (iSIGHT, SYSOPT, others) (Rohl et al 10,Honlinger et al9 ). Some have been tried but the degree ofsuccess is uncertain. In addition commercial distributedcomputing does not seem to be robust (Bartholomew8).Industry wants demonstrated, validated MDO software(Honlinger et al 9 ) that is easy to use.

Databases, Data Flow & StandardsIndustry considers database capability to be veryimportant (Young et al13). It is a repository for current(and past) design data (and the ground rules for generatingthem) and as such should facilitate communication andreduce cycle time for interdisciplinary data exchange(Bennett et al 5). Such a database must be industrialstrength (able to handle huge amounts of data rapidly andshould be able to sustain multi site, heterogeneousoperation and be user friendly (Radovcich and Layton7).A standard set of formats and ground rules for the data(STEP = Standard for The Exchange of Product data)(Bartholomew8) will also greatly increase the speed ofcommunication, reduce errors and greatly reduce cycletime. European experience includes projects supported bya “Software Infrastructure Group” and development (Task8) of database and related tools as follows;- software version management- data definition- database technology- process definition- process execution on distributed networks- data visualization

Computing RequirementsIn the case of the F-22 the size of the structural FEM andresulting database was determined by the computermemory (10 terabites) (Radovcich and Layton7) requiredto house the database. CFD analysis and design alsoposes challenges to computing power. For instance ittakes on the order of 10 hours for analysis and about 10-20 hours per design variable for aerodynamic design usingthe C-90 supercomputer. Thus, if 20 d.v. are used for adesign problem then the design would take approximately

11

300 computer hours. NASTRAN Solution 200 can easilyrun several days on a high end work station. Distributedcomputing is probably a necessity for the future to garnerenough computing power to perform some of the requiredanalysis functions and to drive multi site operations.

Design Space VisualizationConfiguration designers can sometimes be more interestedin the design space than the optimum design point. Howflat or narrow is the design space near the optimum? Howmuch is lost if an adjacent point is chosen because theoptimum point is undesirable? Is the design spaceprecipitous and overly sensitive to errors/noise in thedisciplinary modules? How did the optimizer reach theoptimum design point? The end result is that it isimportant to the designer to have user friendly processesfor displaying the design space and interpreting the resultsof the optimization (Tarzanin and Young6, Honlinger etal 9).

Organizational StructureIndustry is organized along disciplinary lines where eachtechnology group is responsible for maintaining technicalexcellence, and ensuring that the data generated in thatdiscipline is correct. It is absolutely necessary that thisdisciplinary control be maintained in any MDO processthat is developed. One of these disciplines ortechnologies is contained in the Advanced/ConceptualDesign group. This group is responsible for configurationdesign and global integration methods and applications.Usually, very approximate analysis methods are used thereand so high-fidelity coordination with the variousdisciplines is minimal. However, for future MDO designsuch is not the case. If the Advanced/Conceptual Designgroup is to assume responsibility for MDO at the globallevel then it will have to change tactics somewhat andprovide an integrating function (instead of providing theirown simple disciplinary analyses) while allowing thevarious disciplines to maintain control of the local leveldesign/sub-optimization and data recovery (such asinternal FEM loads). In the papers sampled it is theperception that currently no one is in charge of MDO andthat an improved company organization would benefit theuse of MDO (Honlinger et al9). Ensuring buy-in of thedisciplinary experts to the MDO system may be difficulthowever (Bennett et al5).

MDO Operation within IPD TeamsThe IPD Team is an essential element in industrial design(References 4, 11, 10, 13). When MDO is used in the designthe IPD team is not replaced but interacts with the processto learn about the design, assess the ground rules,add/replace constraints, furnish guidance in areas notmodeled and generally keep the optimization on track(Wakayama and Kroo4). An example of this was thecomposite wing design of Reference 19. MDO is a tool of

the IPD Team which is used to assist in selecting andimplementing the final design.

Acceptance, Validation, Cost, & BenefitsA lack of understanding of MDO and what it meansorganizationally is an obstacle to industrial acceptanceboth by managers and by disciplinary experts (H). Also,Industry can have difficulty in determining both thebenefits and development/deployment costs of MDO(Honlinger et al 9). How does a manager assess if there isa net benefit for developing and using an MDO process?Lack of validated results and quantified benefits in thepractical industrial environment (not just mathematicalprocess validation) is a big obstacle to its acceptance(References4, 9). Specifically, the cost/benefit overconventional design processes is needed. An example of atest that proved that there were benefits of an optimizeddesign is given by (Tarzanin and Young6), however, acomparison of the predicted versus actual benefits was notgiven.

TrainingOnly recently have universities offered MDO orientedtraining and so, for the most part, only those in industrythat are newly trained are intimately familiar with theformalisms associated with optimization. The rest of theengineering force are, to one degree or other, are havingdifficulty (Bennett et al 5). This lack of familiarity is anobstacle to the use of MDO in industry.

4. Development Needs for Future Industry MDO

MDO Needs by CategoryMDO development needs in industry, asinferred/interpreted from the 10 papers and the experienceof the current authors, are presented here. For consistencythese needs are categorized in the same fashion as thesalient points of Section 3, i.e., the categories shown in theMDO Taxonomy given in Figure 3-2 are used.

Design Problem Objectives (Needs)Each industrial problem is different and so the biggestneed is to have MDO frameworks that are flexible enoughto accept whatever objective function is needed. As far asobjective function formulation is concerned, research hasbeen, and is being done to provide ways to formulatemultiple, difficult or nebulous objective functions. ParetoFront techniques help define the biggest bang-for-buck sothat, for instance, the DoD can decide on how muchperformance it can afford. Also, advances in simplifiedcost related objective functions have been made (Giesingand Wakayama et al 18, Bartholomew8) and this type ofwork should continue.

Design Problem Decomposition and Organization(Needs)

12

Most high fidelity process (e.g., CFD, FEM) are currentlynot automated, robust, nor fast enough to be directlycalled by an optimizer. Thus, a very big challenge is tosomehow end up with a design that reflects this highfidelity but does not call it directly by the optimizer.Approximation approaches to this problem are mentionedin that particular category (discussed below). However, inthis section the question is; are there decompositionapproaches that could accomplish this task? For instance,can approaches be developed that converge to a highfidelity result that only require periodic high fidelityupdates to the design process? Currently, analysismethods accommodate and are tailored/adapted to theneeds of the optimizer (smoothness automation, etc.),however, this needs to be reversed. Development work ondecomposition processes that accommodate and aretailored to the needs/deficiencies of the analysis methods(noise, lack of automation, very large computing time,etc.) are needed since analysis methods are the criticallimiting factors in industrial MDO processes.

Optimization Procedure and Issue (Needs)Improvements in optimization techniques are continuouslybeing made and this must continue since industrialstrength, robust, and efficient modules are needed.Industrial strength implies that large problems can behandled (thousands of design variables and constraints).Robust techniques are needed that converge under a widevariety of conditions. Efficient modules are needed tokeep the computing time to a reasonable level. Userfriendly optimization techniques that are insensitive tonoise or are “self smoothing” and that provide their ownscaling (self scaling) are also needed. Finally robustprocesses and procedures for escaping local minima andfinding the global optimum are needed.

Breadth and Depth Requirement (Needs)All critical physical mechanisms and critical constraintsmust be accounted for in an accurate manner for realisticdesign. Breadth indicates the number of disciplinesinvolved (mechanisms and constraints) and depth theaccuracy/fidelity. Identification of all critical constraintsrequires experience in the design of the particular vehicleor artifact involved. Identification of the criticalmechanisms is more subtle and difficult and requiresunderstanding of the underlying physics of the variousdisciplines. Experience with high fidelity codes (e.g.,CFD) does not necessarily mean that the variousmechanisms are understood. Techniques that use theMDO process itself to determine which are the criticalconstraints and mechanisms would be very helpful.

Effective Inclusion of High Fidelity Ana./Test (Needs)Two approaches for including high fidelity analyses inMDO have already been discussed, (using adecomposition approach and by using an approximationand correction approach). This section discusses whatshould be done to the high fidelity methods themselves for

direct use in MDO. Currently, many high fidelityprocesses (such as Navier Stokes, FEM global-localstructural sizing) can not be used directly in MDObecause they are not automated, robust, nor fast enough tobe included. This presents formidable challenges in mostdisciplinary areas and advancements of the state-of-the-artare required. As each high fidelity technology areamatures it becomes more robust and efficient and moresubject to automation. Even after maturity however,computing requirements will still be a problem for highfidelity methods.

Approximations and Corrections (Needs)This may be the single most important need for industrialMDO. As mentioned above many analysis codes (highfidelity or otherwise) can not be put directly into the MDOprocess and thus approximations and corrections must beused. Response surfaces, neural networks, Taylor seriesand Taguchi techniques are in current use but robust,efficient, and user-friendly software packages are needed.Procedures that use high fidelity analysis or test data tocorrect lower fidelity methods are also currently underdevelopment but improved techniques are needed in alldisciplines. Simple mathematical (non physical)techniques of fudging low fidelity analysis methods needto be upgraded to those that isolate and correct eachseparate physical mechanism separately. Wakayama’s(Reference 4 ) use of calibrated simple 3-D source flowterms to simulate transonic 3-D effects is an example.Baker et al 15,16 have also develoed advanced correctionprocedures for steady and unsteady aerodynamics andloads. An even more sophisticated approximationprocedure is produced using reduced order or parameteridentification methods and models (Ref. Baker17). Simpleexamples are state space representations of dynamicaeroelastic models and associated rational functionapproximations for the unsteady aerodynamics. Othermore sophisticated procedures require development,refinement, and extension. Finally, intermediate levelfidelity methods can, themselves, be considered anapproximation method whose approximating equations arebased on physical mechanisms. If all of the criticalphysical mechanisms are present and a process, includinghigh fidelity adjustments/corrections to each one, are inplace then the intermediate fidelity level methods might bethought of as a physics-based interpolation/extrapolationmedium for high fidelity codes. This is highly desirablesince physics and not mathematics forms the basis of theapproximation equations and not just mathematical fittingfunctions. Figure 4-1 presents examples of theapproximation and correction procedures outlined here.

Parametric Geometric Modeling (Needs)MDO processes require parametric models and automatedmodeling techniques. Tool kits such as the one beinggenerated under DARPA sponsorship (PMTK)(Reference10) will be helpful. Parametric models need to

13


GenericApproximations -Responce Surf. -Neural Nets -Taylor Series

IntermediateFidelilty Analysis - Phy. Mech. A - Phy. Mech. B - Phy. Mech. C


Correction Process


Reduced OrderRepresentations

Optimizer

Optimizer

Optimizer

A BC D[ ]{x}{X}=

•Example

Figure: 4-1 Three Approximation and/or Correction Processes

maintain accuracy and realism as design variables arechanged. Thus, for instance, morphing techniques maynot be adequate for structural layouts since best industrialpractices usually require changing the topology as designvariables are changed. Also, straight structural membersthat become curved during morphing will probably not beacceptable. Robust, automated, and accurate non-parametric models are also required in industry as areinterdisciplinary grid/mesh mapping techniques. Wellproven software modules for these are needed.

Analysis and Sensitivity Capability (Needs)Automation is one of the biggest needs with respect todisciplinary analysis methods. An automated analysis willallow the possibility of direct integration into an MDOprocess and will facilitate the generation and updating ofapproximations (response surfaces, etc.). Anotherchallenge is the quantification of manufacturing andmaintenance cost and constraint requirements into usablemodels. Cost is usually based on weight even though partcount and complexity are much more important for costthan weight. The development and quantification of suchmodels is a definite need in industry. Robust, efficientnonlinear loads analysis methods are also needed as wellas well developed aeroservoelastic techniques. A currentindustry trend is to use well proven, over the counter(OTC) analysis modules and thus development of these isneeded in all disciplines. Sensitivity analysis methods didnot seem to be high on the list of required technologies,however, such methods are desirable to increaseefficiency both for direct inclusion into optimizationprocess or indirectly through the generation of responsesurfaces and other approximations. Robust CFD (Navier-Stokes) codes both rigid and aeroelastic are needed. Also,an efficient robust global-local structural sizing process isneeded that accounts for all of the major structural effects;

stress, buckling, aeroelastic loads, local panel design,durability and damage tolerance, flutter, and reversal.

MDO Frameworks and Architecture (Needs)A mature, efficient, flexible, robust, industrial strengthcommercial MDO framework is desired by industry.Preferably, a loosely coupled reconfigurable system thatcan use legacy and other commercial software is best.The architecture should be flexible enough to accept awide variety of MDO problems.

Databases, Data Flow & Standards (Needs)Data standards for format, access, and monitoring areneeded to facilitate analysis module integration and datatransfer. An industrial strength, efficient, and easy to usecommercial database system for multi-site, multi-platformoperation is also needed. Possibly an Internet basedsystem could be the system of the future if and when it isable to handle large engineering data sets in an efficientmanner.

Computing Requirement (Needs)Current CFD and FEM sizing (e.g., NASTRAN Solution200) require hours and even days of computer time onhigh end work stations. This is a formidable barrier totheir use in optimization processes. Improving computingpower with the use of massively parallelized machineswill improve this situation especially if analysis codes canbe re-programmed to take advantage of them. Specifically,new subroutines and algorithms (e.g. matrix operations,eiganvalue analysis etc.) designed to take advantage ofmultiple processors are needed. In this case astraightforward process of re-programming existinganalysis codes would be desirable. In this regard theHPCCP (High Performance Computing andCommunication Program is dedicated to demonstrating

14

“teraflops computing” since it is assumed that this is thewave of the future. If clusters of work stations are usedinstead then efficient and robust distributed computingcontroller systems are needed. If the controller can spanmultiple sites then this will potentially open up a largeresource for computing. This system, however, must bevery versatile since work stations are usually availableonly on an intermittent basis and scheduling andcoordinating would be a very challenging task.

Design Space Visualization (Needs)Commercial MDO frameworks must provide easy to useoptimization and design space visualization/interpretingsince designers are sometimes more interested in the spacearound an optimum than the optimum itself. The largestchallenge in this regard are techniques for visualizing amultidimensional design space. Since it is impossible tovisualize anything beyond three-dimensions creative waysof interpreting or depicting the design space need to beinvented. These depictions could require a lot morecomputing operations than the optimization process itself.

Organizational Structure (Needs)Industry itself needs to adjust their organizations tofacilitate MDO. Disciplinary groups would still developand maintain technical excellence and be responsible forthe accuracy and integrity of design data in anautonomous fashion. The responsibility for interfacingand coordinating all of the disciplines into an MDOprocess will have to be assigned to an MDO group. All ofthe disciplines will work together with the MDO Group asa team to decide on the interface processes. It makessense that the MDO Group also is responsible for globalconfiguration optimization and this job is currently beingdone by the Advanced Design Group. Does it then makesense to broaden the role of the Advanced Design Groupto assume the responsibility of the MDO function?

MDO Operation in IPD Teams (Needs)Industry itself needs to address this issue since IPD Teamsare now a permanent part of the industrial landscape andare an ideal place to direct the MDO efforts. The MDOGroup (or Advanced Design Group) may conduct theconfiguration optimum operations and perform trade

studies that may not fit the optimization process, however,the IPD Team will direct this effort at a higher level. TheIPD Team must get used to using MDO as a tool that theycan direct. Design philosophy, ground rules for design,critical constraint selection and definition, restraints onthe design, trade studies, etc. will all be directed by thedisciplinary, tooling, manufacturing, maintenance, andcost experts that comprise the ITD Team.

Acceptance, Validation, Cost & Benefits (Needs)The major need in this category is to produce a series offull industrial validation cases. These validations must bepractical industrial strength cases and preferably done onactual vehicles. A firm validation based on test ispreferred where the additional benefits of MDO, over andabove current design practices, are quantified andcompared to the additional effort/cost of MDO.

Training (Needs)Industry is not used to the formalisms and use ofoptimization and MDO and thus training materials andcourses that are meaningful to industry are needed. Also,new university graduates that are already properly trainedare also needed.

Satisfying MDO Development Needs

The needs outlined in this section impact every sector ofthe MDO technology development community includinguniversities, government labs, commercial softwarecompanies, and industry itself. Universities andgovernment labs can help advance the state-of-the-art fordisciplinary and MDO technology. Industry, canefficiently transfer this technology into practical use inindustrial design. Commercial software companies canprovide off-the-shelf industrial strength capability to set-up and execute major multi-site design problems. Theresources required for this development are very large andwill have to come from multiple sources with maximumleveraging. A team approach is needed that coordinatesplans and resources to ensure long range success. Figure4-1 presents an illustration of this cooperative thrust.

15

Figure 4-2: Teaming of MDO and Disciplinary Technology Development Community

5. Conclusions

A series of 10 invited design papers has been reviewedwith the purpose of providing the MDO technologydevelopment community with a distilled view of industryapplications, challenges, and needs. A wide variety ofindustries (airframe, automobile, rotorcraft, jet engine,space), and design problems (feasible design, tradestudies, structural sizing, sub-optimization, dynamicresponse minimization, and full configuration MDO) werecontained in the papers reviewed. The process ofsummarizing the papers and presenting the final resultswas as follows. First, a brief synopsis of each paper waspresented to give an overview of the applicationsreviewed. Second, the challenges and salient points fromeach paper were delineated into one-line sentences whichwere then sorted into logical categories for variouselements of MDO (Appendix I). These logical categorieswere based on an extension/revision of an existingtaxonomy (classification of MDO elements) by Sobieski3.Thirs, a general summary of each category was thenwritten which was based on the salient points contained inthat category. Finally, a summary of the MDOdevelopment needs for industry was given after distillingthem from all of the categorized data. Even though thesample of papers was limited it is felt that a very goodrepresentation and cross-section of industrial applications,challenges and needs has been given and that theconclusions of the data contained here will be helpful tothe MDO technology development community forprioritizing future MDO development. The technologydevelopment needs are wide ranging and will require thecooperative involvement of universities, government labs.,

industry, and commercial software developers to answerthese needs.

1 American Institure for Aeronautics and Astronautics Inc.(AIAA), “Current State-of-the-Art in MultidisciplinaryDesign Optimization,” prepared by the MDO TechnicalCommittee, Jan 1991, AIAA, Reston, VA. (see also:http://endo.sandia.gov/AIAA_MDOTC/sponsored/aiaa_paper.html)2 Sobieszczanski-Sobieski, J. Haftka, R.T.,“Multidisciplinary Aerospace Design Optimization,Survey of Recent Developments,” StructuralOptimization, Vol. 14, No. 1, Aug. 1997.3 Sobieszczanski-Sobieski, J. “Multidisciplinary DesignOptimization: An Emerging New Engineering Discipline,”in Advances in Structural Mechanics, J. Herkovitch, Ed.Kluwer Academic Publishers, Doordrecht, pp. 488-496,1995.4Wakayama, S., and Kroo, I, “The Challenge and Promiseof Blended-Wing-Body Optimization,” AIAA Paper 98-4736, presented at the 7th AIAA/USAF/NASA/ISSMOSymposium on Multidisciplinary Analysis andOptimization, St. Louis, MO, Sep 98.5Bennett, J., Fenyes, P., Haering, W., and Neal, M.,“Issues in Industrial Multidisciplinary Optimization,”AIAA Paper 98-4727, presented at the 7thAIAA/USAF/NASA/ISSMO Symposium onMultidisciplinary Analysis and Optimization, St. Louis,MO, Sep 98.6Tarzanin, F., and Young, D.K., “Boeing RotorcraftExperience with Rotor Design and Optimization,” AIAA

• Industry •Government Labs.• University •Commercial Software

GovernmentContracts &R&T Base

IndustryIR&D

University

Commer. SoftwareInvest.

Financial Resources

IndustrialStrength

M D O

Technical Challenges and Needs

Financial Leveraging

TechnologyDevelopment& V erification

TechnologyT ransfer &Applications

TechnologyCommercialization

T E A M• University• Gov. Labs• Com. Soft.• Industry

Technology Development Community

16

Paper 98-4733, presented at the 7th AIAA/USAF/NASA/ISSMO Symposium on MultidisciplinaryAnalysis and Optimization, St. Louis, MO, Sep 98.7Radovcich, N., and Layton, D., “The F-22 StructuralAeroelastic Design Process with MDO Examples,” AIAAPaper 98-4732, presented at the 7th AIAA/USAF/NASA/ISSMO Symposium on MultidisciplinaryAnalysis and Optimization, St. Louis, MO, Sep 98.8Bartholomew, P. “The Role of MDO within AerospaceDesign and Progress Towards an MDO CapabilityThrough European Collaboration,” AIAA Paper 98-4705,presented at the 7th AIAA/USAF/NASA/ISSMOSymposium on Multidisciplinary Analysis andOptimization, St. Louis, MO, Sep 98.9Hoenlinger, H., Krammer, J., and Stettner, M., “MDOTechnology Needs in Aeroservoelastic Structural Design,”AIAA Paper 98-4731, presented at the 7th AIAA/USAF/NASA/ISSMO Symposium on MultidisciplinaryAnalysis and Optimization, St. Louis, MO, Sep 98.10Rohl, P., He, B., and Finnigan, P., “A CollaborativeOptimization Environment for Turbine EngineDevelopment,” AIAA Paper 98-4734, presented at the7th AIAA/USAF/NASA/ISSMO Symposium onMultidisciplinary Analysis and Optimization, St. Louis,MO, Sep 98.11Lillie, C., Wehner, M., and Fitzgerald, T, R.,“Multidiscipline Design as applied to Space,” AIAAPaper 98-4703, presented at the 7th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysisand Optimization, St. Louis, MO, Sep 98.12Love, M. H., “Multidisciplinary Design Practices fromthe F-16 Agile Falcon,” AIAA Paper 98-4704, presentedat the 7th AIAA/USAF/NASA/ISSMO Symposium onMultidisciplinary Analysis and Optimization, St. Louis,MO, Sep 98.13Young, J.A., Anderson, R.D., and Yurkovitch, R.N., “ADescription of the F/A-18E/F Design and DesignProcess,“ AIAA Paper 98-4701, presented at the 7thAIAA/USAF/NASA/ISSMO Symposium on Multi-disciplinary Analysis and Optimization, St. Louis, MO,Sep 98.14Chang, K.J., Haftka, R.T., Giles, G.L., and Kao, P.J.,“Sensitivity-based scaling for approximating structuralresponse,” Vol. 30, pp. 283-287, 1993.15Baker, M. L., Yuan, K., Goggin, P. J., “Calculation ofCorrections to Linear Aerodynamic Methods for Staticand Dynamic Analysis and Design,” AIAA Paper 98-2072, presented at the 39thAIAA/ASME/ASCE/AHS/ASC Structures, StructuralDynamics and Materials Conference, Long Beach, CA,April 1998.16Baker, M. L., “CFD Based Corrections for LinearAerodynamic Methods,” presented at the 85th Meeting ofthe AGARD Structures and Materials Panel, Aalborg,Denmark, 14-15 Oct. 1997 and contained in AGARD

Report 822, entitled, “Numerical Unsteady Aerodynamicand Aeroelastic Simulation,” March 1998, pg. 8.17Baker, M. L., Mingori, D. L., Goggin, P. J.,“Approximate Subspace Iteration for ConstructingInternally Balanced Reduced Order Models of UnsteadyAerodynamic Systems,” AIAA Paper 96-1441, presentedat the 37th AIAA/ASME/ASCE/AHS/ASC Structures,Structural Dynamics and Materials Conference, Salt LakeCity UT, April 15-17, 1996.18Giesing, J. P., Wakayama, S., “A Simple Cost RelatedObjective Function for MDO of Transport Aircraft,”AIAA Paper 97-0356, presented at the 35th AerospaceSciences Meeting & Exhibit, Jan. 1997, Reno, NV.19Wakayama, S., Page, M., Liebeck, R. H.,“Multidisciplinary Optimization on an AdvancedComposite Wing”, AIAA Paper 96-4003-CP, presented atthe 6th AIAA/NASA/ISSMO Symposium onMultidisciplinary Analysis and Optimization, Bellevue,WA., Sept. 4-6, 1996

Appendix I: Salient Points from the 11 Papers Sortedby Catagories

LegendY=Young, Anderson, & YurkovichW=Wakayama & KrooL=LoveH=Honlinger, Krammer, & StettnerR=Rhol, Beichang, & FinniganB=Bennett, Fenyes, Haering, & NealBw=BartholomewRl=Radovcich/LaytonT=Tarzanin & YoungLwf=Lillie/Wehner/Fitzgerald

Design Formulations & Solutions• Design Objectives

- Minimize TOGW. W- Large aeroelastic structural sizing/optimization of

aircraft. H- MDO math. formulation and definition of “Best” is a

problem. W Y L- Accuracy of Obj. Funct. may be off by 100%. Y- Objective function hard to formulate. Y L- Most important is satisfying constraints. Y Lwf- Second most important is Robustness. Y- Better design, Nearest local opt. mostly continuous

design variables. H- Feasible, better design, mostly continuous, large no.

of d.v. and constr. and multiple obj. H- Reduced cycle time, discover critical aspects early,

model manufacturing; continuous process definitionfrom definition to product. H

- Reduced weight, maintain performance, detaileddesign. Rl

17

- DOC (Direct Operating Cost) heuristically

approximated with very simple linear combination ofweight and drags at various flight conditions. Bw

- Frontier Project uses Pareto Front techniques toidentify “biggest bang for buck.” Bw

• Design Problem Decomposition and Organization- Definition of MDO excludes flight simulation. Bw- BWB is a highly coupled problem and used an all-at-

once MDO formulation i.e., included all d.v.(mission, aero, str., etc) at the global level. W

- Industry processes are sequential; MDO requiresconcurrent processes. H

- MDO technology not yet mature enough for industrialuse. H

- MDO strategies for multisite operation is needed. H- How do you use CO to include high fidelity results.

W- Just use trade study and DOE approach. L Y- Just add more analyses to trade study approach

(MDO not needed). L Y- Advanced decomposition processes are OK if they

prove themselves. Y- Needed when high fidelity involved. W- Interdigitized optimization could prove useful. R- CO and CSSO may not be very practical. R- How do you make CO practical. W- Need a distributed optimization process. R- Going to loosely coupled processes for MDO since

source code not available for OTS MDO codes likeiSIGHT. R

- Global/Local process needed; Sub optimization ofdisciplines used. B

- Complex large scale optimizer not needed; usedADS. B

- Must go toward loose coupling to make furtherprogress in MDO. B

- Suitable MDO method is needed. H- CSSO CO class of decomposition too complex and

little understood. H- CSSO CO class of decomposition immature for

Industry. H- CSSO CO class of decomposition lacks software. H- Loosely coupled process needed to handle all of the

constraints. Y- Tightly coupled needed to be efficient and enable to

perform in reasonable time. W- Had to use tightly coupled system to make practical.

T- Loosely coupled preferred. H- Tightly coupled needed for special problems. H- Both are needed. H- Design done in cycles requiring sequential steps.

Various disciplines lagged several cycles; loads/flutter; one cycle, fatigue; 2 cycles, elastic-to-rigideffects on maneuver flight simulator; three cycles.

Some disciplines performed several cycles for eachglobal cycle. Rl

- Sometimes had to get “forward looking” models tohelp leap frog the global iteration. Rl

- Challenge is to provide tools to integrate disciplines.Bw

- Europe moving toward loosely coupled systems. Bw- Conceptual Design is multidisciplinary but low

fidelity. Bw- Preliminary Design is fragmented and configuration

flexibility lost. Bw- Don’t need close coupled black box. Bw- Do need loose coupled modular framework that can

use legacy codes. Bw- Multilevel Global-Local Structural optimization

employed in GARTEUR. Bw- MDO process needs to be flexible and

reconfigurable. Bw- Industrial design process does not necessarily fit a

particular MD procedure. Need flexibility andreconfigurability. B

• Optimization Procedures and Issues- MDO problem Size is an issue. W- Smoothness can be a problem. W R- Local minimums is a challenge. T- Ways to get around local minimums. T

1) New starting point.2) Broaden move and constraint limits initially.3) Change wt. factors in Obj. Function.4) Update fixed parameters periodically.

- Increase robustness by incrementally tighteningconstraints as optim. progresses.

- Structural sizing optimization. Rl- Manual “optimization” facilitated by common high

fidelity FEM, rapid turn- around on cycle updates,performing strength/fatigue sizing first then iteratingfor flutter. Rl

- Direct use of optimizer less attractive due to thenumber of possible function calls to high fidelityanalysis routines; possibly use a hybrid scheme. Bw

- Design variable linking needed. B- MDO robustness is lacking since different sites end

up with different optimums. Bw- Optimize system by running simulation of

interdisciplinary process. Lwf

Analysis Capabilities & Approximations• Breadth vs Depth Requirements

- All critical constraints needed otherwise weird results(e.g. pointed wing tips). W

- All critical mechanisms needed otherwise may loose amechanism to optimize. W

- Must have high fidelity or results are useless. Y- Inability to analytically (instead of wind tunnel tests)

determine design variable sensitivities is a need. Y

18

- Need capability for multiple configurations, fuel,

stores, actuator failure. H

• Approximation and Correction Processes- Response Surfaces may be too big and expensive. W- Response Surfaces play a key role. L- Use RS, Neural Nets, Taylor Series. B- Lack of Intermediate Fidelity Codes is a problem. W- Approximation based processes. H- Use Approximation models. B W- Approximation generation software needed. H- Correct intermediate methods using high fidelity data

(References 15 and 16)- Used nonlinear wind tunnel test data to correct linear

loads (especially with reference to control surfaces).Rl

- Use Response Surfaces to reduce cost. B W

• Effective Inclusion of High Fidelity Analyses/TestResults in Optimization and Design- Close coupled process made high fidelity possible. T- Replacing Wind Tunnel data in design process. Y- Possibly use Response Surfaces. W- Used an automatic FEM generator. H- Rapid CFD for air loads. Y- Need high fidelity propulsion integration for BWB;

cannot effectively include it at intermediate fidelitylevel. W

- Fidelity levels categorized; BwLevel 1- empirical modes (e.g. conceptual design)Level 2- intermediate level (e.g. beam str. models,panel aero etc.)Level 3- State-of-art high fidelity (e.g. CFE, FEM)

- MDO is moving toward Level 3 fidelity. Bw- Analysis times for high fidelity codes can make MDO

problem intractable. R

• Parametric Modeling- Need parametric geometric mode compatible with

current CAD systems. R- Parametric CAD not robust enough for topology

optimization. R- Need parametric/associative modeling and speed up

analysis. L- Automated robust model generation needed. H W- Lack of layout/material distribution algorithms. H- Discipline models too complex. H- Standard product process models and interfaces

catering to it. H- Automation and ease of use and checking is a barrier

in disciplinary analysis integration to MDO. H- Single high fidelity FEM used for stress, loads,

flutter, allowables, fatigue, aeroelastic effectiveness.Rl

- TDMB (Technical Data Modeller and Browser) candevelop a fully parameterized aircraft configuration

and associated aero, FEM and AE models. Dataconforms to STEP standards. Bw

- Unified parameterized geometry description. B- Need shareable common vehicle description and

approximations, easy ways to interact back and forthwith various disciplines. B

- Common models for structural, thermal, and opticalneeded. Lwf

• Analysis and Sensitivity Capability- Affordability missing in design. L- Move away from weight based cost. L- Manufacturability in design is needed. L- Need missing engine-out constraint for BWB. W- Bring controls into structural design process (early).

H L- Need to bring controls further up in the design

process. L- Nonlinear loads database is a big barrier. Y- Sensitivities will be used when design community

gets use to them. Y- Aero wind tunnel data can not produce sensitivities

needed in optimization. Y- Need aeroservoelastic integration. H- Need capability for multiple configurations, fuel,

stores, actuator failure. H- Lack of Robustness is a barrier to use of disciplinary

analysis in MDO. H- Employed high fidelity fuel tank loads. Rl- Used CFD and test to determine Hammer Shock. Rl- Maneuver load active controls used to reduce weight.

Rl- STARS code used for structural optimization. B- LAGRANGE code used for structural optimization.

H- Need nonlinear aero but is complex. H- Need standardized tool interfaces and disciplinary

analysis tools which are developed withinterdisciplinary interfacing in mind. H

- Moving to OTS codes for analysis and MDO (UG,PRO-E, I-DEAS, PATRAN, ANSYS, ABAQUS,NASTRAN, DEFORM etc.) R

- Applicability to MDO is a barrier to use ofdisciplinary analysis in MDO. H

- Working on a full structural, thermal, opticalsimulation process. Lwf

- Large internal load changes due to FEM gridchanges/refinements. Rl

- Important margin of safety changes due to smallinternal load or FEM changes. Rl

- Composite tailoring impractical due to costsassociated with testing/databases. Rl

Information Management, Data Flow & Processing• MDO Frameworks & Architecture

- Use the iSIGHT generic MDO system. R

19

- Developed GM system tailored to car design. B- Industry wants to use off the shelf tools AD included.

Y- Use generic Genie system and GUI but not

commercially available system. W- Stopped developing LAGRANGE because it might

be better to go to a general architecture. H- Need a flexible user configurable MDO architecture.

Current commercial optimization codes over sensitiveto details, do not always converge to optimum, andare not very flexible. Bw

- Commercial distributed computing not robust. Bw- GSE software is lacking. H- Software integration tools needed for design process

organization. H- Demonstrated Validated MDO software is needed. H- Interfacing various systems together and keeping

versions of software working is a concern. Lwf

• Databases, Data Flow & Standards- Common databases, database management all help. Y- Tool interfaced do not match. H- Tool interfaces is a barrier. H- Data standards are needed. H- Database updating of data and the ground rules that

generated them. B- ORACLE database used. Rl- Terabyte data handling needed. Rl- Recent projects supported by “Software Infrastructure

Group.” Bw- Need common standards for data definition and

intercommunication vehicle (STEP=standard for theexchange of product data). Bw

- Multi-site MDO and data capability needed. Bw- European MDO Process (Task 8) have provided the

following tools; software version management, datadefinition, database technology, process definition,process execution on distributed networks, datavisualization and optimization. Bw

- Common database used for F/A-18 E/F. Y- Database as support for Response Surfaces. T- Interdisciplinary data conversion and transmission is

a problem. Lwf

• Computing Requirements- F-22 needed a widely distributed, very heterogeneous

computing system. Rl- Very large memory 10 Terabytes needed for F-22. Rl- The amount of data stored for the F-22 was so large

that the process had constraints on the amount ofinformation manipulated. Plies were optimized laterin the process because of that. Rl

- Distributed computing used for F/A -18 E/F. Y- Computing infrastructure and available deployed

hardware not kept up to demands of MDO. R

• Design Space Visualization

- Need graphic visualization. H- Need process for extracting characteristic features of

a family of designs. H- Interpreting results is an obstacle to optimization. H- User friendly monitoring tools would be beneficial.

H- Need design space display. T

Management & Cultural Implementation• Organizational Structure

- No coordinating person for MDO in industry. H- MDO should be someone’s job in industrial

organization. B- Conflicting requirements impedes MDO

implementation - Conceptual Design. Dept. says thatMDO code does not include all the neededdisciplines- Analysis Dept. says that your models aretoo simple. B

- Industry sometimes lacks math skills and hasdifficulty with MDO formalisms. B

- Improved company organization would benefit useof MDO. H

- Loss of control by disciplinary experts is an issue. H- Coordinated Tri-company team by instituting detailed

ground rules, guidelines, and policies. Rl- Cross functional, partners, different locations. Rl- Coordinating and scheduling of multisite MDO

important. Bw

• MDO Operation within IPD Teams- Integration of IPT and MDO needed. W- IPT Needed to keep opt. on track. W- IPT (key organizational element) Needed to help

decide on opt. config. Y- Common High fidelity FEM model used to facilitate

communication and reduce cycle time and errors. Rl- Non optimal design of long lead time items

(actuators). Rl- Budget profiles did not always match process flow

requirements. Rl- Variations in results due to ground-rules which may

not be known ahead of time. Bw- IPD team direction and learning is needed to help

direct MDO and its ground rules. Bw

• Acceptance, Validation, Cost & Benefits- Validation is a MDO Cultural Acceptance Problem.

W- MDO produces benefits over conventional design in

performance, weight and in providing baselines forfurther analysis. W

- Lack of acceptance of MDO by management anddisciplinary experts. H

- Lack of validated MDO results is an obstacle to useof optimization. H

20

- Quantification of MDO benefit versus MDO

development cost is missing. H- Imbalance of fidelity is an obstacle to acceptance of

discipline trading via MDO. W- Large common FEM (and associated computing bill)

paid for itself many times over in reduced cycle time,increased communications and reduced errors. Rl

- Future European SM to verify MDO and its benefits.Bw

- MDO gives clear benefits as shown by tests (howeverno comparison with expected improvement). T

• Training- User familiarity and training is an obstacle to using

optimization. H- Engineering force having difficulty with optimization

formalisms and precisely defining objective function.B

- New engineers are more familiar with optimizaitontools. B

MDO TC Home Page

MULTIDISCIPLINARY DESIGN OPTIMIZATION TECHNICAL COMMITTEE

The AIAA Multidisciplinary Design Optimization Technical Committee (MDO TC) works to provide an AIAA forum for those active in development, application, and teaching of a formal design methodology based on the integration of disciplinary analyses and sensitivity analyses, optimization, and artificial intelligence, applicable at all stages of the multidisciplinary design of aerospace systems.

News and Announcements

● In May 2003 the AIAA MDO TC web site moved from its current location to a server at AIAA headquarters. Once the move occurs, the old MDO TC web site will remain available, but it will not be updated with new information. I'll send out email to the TC members once the new site is up and running.

Comments on the new web site are welcome.

-Tony Giunta, webmaster, [email protected]

Click here for a link to the NEW MDO TC web site.

Click here for a link to the old MDO TC web site.

Last Updated: 29 May 2003

Anthony A. Giunta, [email protected]

http://endo.sandia.gov/AIAA_MDOTC/main.html12/29/2006 12:28:43 PM

http://www.aiaa.org/tc/mdo


MDO TC Home Page

MULTIDISCIPLINARY DESIGN OPTIMIZATION TECHNICAL COMMITTEE

The AIAA Multidisciplinary Design Optimization Technical Committee (MDO TC) works to provide an AIAA forum for those active in development, application, and teaching of a formal design methodology based on the integration of disciplinary analyses and sensitivity analyses, optimization, and artificial intelligence, applicable at all stages of the multidisciplinary design of aerospace systems.

News and Announcements

● In May 2003 the AIAA MDO TC web site moveed from this site to its current location on a server at AIAA headquarters. The infomation on this web site is no longer updated.

-Tony Giunta, webmaster, [email protected]

Click here for a link to the new MDO TC web site.

Communications

● White Paper on Industrial Experience with MDO. ● TC email lists and browsable web archives:

Two electronic mailing lists are available for internal TC announcements.Contact Tony Giunta if you need the email addresses for these lists. (They are not listed here to reduce spam email.)

● TC meeting minutes ● TC subcommittee reports ● TC newsletters ● Download TC files

Sponsored Activities

http://endo.sandia.gov/AIAA_MDOTC/original.html (1 of 2)12/29/2006 12:28:44 PM

http://www.aiaa.org/participate/tcs/index.hfm?GetComm=80&tc=tc


MDO TC Home Page

(need to update)

● Conferences supported by the MDO TC ● Short courses supported by the MDO TC ● White papers and publications prepared by the MDO TC

Frequently Asked Questions

● What is MDO? ● What functions are performed by the AIAA MDO TC? ● How do I join the AIAA MDO TC?

Operations Information

● MDO TC Operations Manual ● TC operations slides in:

❍ HTML format. ❍ Portable Document Format (PDF file, Adobe Acrobat required). ❍ Downloadable postscript.

Last Updated: 17 October 2002


http://endo.sandia.gov/AIAA_MDOTC/original.html (2 of 2)12/29/2006 12:28:44 PM

http://www.adobe.com/Acrobat

ftp://endo.sandia.gov/pub/docs/mseldre/operations.ps


MDO Electronic Mailing Lists


The MDO Electronic Mailing Lists are serviced by Majordomo and the browsable archives are managed by LWGate. The following FAQ and help information provide additional details on Majordomo, although this background is not required to be able to use the MDO mailing lists.

mdotc_public:

The mdotc_public majordomo list sends electronic mail to all current AIAA MDO TC members. It is a public list in that all email traffic on the list is archived here on the TC Web site (accessible to anyone on the Web, not just members of the TC).

Therefore, TC members should use the mdotc_public list for matters concerning all TC members which are appropriate for posting on the Web. The purpose of the Web archives is to provide a convenient reference and central database of TC information and announcements. To send a public message to the TC, TC members should use the address [email protected]. Postings to the list are restricted to members of the TC.

With assistance from the MDO TC chairman's database, the Internet subcommittee chairman keeps the email addresses current. No action is required from the TC membership.

Archives from the mdotc_public list:

● mdotc_public 2006 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2005 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2004 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2003 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2002 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2001 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 2000 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 1999 sorted by: [ date ] [ subject ] [ author ] ● mdotc_public 1998 sorted by: [ date ] [ subject ] [ author ]

mdotc_private:

http://endo.sandia.gov/AIAA_MDOTC/forum/MDOTC_forum.html (1 of 2)12/29/2006 12:28:45 PM

http://www.greatcircle.com/majordomo/

http://www.netspace.org/users/dwb/lwgate.html

http://www.greatcircle.com/majordomo/majordomo-faq.html


http://endo.sandia.gov/cgi-bin/lwgate/MDOTC_PUBLIC/archives/archive.06/date/

http://endo.sandia.gov/cgi-bin/lwgate/MDOTC_PUBLIC/archives/archive.06/subject/

http://endo.sandia.gov/cgi-bin/lwgate/MDOTC_PUBLIC/archives/archive.06/author/


























Like the mdotc_public list, the mdotc_private list will send electronic mail to all AIAA MDO TC members. However, it is a private list in that its email traffic is not posted on the Web for all to see.

Therefore, TC members should use the mdotc_private list for matters concerning all TC members for which posting on the Web is inappropriate. To send a private message to the TC, TC members should use the address [email protected]. Postings to this list are restricted to members of the TC.

[ NO ARCHIVES ]


Last Updated: 12 September 2002

[email protected]

http://endo.sandia.gov/AIAA_MDOTC/forum/MDOTC_forum.html (2 of 2)12/29/2006 12:28:45 PM



http://endo.sandia.gov/AIAA_MDOTC/forum/help.txt

If you're familiar with mail servers, an advanced user's summary ofMajordomo's commands appears at the end of this message.

Majordomo is an automated system which allows users to subscribeand unsubscribe to mailing lists, and to retrieve files from listarchives.

You can interact with the Majordomo software by sending it commandsin the body of mail messages addressed to "[email protected]".Please do not put your commands on the subject line; Majordomo doesnot process commands in the subject line.

You may put multiple Majordomo commands in the same mail message.Put each command on a line by itself.

If you use a "signature block" at the end of your mail, Majordomo maymistakenly believe each line of your message is a command; you willthen receive spurious error messages. To keep this from happening,either put a line starting with a hyphen ("-") before your signature,or put a line with just the word

end

on it in the same place. This will stop the Majordomo software fromprocessing your signature as bad commands.

Here are some of the things you can do using Majordomo:

I. FINDING OUT WHICH LISTS ARE ON THIS SYSTEM

To get a list of publicly-available mailing lists on this system, put thefollowing line in the body of your mail message to [email protected]:

lists

Each line will contain the name of a mailing list and a brief descriptionof the list.

To get more information about a particular list, use the "info" command,supplying the name of the list. For example, if the name of the list about which you wish information is "demo-list", you would put the line

http://endo.sandia.gov/AIAA_MDOTC/forum/help.txt (1 of 6)12/29/2006 12:28:46 PM


info demo-list

in the body of the mail message.

II. SUBSCRIBING TO A LIST

Once you've determined that you wish to subscribe to one or more lists onthis system, you can send commands to Majordomo to have it add you to thelist, so you can begin receiving mailings.

To receive list mail at the address from which you're sending your mail,simply say "subscribe" followed by the list's name:

subscribe demo-list

If for some reason you wish to have the mailings go to a different address(a friend's address, a specific other system on which you have an account,or an address which is more correct than the one that automatically appears in the "From:" header on the mail you send), you would add that address tothe command. For instance, if you're sending a request from your workaccount, but wish to receive "demo-list" mail at your personal account(for which we will use "[email protected]" as an example), you'd putthe line

subscribe demo-list [email protected]

in the mail message body.

Based on configuration decisions made by the list owners, you may be added to the mailing list automatically. You may also receive notificationthat an authorization key is required for subscription. Another messagewill be sent to the address to be subscribed (which may or may not be thesame as yours) containing the key, and directing the user to send acommand found in that message back to [email protected]. (This can bea bit of extra hassle, but it helps keep you from being swamped in extraemail by someone who forged requests from your address.) You may alsoget a message that your subscription is being forwarded to the list ownerfor approval; some lists have waiting lists, or policies about who maysubscribe. If your request is forwarded for approval, the list ownershould contact you soon after your request.

Upon subscribing, you should receive an introductory message, containinglist policies and features. Save this message for future reference; it



will also contain exact directions for unsubscribing. If you lose theintro mail and would like another copy of the policies, send this messageto [email protected]:

intro demo-list

(substituting, of course, the real name of your list for "demo-list").

III. UNSUBSCRIBING FROM MAILING LISTS

Your original intro message contains the exact command which should beused to remove your address from the list. However, in most cases, youmay simply send the command "unsubscribe" followed by the list name:

unsubscribe demo-list

(This command may fail if your provider has changed the way youraddress is shown in your mail.)

To remove an address other than the one from which you're sendingthe request, give that address in the command:

unsubscribe demo-list [email protected]

In either of these cases, you can tell [email protected] to removethe address in question from all lists on this server by using "*"in place of the list name:

unsubscribe * unsubscribe * [email protected]

IV. FINDING THE LISTS TO WHICH AN ADDRESS IS SUBSCRIBED

To find the lists to which your address is subscribed, send this commandin the body of a mail message to [email protected]:

which

You can look for other addresses, or parts of an address, by specifyingthe text for which Majordomo should search. For instance, to find whichusers at my-isp.com are subscribed to which lists, you might send thecommand



which my-isp.com

Note that many list owners completely or fully disable the "which"command, considering it a privacy violation.

V. FINDING OUT WHO'S SUBSCRIBED TO A LIST

To get a list of the addresses on a particular list, you may use the"who" command, followed by the name of the list:

who demo-list

Note that many list owners allow only a list's subscribers to use the"who" command, or disable it completely, believing it to be a privacyviolation.

VI. RETRIEVING FILES FROM A LIST'S ARCHIVES

Many list owners keep archives of files associated with a list. Thesemay include:- back issues of the list- help files, user profiles, and other documents associated with the list- daily, monthly, or yearly archives for the list

To find out if a list has any files associated with it, use the "index"command:

index demo-list

If you see files in which you're interested, you may retrieve them byusing the "get" command and specifying the list name and archive filename.For instance, to retrieve the files called "profile.form" (presumably aform to fill out with your profile) and "demo-list.9611" (presumably themessages posted to the list in November 1996), you would put the lines

get demo-list profile.form get demo-list demo-list.9611

in your mail to [email protected].

VII. GETTING MORE HELP

To contact a human site manager, send mail to [email protected].



To contact the owner of a specific list, send mail to that list'sapproval address, which is formed by adding "-approval" to the user-nameportion of the list's address. For instance, to contact the list ownerfor [email protected], you would send mail [email protected].

To get another copy of this help message, send mail [email protected] a line saying

help

in the message body.

VIII. COMMAND SUMMARY FOR ADVANCED USERS

In the description below items contained in []'s are optional. Whenproviding the item, do not include the []'s around it. Items in anglebrackets, such as <address>, are meta-symbols that should be replacedby appropriate text without the angle brackets.

It understands the following commands:

subscribe <list> [<address>] Subscribe yourself (or <address> if specified) to the named <list>. unsubscribe <list> [<address>] Unsubscribe yourself (or <address> if specified) from the named <list>. "unsubscribe *" will remove you (or <address>) from all lists. This _may not_ work if you have subscribed using multiple addresses.

get <list> <filename> Get a file related to <list>.

index <list> Return an index of files you can "get" for <list>.

which [<address>] Find out which lists you (or <address> if specified) are on.

who <list> Find out who is on the named <list>.



info <list> Retrieve the general introductory information for the named <list>.

intro <list> Retrieve the introductory message sent to new users. Non-subscribers may not be able to retrieve this.

lists Show the lists served by this Majordomo server.

help Retrieve this message.

end Stop processing commands (useful if your mailer adds a signature).

Commands should be sent in the body of an email message to"[email protected]". Multiple commands can be processed providedeach occurs on a separate line.

Commands in the "Subject:" line are NOT processed.

If you have any questions or problems, please contact"[email protected]".


MDO TC Meeting Minutes


● September 2002 TC Meeting Minutes (Atlanta)

● April 2002 TC Meeting Minutes (Denver)

● January 2002 TC Meeting Minutes (Reno)

● April 2001 TC Meeting Minutes (Seattle)


● September 2000 TC Meeting Minutes (Long Beach)

● April 2000 TC Meeting Minutes (Atlanta)


● October 1999 TC Meeting Minutes (Internet meeting)

● April 1999 TC Meeting Minutes (St. Louis)


● September 1998 TC Meeting Minutes (St. Louis)

● April 1998 TC Meeting Minutes (Long Beach)


● October 1997 TC Meeting Minutes (St. Louis)

● April 1997 TC Meeting Minutes (Kissimmee)


● September 1996 TC Meeting Minutes (Bellevue)

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.html (1 of 2)12/29/2006 12:28:54 PM

http://www.nd.edu/~gtorres/mdotc/


● April 1996 TC Meeting Minutes (Salt Lake City)


● September 1995 TC Meeting Minutes (Los Angeles)


Last Updated: 26 June 2002

Tony Giunta, [email protected]

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.html (2 of 2)12/29/2006 12:28:54 PM


mdotc_apr02

MINUTES OF THE MDO TC MEETING

April 22, 2002

Denver,CO

PRELIMINARIES

Achille Messac called the meeting to order at 6:30 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Balabanov, Basu, Canfield, Gurdal, Kodiyalam, Smith, Striz, de Weck,Willcox ,Zang,Schweiger, Giunta,Tappeta. Other guests were Mattson, Mullur, Ismail-yahaya, Fadel, Engelsen,Wang, Malare, Chris Horton (AIAA), Thanks to these and their sponsoring organizations for attending.

MEMBERSHIP DATABASE INFORMATION REVIEW AND VERIFICATION

Current membership information, i.e., names, email addresss, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Achille Messac in order to maintain a consistent and up to date database. Changes will be passed on to Anthony Giunta for inclusion on the web site.

Reno MDO-TC (January 14, 02) MEETING MINUTES REVIEWED AND APPROVED

MA &O 2002

Balabanov discussed the progress on organizing this conference, which will be held at Grand Hyatt Atlanta from 4 - 6 Sept. 2002. The number of abstracts submitted for the conference was 285 and out of these 270 papers were accepted. The acceptance percentage was 94%. Participants upon registration will get a Book of Abstracts to plan their 3 days effectively. The General co-chairs for this conference are Farrokh Mistree and Dan Schrage from GT. The technical co-chairs for this conference are Dan DeLaurentis from GT School of AE and Pradeep Raj from Lockheed Martin Aeronautics Company. The theme of the conference will be System Affordability.

SDM 2002 meeting at Denver.

Kodiyalam represented MDO TC at this conference. Gurdal represented MDO TC at the SDM long range planning committee. Zafar mentioned that this committee is proposing new guidelines for chairing future conferences and it may not be possible for MDO TC representative to chair the SDM conference according to new rules. There was a discussion amongst the TC members about the small size of the

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.apr02.html (1 of 3)12/29/2006 12:28:55 PM

mdotc_apr02

rooms provided for the MAO sessions. Even though the room was supposed to accommodate 50 persons there were only 30 chairs and for some presentations there was no standing room.

SDM meeting at Norfolk, April. 2003

Srinivas and Khot will represent the MDO TC at this Conference.

Awards Committee (Canfield)

Bob Canfield discussed the activities of this subcommittee regarding membership upgrades, best paper award, MDO TC award etc. Top 12 papers were selected for the best paper award and this number will be reduced to three papers during the next step in the selection process.

Application Subcommittee (Purcell)

Guruswamy was not present but Frode Engelsen mentioned that the committee is still working on collecting and condensing inputs from MAO community. The target is to write a white paper based on the information collected.

Education Subcommittee(Kemper)

Kemper was not present but Doug Smith summarized the activities of the subcommittee. The committee is planning to have MAO paper competition during 2004 MAO conference. Their future plan includes marketing an AIAA endorsed home study course on MDO.

Publication Subcommittee (Striz)

Striz discussed future plans on setting up a web site for benchmark problems and definitions.

Presentation by Dr. Schweiger

Schweiger made presentation on the Active Aeroelastic Aircraft Structures research work being conducted at EADS Deutschland.

Plaque for Messac

Balabanov presented a plaque to Achille Messac on behalf of AIAA for his excellent work as the chairman of the MDO TC.


mdotc_apr02

White Paper Issues.

Achille Messac discussed the future plans for the white paper and then divided the members into four groups for brainstorming exercise. At the end representative form each group presented their ideas. Achille Messac requested each team representative to e-mail him summary of team’s ideas.

MEETING SCHEDULE

NEXT MEETING: 4-6 September 2002 at Atlanta, Georgia

The meeting was adjourned at 10:30 PM.

Respectfully submitted on 6/21/2002Narendra Khot AFRL/VASD 2210 Eighth Street Wright -Patterson AFB, OH, 45433-7531 Tel: (937) 255-8474 Fax: (937) 656-4945 Email:Narendra [email protected]


MDO TC Meeting Minutes (11 January 2000)


January 14, 2002

Reno, NV

PRELIMINARIES

Chairman Achille Messac called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Messac, Anderson, Balabanov, Batill, Bounajem, Canfield, DeLaurentis, Grossman, Guruswamy, Khot, Kramer, Purcell, Striz, Giunta. Other guests were Raj (Lockheed), and members of AIAA staff and members of the AIAA Technical Activities Committee. Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Achille Messac in order to maintain a consistent and up to date database. Changes will be passed on to Anthony Giunta for inclusion on the web site.

AIAA Activity

Steve Schultz from AIAA staff discussed the new Strategic Plan. He mentioned that the MDO TC would be moved from its current organizational home (Aircraft Technology Integration & Operation ) to a new home (Engineering & Technology and Management) . His presentation gave an impression that this decision taken by the members of the board of directors of the AIAA Technical Activities Committee, without consultation with MDO TC, was final. However, Robert Winn, Vice President-Elect, Technical Activities who came afterwards mentioned that this decision is not final and he will communicate to the board of directors the opinion of the MDO TC. It was then agreed that the AIAA staff will send the pertinent information to MDO TC Chair, who will pursue the matter further, taking into consideration the opinion of the members of the MDO TC committee. As of the writing of this report, Achille Messac has led a successful discussion with the whole TC and with AIAA, and the matter has been resolved to the satisfaction of all parties.

Seattle MDO-TC (Aprill 17 , 01) MEETING MINUTES REVIEWED AND APPROVED

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.jan02.html (1 of 4)12/29/2006 12:28:56 PM


MA &O 2002 STATUS REPORT

The two technical co-chairs, Dan DeLaurentis from GT School of AE and Pradeep Raj, discussed the progress on organizing this conference. Abstract submission deadline has been extended to 11 Feb. 2002. Super chairs and Special session organizers are in place. The conference will be held at Grand Hyatt Atlanta on 4-6 Sept. 2002.

SDM meeting at Denver, April 2001

Srinivas Kodiyalam represented the MDO TC at this Conference. He was unable to attend MDO TC meeting at Reno.


Bob Canfield discussed requirements for membership upgrades to Senior Member, Associate Fellow and Fellow. Bernie Grossman and Frank Eastep (previous member of TC) were elected Fellows. Evin Cramer and Ram Krishnamachari were elected Associate Fellows. Best paper nominations have been received and the selection committee is in the process of picking up the best paper for award during MA & O 2002 conference. The nominees for MDO award are due soon.

Education Subcommittee (Anderson)

Kurt Anderson presented his thoughts on MAO graduate and undergraduate paper competitions and marketing of AIAA endorsed home study course on MDO.

Survey

Guru Guruswamy has sent out a questioner to assess the state of the art on the use of high fidelity methods in multidisciplinary optimization. He discussed his related activities.

Publication Subcommittee (Stritz)

Stritz discussed future plans on publication of a color brochure explaining MDO to industry. He



also mentioned that he would like to collect segments of the Aerospace America Highlight articles of the last few years into a comprehensive MDO development story and put it on the MDO web page.

MDO TC Chair

Chairman Achille Messac was reelected unanimously by the TC for another term. Due to his upcoming responsibilities as the General Chair of the Multidisciplinary Analysis and Optimization Symposium in 2004, he will step down as TC Chair in one year. This will allow him to devote his full TC attention to the success of the conference. It was also proposed that the TC should consider changing the term of the TC Chair to three years for the next chair. This matter will be addressed further.

MEETING SCHEDULE

NEXT MEETING: Monday April 22 2002 at Denver CO , 6:30PM-10:00 PM.


Respectfully submitted on 3/6/2002 Narendra Khot AFRL/VASD 2210 Eighth Street Wright -Patterson AFB ,OH, 45433-7531 Tel: (937) 255-8474 Fax: (937) 656-4945 Email:Narendra [email protected]

Back to Meeting Minutes list


Last Updated: March 22, 2000





http://endo.sandia.gov/~mseldre/


Michael S. Eldred - Sandia National Laboratories

Michael S. Eldred

Telephone:Voice: (505) 844-6479Fax: (505) 284-2518

Address:Sandia National LaboratoriesP. O. Box 5800MailStop: 1318Org: 01411Albuquerque, NM 87185-1318

Location:Bldg: CSRIRoom: 230

E-mail:[email protected] (Personal)dakota-developers (DAKOTA project)

Experience

Mike received his B.S. in Aerospace Engineering from Virginia Tech in 1989, his M.S.E. and Ph.D. in Aerospace Engineering from the University of Michigan in 1990 and 1993, and is currently a Principal Member of the Technical Staff in the Optimization and Uncertainty Estimation Department within the Computation, Computers, Information, and Mathematics Center at Sandia National Laboratories.

Research

Mike is currently the project leader for the DAKOTA effort. Mike's research interests include surrogate-based optimization, uncertainty quantification, optimization under uncertainty, parallel processing, and object-oriented software development. A number of his publications are available on the DAKOTA web site.

Professional Activities

Mike is an Associate Fellow of the American Institute of Aeronautics and Astronautics (AIAA) and a member of the Society for Industrial and Applied Mathematics (SIAM), the International Society for Structural and Multidisciplinary Optimization (ISSMO) and the United States Association for Computational Mechanics (USACM). Mike served as a member of the AIAA Multidisciplinary Design Optimization Technical Committee from 1996-2001, and currently serves on the editorial board for Structure and Infrastructure Engineering: Maintenance, Management, Life-Cycle Design and Performance and as a member of the AIAA Nondeterministic Approaches Working Group.

http://www.cs.sandia.gov/~mseldre/ (1 of 2)12/29/2006 12:28:58 PM

http://www.sandia.gov/



http://www.cs.sandia.gov/optimization/

http://www.cs.sandia.gov/

http://www.cs.sandia.gov/

http://www.sandia.gov/

http://www.aiaa.org/

http://www.siam.org/

http://www.issmo.org/menu_page.html

http://www.issmo.org/menu_page.html

http://www.usacm.org/

http://www.tandf.co.uk/journals/titles/15732479.asp

http://www.tandf.co.uk/journals/titles/15732479.asp

Michael S. Eldred - Sandia National Laboratories

Quarterly Reports (Restricted Access)

© Sandia Corporation | Site Contact | Site Map | Privacy and Security

http://www.cs.sandia.gov/~mseldre/ (2 of 2)12/29/2006 12:28:58 PM

http://www.cs.sandia.gov/~mseldre/sandia_only

mailto:[email protected]?Subject=Web Site Comment

http://www.sandia.gov/general/site-map/index.html

http://www.sandia.gov/general/privacy-security/index.html

Large-scale Engineering Opt and UQ

DAKOTA software

Download DAKOTA

Applications

Publications

Research

Contacts

The DAKOTA Project:Large-scale Engineering Optimization and Uncertainty Analysis

DAKOTA Version 4.0 Released May 12, 2006

DAKOTA in the News

2006 Risk-Informed Design Flyer

April 2002 Press Release

November 2001 Lab News article

Supercomputing 2000 flyer

Overview

Computational methods developed in structural mechanics, heat transfer, fluid mechanics, shock physics, and many other fields of engineering can be an enormous aid to understanding the complex physical systems they simulate. Often, it is desired to use these simulations as virtual prototypes to obtain an acceptable or optimized design for a particular system. This effort seeks to enhance the utility of these computational methods by enabling their use as design tools, so that simulations may be used not just for single-point predictions, but also for automated determination of system performance improvements throughout the product life cycle. This allows analysts to address the fundamental engineering questions of foremost importance to our programs, such as "what is the best design?", "how safe is it?", and "how much confidence do I have in my answer?". System performance objectives can be formulated to minimize weight, cost, or defects; to limit a critical temperature, stress, or vibration response; or to maximize performance, reliability, throughput, reconfigurability, agility, or design robustness. A systematic, rapid method of determining these optimal solutions will lead to better designs and improved system performance and will reduce dependence on prototypes and testing, which will shorten the design cycle and reduce development costs.

Toward these ends, a general purpose software toolkit is under continuing development for the integration of commercial and in-house simulation capabilities with broad

http://www.cs.sandia.gov/DAKOTA/ (1 of 4)12/29/2006 12:29:00 PM

http://www.sandia.gov/Main.html


http://www.sandia.gov/search.html

http://www.sandia.gov/News.htm

http://www.sandia.gov/Contacting.htm

http://www.sandia.gov/Working.htm

http://www.sandia.gov/Solution.htm

http://www.sandia.gov/About.htm

http://www.cs.sandia.gov/DAKOTA/software.html


http://www.cs.sandia.gov/DAKOTA/licensing/license.html



http://www.cs.sandia.gov/DAKOTA/papers/Dept1533_FactSheet_2006.pdf

http://www.sandia.gov/media/NewsRel/NR2002/DAKOTA.htm

http://www.sandia.gov/LabNews/LN11-16-01/key11-16-01_stories.html#dakota

http://www.cs.sandia.gov/DAKOTA/papers/SC2000Final.pdf


classes of systems analysis tools. Written in C++, the DAKOTA (Design Analysis Kit for Optimization and Terascale Applications) toolkit is intended as a flexible, extensible interface between simulation codes and iterative systems analysis methods. In addition to optimization methods, DAKOTA implements uncertainty quantification with sampling, reliability, and stochastic finite element methods, parameter estimation with nonlinear least squares methods, and sensitivity/variance analysis with design of experiments and parameter study capabilities. These capabilities may be used on their own or as components within advanced strategies such as surrogate-based optimization, mixed integer nonlinear programming, or optimization under uncertainty. By employing object-oriented design to implement abstractions of the key components required for iterative systems analyses, the DAKOTA toolkit provides a flexible and extensible problem-solving environment as well as a platform for rapid prototyping of advanced methodologies which focus on increasing robustness and efficiency for computationally complex engineering problems. Several of these research programs are discussed below.

Large-scale Applications

Multilevel Parallel Optimization with Salinas

CVD Reactor Design with MPSalsa

Publications

A list of recent and upcoming publications is available. Many can be downloaded in PDF or postscript form.

Research Directions

Research programs have focused on several areas:

● Parallel processing. Selected publications include:

❍ "Multilevel Parallel Optimization Using Massively Parallel Structural Dynamics".



http://www.cs.sandia.gov/DAKOTA/papers/StrOpt_JPaper.pdf





http://www.cs.sandia.gov/DAKOTA/papers/CVD_2004.pdf









❍ "Multilevel Parallelism for Optimization on MP Computers: Theory and Experiment".

● Surrogate-based optimization.

❍ SBO with data fits.

❍ SBO with multifidelity models.

❍ SBO with reduced-order modeling (ROM).

● Uncertainty quantification. Selected publications include:

❍ "Investigation of Reliability Method Formulations in DAKOTA/UQ".

❍ "A Toolkit For Uncertainty Quantification In Large Computational Engineering Models".

● Optimization under uncertainty.

❍ Surrogate-based.

❍ Reliability-based.

Contacts

DAKOTA Software Contact: dakota-developers. Technical Contact: Michael S. Eldred, Principal Member of Technical Staff. Business Contact: Scott A. Mitchell, Manager of Optimization and Uncertainty Estimation Department. Location

Sandia National LaboratoriesP. O. Box 5800, Mail Stop 0370Albuquerque, NM 87185-0370

Back to top of page || Questions and Comments || Acknowledgment and Disclaimer


dakota-developers


http://www.cs.sandia.gov/DAKOTA/papers/MDO2000_paper.pdf


http://www.cs.sandia.gov/DAKOTA/papers/MDO2000_sao_paper.pdf

http://www.cs.sandia.gov/DAKOTA/papers/SBO_Corr_MAO_paper.pdf

http://www.cs.sandia.gov/DAKOTA/papers/SIE-KM-005.pdf

http://www.cs.sandia.gov/DAKOTA/papers/SDM2001_uq_paper.pdf


http://www.cs.sandia.gov/DAKOTA/papers/MAO2002paper4875.pdf


mailto: [email protected]

http://www.cs.sandia.gov/~mseldre

http://www.cs.sandia.gov/~samitch



http://www.sandia.gov/feedback.htm

http://www.sandia.gov/dis.htm




Recent and Upcoming Publications

Optimization Home

Journal Papers

Conference Papers

Conference Abstracts

SAND Reports


Journal Papers

● Giunta, A.A., McFarland, J. M., Swiler, L.P., and Eldred, M.S., "The promise and peril of uncertainty quantification using response surface approximations," Structure & Infrastructure Engineering: Maintenance, Management, Life-Cycle Design & Performance, special issue on Uncertainty Quantification and Design under Uncertainty of Aerospace Systems, Vol. 2, Nos. 3-4, Sept.-Dec. 2006, pp. 175-189.

● Eldred, M.S., Agarwal, H., Perez, V.M., Wojtkiewicz, S.F., Jr., and Renaud, J.E., "Investigation of Reliability Method Formulations in DAKOTA/UQ," Structure & Infrastructure Engineering: Maintenance, Management, Life-Cycle Design & Performance (to appear), Taylor & Francis Group.

❍ Also appears in Proceedings of the 9th ASCE Joint Specialty Conference on Probabilistic Mechanics and Structural Reliability, Albuquerque, NM, July 26-28, 2004.

● Lemke, R.W., Knudson, M.D., Bliss, D.E., Cochrane, K., Davis, J.-P., Giunta, A.A., Harjes, H.C., and Slutz, S.A., "Magnetically accelerated, ultrahigh velocity flyer plates for shock wave experiments," J. Applied Physics, Vol. 98, 2005.

● Salinger, A.G., Pawlowski, R.P., Shadid, J.N., and van Bloemen Waanders, B., "Computational Analysis and Optimization of a Chemical Vapor Deposition Reactor with Large-Scale Computing," Industrial and Engineering Chemistry Research, in press, 2004.

● Simpson, T. W., Booker, A. J., Ghosh, D., Giunta, A. A., Koch, P. N. and Yang, R.-J., "Approximation Methods in Multidisciplinary Analysis and Optimization: A Panel Discussion," Structural and Multidisciplinary Optimization, Vol. 27, No. 5, 2004, pp. 302-313.

● Eldred, M.S., Giunta, A.A., and van Bloemen Waanders, B.G., "Multilevel Parallel Optimization Using Massively Parallel

http://www.cs.sandia.gov/DAKOTA/references.html (1 of 15)12/29/2006 12:29:02 PM



http://www.sandia.gov/search.html

http://www.sandia.gov/News.htm

http://www.sandia.gov/Contacting.htm

http://www.sandia.gov/Working.htm

http://www.sandia.gov/Solution.htm

http://www.sandia.gov/About.htm

http://www.cs.sandia.gov/DAKOTA/index.html

http://www.cs.sandia.gov/DAKOTA/index.html



http://www.cs.sandia.gov/DAKOTA/papers/UQ_Reliability_paper_PMC04.pdf





Structural Dynamics," Structural and Multidisciplinary Optimization, Springer-Verlag, Vol. 27, Nos. 1-2, May 2004, pp. 97-109.

❍ Also appears as paper AIAA-2001-1625 in Proceedings of the 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, Seattle, WA, April 16-19, 2001.

● Romero, V. J., Swiler L. P., and Giunta, A. A., "Construction of Response Surfaces Based on Progressive-Lattice-Sampling Experimental Designs," Structural Safety, Vol. 26, No. 2, pp. 201-219, 2004.

● Biegler, L.T., Ghattas, O., Heinkenschloss, M., and van Bloemen Waanders, B., "Large-scale PDE-constrained Optimization: An Introduction," Large-Scale PDE-Constrained Optimization, Lecture Notes in Computational Science and Engineering, No. 30, Springer-Verlag, 2003, pp. 3-13.

● Salinger, A.G., Pawlowski, R.P., Shadid, J.N., van Bloemen Waanders, B., Bartlett, R., Itle, G.C., and Biegler, L., "rSQP Optimization of Large Scale Reacting Flow Applications with MPSalsa," Large-Scale PDE-Constrained Optimization, Lecture Notes in Computational Science and Engineering, No. 30, Springer-Verlag, 2003, pp. 45-59.

● Bartlett, R.A. and Biegler, L.T., "rSQP++: An Object-Oriented Framework for Successive Quadratic Programming," Large-Scale PDE-Constrained Optimization, Lecture Notes in Computational Science and Engineering, No. 30, Springer-Verlag, 2003, pp. 316-330.

● Greenberg, D.S., Hart, W.E., and Phillips, C.A., "Enabling Department-Scale Supercomputing," Algorithms for Parallel Processing, IMA Volumes in Mathematics and Its Applications, 105:321-344. 1999.

● Hart, W.E., "Sequential Stopping Rules for Random Optimization Methods with Applications to Multistart Local Search," SIAM Journal of Optimization, 1999, pp. 270-290. Also appears as Sandia



http://www.cs.sandia.gov/DAKOTA/papers/SantaFePDE_rSQP.pdf

http://www.cs.sandia.gov/DAKOTA/papers/SantaFePDE_rSQP.pdf

http://www.cs.sandia.gov/DAKOTA/papers/DeptScaleSC.pdf

http://www.cs.sandia.gov/DAKOTA/papers/DeptScaleSC.pdf

http://www.cs.sandia.gov/~wehart/abstracts.html#Har99

http://www.cs.sandia.gov/~wehart/abstracts.html#Har99


Technical Report SAND94-2666, Nov. 1994.

● Dowding, K.J., and Blackwell, B.F., "Joint Experimental/Computational Techniques to Measure Thermal Properties of Solids," Measurement Science and Technology, Vol. 9, No. 6, June 1998, pp. 877-887.

● Eldred, M.S., Outka, D.E., Bohnhoff, W.J., Witkowski, W.R., Romero, V.J., Ponslet, E.R., and Chen, K.S., "Optimization of Complex Mechanics Simulations with Object-Oriented Software Design," Computer Modeling and Simulation in Engineering, Vol. 1, No. 3, August 1996, pp. 323-352.

❍ Also appears as paper AIAA-95-1433 in Proceedings of the 36th AIAA/ASME/ ASCE/AHE/ASC Structures, Structural Dynamics, and Materials Conference, New Orleans, LA, April 10-13, 1995, pp. 2406-2415.

Conference Papers

● Bichon, B.J., Eldred, M.S., Swiler, L.P., Mahadevan, S., and McFarland, J.M., "Multimodal Reliability Assessment for Complex Engineering Applications using Sequential Kriging Optimization," abstract submitted for 48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference (9th AIAA Non-Deterministic Approaches Conference), Honolulu, HI, April 23-26, 2007.

● Eldred, M.S., Adams, B.M., Copps, K.D., Carnes, B., Notz, P.K., Hopkins, M.M., and Wittwer, J.W., "Solution-Verified Reliability Analysis and Design of Compliant Micro-Electro-Mechanical Systems," abstract submitted for 48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference (9th AIAA Non-Deterministic Approaches Conference), Honolulu, HI, April 23-26, 2007.

● Adams, B.M., Eldred, M.S., Wittwer, J., and Massad, J., "Reliability-Based Design Optimization for Shape Design of Compliant Micro-Electro-Mechanical Systems," paper AIAA-2006-7000 in the


http://www.cs.sandia.gov/DAKOTA/papers/mstjoint.pdf



http://www.cs.sandia.gov/DAKOTA/papers/CMSE96.html



http://www.cs.sandia.gov/DAKOTA/papers/AIAA-2006-7000.pdf




Proceedings of the 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth, VA, Sept. 6-8, 2006.

● Eldred, M.S. and Dunlavy, D.M., "Formulations for Surrogate-Based Optimization with Data Fit, Multifidelity, and Reduced-Order Models," paper AIAA-2006-7117 in the Proceedings of the 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth, VA, Sept. 6-8, 2006.

● Giunta, A.A., Swiler, L.P., Brown, S.L., Eldred, M.S., Richards, M.D., and Cyr, E.C., "The Surfpack Software Library for Surrogate Modeling of Sparse Irregularly Spaced Multidimensional Data," paper AIAA-2006-7049 in the Proceedings of the 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth, VA, Sept. 6-8, 2006.

● Robinson, T.D., Willcox, K.E., Eldred, M.S., and Haimes, R., "Multifidelity Optimization for Variable-Complexity Design," paper AIAA-2006-7114 in the Proceedings of the 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth, VA, Sept. 6-8, 2006.

● Eldred, M.S. and Bichon, B.J., "Second-Order Reliability Formulations in DAKOTA/UQ," paper AIAA-2006-1828 in Proceedings of the 47th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference (8th AIAA Non-Deterministic Approaches Conference), Newport, Rhode Island, May 1 - 4, 2006.

● Robinson, T.D., Eldred, M.S., Willcox, K.E., and Haimes, R., "Strategies for Multifidelity Optimization with Variable Dimensional Hierarchical Models," paper AIAA-2006-1819 in Proceedings of the 47th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference (2nd AIAA Multidisciplinary Design Optimization Specialist Conference), Newport, Rhode Island, May 1 - 4, 2006.

● Weickum, G., Eldred, M.S., and Maute, K., "Multi-point Extended Reduced Order Modeling For Design Optimization and Uncertainty Analysis," paper AIAA-2006-2145 in Proceedings of the 47th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and
















Materials Conference (2nd AIAA Multidisciplinary Design Optimization Specialist Conference), Newport, Rhode Island, May 1 - 4, 2006.

● Eldred, M.S., Bichon, B.J., and Adams, B.M., "Overview of Reliability Analysis and Design Capabilities in DAKOTA," Proceedings of the NSF Workshop on Reliable Engineering Computing (REC 2006), Savannah, GA, February 22-24, 2006.

● Eldred, M.S., Giunta, A.A., and Collis, S.S, "Second-Order Corrections for Surrogate-Based Optimization with Model Hierarchies," paper AIAA-2004-4457 in Proceedings of the 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Albany, NY, Aug. 30 - Sept. 1, 2004.

● Giunta, A.A., Eldred, M.S., Swiler, L.P., Trucano, T.G., and Wojtkiewicz, S.F., Jr., "Perspectives on Optimization Under Uncertainty: Algorithms and Applications" paper AIAA-2004-4451 in Proceedings of the 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Albany, NY, Aug. 30 - Sept. 1, 2004.

● Perez, V.M., Eldred, M.S., and Renaud, J.E., "Solving the Infeasible Trust-region Problem Using Approximations," paper AIAA-2004-4312 in Proceedings of the 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Albany, NY, Aug. 30 - Sept. 1, 2004.

● Giunta, A.A., Eldred, M.S., and Castro, J.P., "Uncertainty Quantification Using Response Surface Approximations," Proceedings of the 9th ASCE Joint Specialty Conference on Probabilistic Mechanics and Structural Reliability, Albuquerque, NM, July 26-28, 2004.

● Perez, V.M., Eldred, M.S., and Renaud, J.E., "An rSQP Approach for a Single-Level Reliability Optimization," Proceedings of the 9th ASCE Joint Specialty Conference on Probabilistic Mechanics and Structural Reliability, Albuquerque, NM, July 26-28, 2004.

● van Bloemen Waanders, B., "Application of Optimization Methods to the Calibration of Water Distribution Systems," Proceedings of the World Water and Environmental Resources Congress (EWRI),


http://www.cs.sandia.gov/DAKOTA/papers/Overview_Reliability_color.pdf

http://www.cs.sandia.gov/DAKOTA/papers/Overview_Reliability_color.pdf




http://www.cs.sandia.gov/DAKOTA/papers/OUU_MAO2004.pdf

http://www.cs.sandia.gov/DAKOTA/papers/OUU_MAO2004.pdf

http://www.cs.sandia.gov/DAKOTA/papers/Infeas_TR_MAO2004.pdf

http://www.cs.sandia.gov/DAKOTA/papers/Infeas_TR_MAO2004.pdf

http://www.cs.sandia.gov/DAKOTA/papers/UQ_RSM_PMC04.pdf

http://www.cs.sandia.gov/DAKOTA/papers/UQ_RSM_PMC04.pdf

http://www.cs.sandia.gov/DAKOTA/papers/rSQP_PMC04.pdf

http://www.cs.sandia.gov/DAKOTA/papers/rSQP_PMC04.pdf


Salt Lake City, UT, June 27 - July 1, 2004.

● Giunta, A.A., Wojtkiewicz, S.F., Jr., and Eldred, M.S., "Overview of Modern Design of Experiments Methods for Computational Simulations," paper AIAA-2003-0649 in Proceedings of the 41st AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, Jan. 6-9, 2003.

● Simpson, T.W., Booker, A.J., Ghosh, D., Giunta, A.A., Koch, P.N., and Yang, R.-J., "Approximation Methods in Multidisciplinary Analysis and Optimization: A Panel Discussion," 3rd ISSMO/AIAA Internet Conference on Approximations in Optimization, Oct. 14-18, 2002.

● Eldred, M.S., Giunta, A.A., Wojtkiewicz, S.F., Jr., and Trucano, T.G., "Formulations for Surrogate-Based Optimization Under Uncertainty," paper AIAA-2002-5585 in Proceedings of the 9th AIAA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Atlanta, GA, Sept. 4-6, 2002.

● Giunta, A.A., Eldred, M.S., Trucano, T.G., and Wojtkiewicz, S.F., Jr., "Optimization Under Uncertainty Methods for Computational Shock Physics Applications," paper AIAA-2002-1642 in Proceedings of the 43rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference (Nondeterministic Approaches Forum), Denver, CO, April 22-25, 2002.

● Giunta, A. A., "Use of Data Sampling, Surrogate Models, and Numerical Optimization in Engineering Design," paper AIAA-2002-0538 in Proceedings of the 40th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, Jan. 2002.

● Wojtkiewicz, S.F., Jr., Eldred, M.S., Field, R.V., Jr., Urbina, A., and Red-Horse, J.R., "A Toolkit For Uncertainty Quantification In Large Computational Engineering Models," paper AIAA-2001-1455 in Proceedings of the 42nd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, Seattle, WA, April 16-19, 2001.

● Eldred, M.S., Hart, W.E., Schimel, B.D., and van Bloemen Waanders, B.G., "Multilevel Parallelism for Optimization on MP


http://www.cs.sandia.gov/DAKOTA/papers/sand2003-0209C.pdf



http://www.cs.sandia.gov/DAKOTA/papers/panel_review.pdf

http://www.cs.sandia.gov/DAKOTA/papers/panel_review.pdf



http://www.cs.sandia.gov/DAKOTA/papers/sdm2002_1642.pdf

http://www.cs.sandia.gov/DAKOTA/papers/sdm2002_1642.pdf

http://www.cs.sandia.gov/DAKOTA/papers/reno2002_0538.pdf

http://www.cs.sandia.gov/DAKOTA/papers/reno2002_0538.pdf





Computers: Theory and Experiment," paper AIAA-2000-4818 in Proceedings of the 8th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Long Beach, CA, September 6-8, 2000.

● Giunta, A.A., and Eldred, M.S., "Implementation of a Trust Region Model Management Strategy in the DAKOTA Optimization Toolkit," paper AIAA-2000-4935 in Proceedings of the 8th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Long Beach, CA, September 6-8, 2000.

● Eldred, M.S., and Schimel, B.D., "Extended Parallelism Models for Optimization on Massively Parallel Computers," paper 16-POM-2 in Proceedings of the 3rd World Congress of Structural and Multidisciplinary Optimization (WCSMO-3), Amherst, NY, May 17-21, 1999.

● Eldred, M.S., and Hart, W.E., "Design and Implementation of Multilevel Parallel Optimization on the Intel TeraFLOPS," paper AIAA-98-4707 in Proceedings of the 7th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, St. Louis, MO, Sept. 2-4, 1998, pp. 44-54.

● Blackwell, B.F., and Eldred, M.S., "Application of Reusable Interface Technology for Thermal Parameter Estimation," Proceedings of the 32nd National Heat Transfer Conference, Vol. 2, Eds. Dulikravitch, G.S., and Woodbury, K.E., HTD-Vol. 340, August 1997, pp. 1-8.

● Chen, K.S., and Witkowski, W.R., "Design Optimization of Liquid-Distribution Chamber-Slot Dies Using the DAKOTA Toolkit," 50th Annual Conference of the Society for Imaging Science and Technology, Cambridge MA, May 18-23, 1997.

● Hobbs, M. L., "A Global HMX Decomposition Model," 1996 JANNAF Propulsion Systems Hazards Subcommittee Meeting, Naval Postgraduate School, Monterey, CA, Nov. 4-8, 1996.

● Eldred, M.S., Hart, W.E., Bohnhoff, W.J., Romero, V.J., Hutchinson, S.A., and Salinger, A.G., "Utilizing Object-Oriented Design to Build Advanced Optimization Strategies with Generic






http://www.cs.sandia.gov/DAKOTA/papers/WCSMO99_longpaper.pdf

http://www.cs.sandia.gov/DAKOTA/papers/WCSMO99_longpaper.pdf



http://www.cs.sandia.gov/DAKOTA/papers/97nhtc_2col.pdf

http://www.cs.sandia.gov/DAKOTA/papers/97nhtc_2col.pdf




Implementation," paper AIAA-96-4164 in Proceedings of the 6th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Bellevue, WA, Sept. 4-6, 1996, pp. 1568-1582.

● Moen, C.D., Spence, P.A., Meza, J.C., and Plantenga, T.D., "Automatic Differentiation for Gradient-Based Optimization of Radiatively Heated Microelectronics Manufacturing Equipment," paper AIAA-96-4118 in Proceedings of the 6th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Bellevue, WA, Sept. 4-6, 1996, pp. 1167-1175.

● Ponslet, E.R., and Eldred, M.S., "Discrete Optimization of Isolator Locations for Vibration Isolation Systems: an Analytical and Experimental Investigation," paper AIAA-96-4178 in Proceedings of the 6th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, Bellevue, WA, Sept. 4-6, 1996, pp. 1703-1716. Also appears as Sandia Technical Report SAND96-1169, May 1996.

● Hart, W.E., "A Stationary Point Convergence Theory for Evolutionary Algorithms," Proceedings of Foundations of Genetic Algorithms 4, San Diego, CA, August 3-5, 1996, pp. 325-342.

● Hart, W.E., Baden, S., Belew, R.K., Kohn, S., "Analysis of the Numerical Effects of Parallelism on a Parallel Genetic Algorithm," Proceedings of the 10th International Parallel Processing Symposium(IPPS `96), Honolulu, HI, April 15-19, 1996, pp. 606-612.

● Hart, W.E., "A Theoretical Comparison of Evolutionary Algorithms and Simulated Annealing," Proceedings of the Fifth Annual Conference on Evolutionary Programming (EP `96), San Diego, CA, February 29 - March 2, 1996, pp. 147-154.

● Harding, D.C., Eldred, M.S., and Witkowski, W.R., "Integration of Finite Element Analysis and Numerical Optimization Techniques for RAM Transport Package Design," Proceedings of the 11th International Conference on the Packaging and Transportation of Radioactive Materials (PATRAM `95), Las Vegas, NV, Dec. 3-8, 1995.

● Harding, D.C., and Eldred, M.S., "Radioactive Material



http://www.cs.sandia.gov/DAKOTA/papers/VibIsol_SAND961169.pdf

ftp://ftp.cs.sandia.gov/pub/papers/wehart/1997/Har97-foga.ps.gz

ftp://ftp.cs.sandia.gov/pub/papers/wehart/1997/Har97-foga.ps.gz

ftp://ftp.cs.sandia.gov/pub/papers/wehart/opt/ipps96.ps.gz

ftp://ftp.cs.sandia.gov/pub/papers/wehart/opt/ipps96.ps.gz

ftp://ftp.cs.sandia.gov/pub/papers/wehart/1996/Har96-ep.ps.gz

ftp://ftp.cs.sandia.gov/pub/papers/wehart/1996/Har96-ep.ps.gz


Transportation Package Design Using Numerical Optimization Techniques," Proceedings of the 1995 Joint ASME/JSME Pressure Vessels and Piping Conference, Honolulu, Hawaii, July 23-27, 1995, Vol. PVP-307, pp. 29-36.

● Romero, V.J., Eldred, M.S., Bohnhoff, W.J., and Outka, D.E., "Application of Optimization to the Inverse Problem of Finding the Worst-Case Heating Configuration in a Fire," Proceedings of the 9th International Conference on Numerical Methods in Thermal Problems, Atlanta, GA, July 17-21, 1995, Vol. 9, Part 2, pp. 1022-1033.

● Witkowski, W.R., Eldred, M.S., and Harding, D.C., "Integration of Numerical Analysis Tools for Automated Numerical Optimization of a Transportation Package Design," Proceedings of the 5th AIAA/NASA/USAF/ISSMO Symposium on Multidisciplinary Analysis and Optimization, paper AIAA94-4259, Panama City Beach, FL, Sept. 7-9, 1994.

Conference Abstracts

● Carnes, B., Copps, K.D., Eldred, M.S., Adams, B.M., Wittwer, J.W., "Coupled a posteriori error estimation and uncertainty quantification for a nonlinear elasticity MEMS problem," abstract for SIAM Conference on Computational Science and Engineering (CSE07), Costa Mesa, CA, February 19-23, 2007.

● Dunlavy, D.M. and Eldred, M.S., "Formulations for Surrogate-Based Optimization Using Data Fit and Multifidelity Models," abstract for SIAM Conference on Computational Science and Engineering (CSE07), Costa Mesa, CA, February 19-23, 2007.

● Robinson, T.D., Willcox, K.E., Eldred, M.S., and Haimes, R., "Multifidelity Optimization for Variable-Complexity Design," abstract submitted for Second International Workshop on Surrogate Modeling and Space Mapping for Engineering Optimization, Lyngby, Denmark, Nov. 9-11, 2006.

● Giunta, A.A., Castro, J.P., Hough, P.D.,Gray, G.A., Eldred, M.S., "Multifidelity Modeling Approaches in Simulation-Based



Optimization," abstract for the SIAM Conference on Optimization, Stockholm, Sweden, May 15-19, 2005.

● Giunta, A.A., Eldred, M.S., Hough, P.D., and Castro, J.P., "Overview of Surrogate-Based Optimization Research and Applications at Sandia National Laboratories," abstract for the Surrogate Optimization Workshop, Houston, TX, May 24-25, 2004.

● Giunta, A.A. and Eldred, M.S., "Robust Design Optimization Using Surrogate Models," abstract for the Robust Optimization-Directed Design (RODD) Conference, Shalimar, FL, April 19-21, 2004.

● Giunta, A.A., Eldred, M.S., Wojtkiewicz, S.F., Jr., Trucano, T.G., and Castro, J.P., "Surrogate-Based Optimization Methods for Engineering Design," abstract in Proceedings of the Fifth Biennial Tri-Laboratory Engineering Conference on Computational Modeling, Santa Fe, NM, October 21-23, 2003.

● Giunta, A. A., and Eldred, M. S., "Surrogate-Based Optimization Under Uncertainty: Formulations and Applications" abstract in the Proceedings of the 18th International Symposium on Mathematical Programming, Copenhagen, Denmark, Aug. 2003.

● Eldred, M.S., Giunta, A.A., Wojtkiewicz, S.F., Jr., and Trucano, T.G., "Formulations for Surrogate-Based Optimization Under Uncertainty," abstract in Proceedings of the 7th U.S. National Congress on Computational Mechanics, Albuquerque, NM, July 28-30, 2003.

● Giunta, A.A. and Eldred, M.S., "Engineering Design Optimization Algorithms: Theory and Practice," abstract in Proceedings of the 7th U.S. National Congress on Computational Mechanics, Albuquerque, NM, July 28-30, 2003.

● Eldred, M.S., Giunta, A.A., Wojtkiewicz, S.F., Jr., and Trucano, T.G., "Surrogate-Based Optimization Under Uncertainty: Status and Directions," abstract in SIAM Conference on Computational Science and Engineering. Final Program and Abstracts, San Diego, CA, Feb 10-13, 2003.

● Giunta, A.A. and Eldred, M.S., "Case Studies in Computational Engineering Design Optimization: Challenges and Solutions,"



abstract in SIAM Conference on Computational Science and Engineering. Final Program and Abstracts, San Diego, CA, Feb 10-13, 2003.

● Eldred, M.S., "DAKOTA: Virtual Prototyping with Large-Scale Engineering Simulations," abstract in IMA Workshop 4: Optimization in Simulation-Based Models, Minneapolis, MN, January 9-16, 2003.

● Eldred, M.S., "The DAKOTA Optimization Framework: Virtual Prototyping with ASCI-Scale Simulations," abstract in Proceedings of the Fourth Biennial Tri-Laboratory Engineering Conference on Computational Modeling, Albuquerque, NM, Oct. 23-24, 2001, p. 82.

● Wojtkiewicz, S.F., Jr., Field, R.V., Jr., Eldred, M.S., Red-Horse, J.R., and Urbina, A., "Uncertainty Quantification in Large Computational Engineering Models," abstract in Proceedings of the Fourth Biennial Tri-Laboratory Engineering Conference on Computational Modeling, Albuquerque, NM, Oct. 23-24, 2001, p. 11.

● Hart, W. E., Giunta, A. A., Salinger, A. G., and van Bloemen Waanders, B., "An Overview of the Adaptive Pattern Search Algorithm and its Application to Engineering Optimization Problems," abstract in Proceedings of the McMaster Optimization Conference: Theory and Applications, McMaster University, Hamilton, Ontario, Canada, August 2001, p. 20.

● Wojtkiewicz, S.F., Jr., Eldred M.S., Field, R.V., Jr., Urbina, A., Red-Horse, J.R., and Giunta, A.A., "DAKOTA/UQ: A Toolkit for Uncertainty Quantification in a Multiphysics, Massively Parallel Computational Environment," presented as (1) poster at ODU-NASA Training Workshop on Nondeterministic Approaches and Their Potential for Future Aerospace Systems, held in Langley, Virginia, May 30-31, 2001, (2) abstract (no proceedings) at USNCCM VI (Sixth United States Congress on Computational Mechanics) held in Dearborn, Michigan, August 1-3, 2001, and (3) abstract (no proceedings) at LLNL Sensitivity Analysis Workshop, August 16-17, 2001.

● van Bloemen Waanders, B., "Simultaneous Analysis and Design Optimization of Massively Parallel Simulation Codes using Object Oriented Framework," abstract for Tenth SIAM Conference on


http://www.cs.sandia.gov/DAKOTA/papers/abstract_mopta.pdf




Parallel Processing for Scientific Computing, March 2001.

● Giunta, A.A., "Coupling High-Performance Computing, Optimization, and Shock Physics Simulations," abstract in session CP04 of the Final Program of the First SIAM Conference on Computational Science and Engineering, Washington, DC, September 21-23, 2000, p. 47.

● Hart, W.E., Eldred, M.S., and Giunta, A.A., "Solving mixed-integer nonlinear problems with PICO," abstract in proceedings of the 17th International Symposium on Mathematical Programming (ISMP 2000), Atlanta, GA, August 7-11, 2000.

● van Bloemen Waanders, B.G., Eldred, M.S., Hart, W.E., Schimel, B.D., and Giunta, A.A., "A Review of the Dakota Toolkit, Multilevel Parallelism for Complex PDE Simulations on TeraFLOP Computers," abstract presented in the Optimization in Engineering Minisymposium at the SIAM Annual Meeting, Rio Grande, Puerto Rico, July 10-14, 2000.

● Romero, V.J., Painton, L.A., and Eldred, M.S., "Optimization Under Uncertainty: Shifting of Maximum Vulnerability Point Due to Uncertain Failure Thresholds," 1997 INFORMS Spring Meeting, San Diego, CA, May 1997.

● Eldred, M.S., Outka, D.E., and Bohnhoff, W.J., "Optimization of Complex Engineering Simulations with the DAKOTA Toolkit," abstract in Proceedings of the First Biennial Tri-Laboratory Engineering Conference on Computational Modeling, Pleasanton, CA, Oct. 31-Nov. 2, 1995.

SAND Reports

● Adams, B.M., Bichon, B.J., Carnes, B., Copps, K.D., Eldred, M.S., Hopkins, M.H., Neckels, D.C., Notz, P.K., Subia, S.R., and Wittwer, J.W., "Solution-Verified Reliability Analysis and Design of Bistable MEMS Using Error Estimation and Adaptivity," Sandia Technical Report SAND2006-6286, October 2006.

● Eldred, M.S., Brown, S.L., Adams, B.M., Dunlavy, D.M., Gay, D.


http://www.cs.sandia.gov/DAKOTA/papers/2006_MilestoneFY06_SAND.pdf

http://www.cs.sandia.gov/DAKOTA/papers/2006_MilestoneFY06_SAND.pdf


M., Swiler, L.P., Giunta, A.A., Hart, W.E., Watson, J.-P., Eddy, J.P., Griffin, J.D., Hough, P.D., Kolda, T.G., Martinez-Canales, M.L. and Williams, P.J., "DAKOTA, A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis: Version 4.0 Users Manual," Sandia Technical Report SAND2006-6337, October 2006.

● Eldred, M.S., Brown, S.L., Adams, B.M., Dunlavy, D.M., Gay, D.M., Swiler, L.P., Giunta, A.A., Hart, W.E., Watson, J.-P., Eddy, J.P., Griffin, J.D., Hough, P.D., Kolda, T.G., Martinez-Canales, M.L. and Williams, P.J., "DAKOTA, A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis: Version 4.0 Reference Manual," Sandia Technical Report SAND2006-4055, October 2006.

● Eldred, M.S., Brown, S.L., Adams, B.M., Dunlavy, D.M., Gay, D.M., Swiler, L.P., Giunta, A.A., Hart, W.E., Watson, J.-P., Eddy, J.P., Griffin, J.D., Hough, P.D., Kolda, T.G., Martinez-Canales, M.L. and Williams, P.J., "DAKOTA, A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis: Version 4.0 Developers Manual," Sandia Technical Report SAND2006-4056, October 2006.

● Swiler, L.P. and Wyss, G.D., "A User's Guide to Sandia's Latin Hypercube Sampling Software: LHS UNIX Library Standalone Version," Sandia Technical Report SAND2004-2439, July 2004.

● Eldred, M.S., Giunta, A.A., van Bloemen Waanders, B.G., Wojtkiewicz, S.F., Jr., Hart, W.E., and Alleva, M.P., "DAKOTA, A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis. Version 3.0 Users Manual." Sandia Technical Report SAND2001-3796, April 2002. Updated April 2003 (Version 3.1).

● Eldred, M.S., Giunta, A.A., van Bloemen Waanders, B.G., Wojtkiewicz, S.F., Jr., Hart, W.E., and Alleva, M.P., "DAKOTA, A


http://www.cs.sandia.gov/DAKOTA/papers/Users4.0.pdf




http://www.cs.sandia.gov/DAKOTA/papers/Reference4.0.pdf




http://www.cs.sandia.gov/DAKOTA/papers/Developers4.0.pdf




http://www.cs.sandia.gov/DAKOTA/papers/LhsManual2004.pdf











Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis. Version 3.0 Reference Manual." Sandia Technical Report SAND2001-3515, April 2002. Updated April 2003 (Version 3.1), July 2004 (Version 3.2), and December 2004 (Version 3.3).

● Eldred, M.S., Giunta, A.A., van Bloemen Waanders, B.G., Wojtkiewicz, S.F., Jr., Hart, W.E., and Alleva, M.P., "DAKOTA, A Multilevel Parallel Object-Oriented Framework for Design Optimization, Parameter Estimation, Uncertainty Quantification, and Sensitivity Analysis. Version 3.0 Developers Manual." Sandia Technical Report SAND2001-3514, April 2002. Updated April 2003 (Version 3.1), July 2004 (Version 3.2), and December 2004 (Version 3.3).

● van Bloemen Waanders, B., Bartlett, R., Long, K., Boggs, P., and Salinger, A., "Large Scale Non-Linear Programming for PDE Constrained Optimization," Sandia Technical Report SAND2002-3198, October 2002.

● Gardner, D.R., and Vaughan, C.T., "The Optimization of a Shaped-Charge Design Using Parallel Computers," Sandia Technical Report SAND99-2953, November 1999.

● Hobbs, M.L., Erickson, K.L., and Chu, T.Y., "Modeling Decomposition of Unconfined Polyurethane Foam," Sandia Technical Report SAND99-2758, November 1999.

● Eisler, G.R., and Veers, P.S., "Parameter Optimization Applied to Use of Adaptive Blades on a Variable Speed Wind Turbine," Sandia Technical Report SAND98-2668, December 1998.

● McGee, B.C., Hobbs, M.L., and Baer, M.R., "Exponential 6 Parameterization for the JCZ3-EOS," Sandia Technical Report SAND98-1191, July 1998.

● Eldred, M.S., "Optimization Strategies for Complex Engineering Applications," Sandia Technical Report SAND98-0340, February

















http://www.cs.sandia.gov/DAKOTA/papers/pdeco_ldrd_finalreport.pdf

http://www.cs.sandia.gov/DAKOTA/papers/pdeco_ldrd_finalreport.pdf

http://infoserve.sandia.gov/sand_doc/1999/992953.pdf











1998.

● Zimmerman, D.C., "Genetic Algorithms for Navigating Expensive and Complex Design Spaces," Final Report for Sandia National Laboratories contract AO-7736 CA 02 (year 2), Sept. 1996.

● Hart, W.E., "Evolutionary Pattern Search Algorithms," Sandia Technical Report SAND95-2293, October 1995.

● Zimmerman, D.C., "Genetic Algorithms for Navigating Expensive and Complex Design Spaces," Final Report for Sandia National Laboratories contract AO-7736 (year 1), Sept. 1995.

● Meza, J.C., and Plantenga, T.D., "Optimal Control of a CVD Reactor for Prescribed Temperature Behavior," Sandia Technical Report SAND95-8224, April 1995.

● Moen, C.D., Spence, P.A., and Meza, J.C., "Optimal Heat Transfer Design of Chemical Vapor Deposition Reactors," Sandia Technical Report SAND95-8223, April 1995.

● Meza, J.C., "OPT++: An Object-Oriented Class Library for Nonlinear Optimization," Sandia Technical Report SAND94-8225, March 1994.

Back to top of page || Questions and Comments || Acknowledgment and Disclaimer

Last Updated: February 3, 2004

[email protected]


http://www.cs.sandia.gov/~wehart/papers.html





http://www.sandia.gov/feedback.htm






April 17, 2001

Seattle, WA

PRELIMINARIES

Achille Messac called the meeting to order at 7:30 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Renaud represented Batill, Blair, Bounajem, Canfield, Chen, Cramer, DeLaurentis, Finckenor, Gurdal, Guruswamy, Khot, Long, Purcell, Smith, Stephens, Striz, Zang, Schweiger, Suleman, Toropov, Krishnamachari, Wujek. Other guests were Eldred (Sandia), Thomas (Altair), Batayneh (RPI), Ismail-Yahara (RPI), Chris Horton (AIAA). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., names, email addresss, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through the Chair in order to maintain a consistent and up to date database. Changes will be passed on to Anthony Giunta for inclusion on the web site.

PRESENTATION BY Dr. Zang

Dr. Zang discussed the possibility of application of knowledge management to MDO.


MA &O 2002

Dan DeLaurentis discussed the progress on organizing this conference, which will be held at Grand Hyatt Atlanta from 4 - 6 Sept. 2002. The deadline for the abstracts is 11 January 2002. The General co-chairs for this conference are Farrokh Mistree and Dan Schrage from GT. The Technical co-chairs for this conference are Dan DeLaurentis from GT School of AE and Pradeep Raj from Lockheed Martin Aeronautics Company. The theme of the conference will be System Affordability.

SDM 2001 meeting at Seattle.

Fred Stritz represented MDO TC at this conference. He discussed the number of sessions organized and



the number of papers presented at each session on different topics. Zafer Gurdal will represent MDO TC at the SDM long range planning committee. He mentioned that there was discussion about MDO TC representative chairing the SDM conference sometimes in the near future.

SDM meeting at Denver, April. 2001

Srinivas will represent the MDO TC at this Conference.


Bob Canfield discussed the activities of this subcommittee regarding membership upgrades, best paper award, MDO TC award etc.

Application Subcommittee (Purcel)

The new subcommittee chair will be Tim Purcell. He outlined the future planned activities of the committee

Publication Subcommittee (Striz) Stritz proposed a possible subcommittee effort to make the management aware of the benefits of MDO in different companies.

Tony Giunta will take over the job of maintenance of MDO TC web page from Mike Eldred.

MEETING SCHEDULE

NEXT MEETING: Monday, January 14, 2002 at Reno, NV, 7:00-11:00 PM.


Respectfully submitted on 12/12/2001Narendra Khot AFRL/VASD 2210 Eighth Street Wright -Patterson AFB ,OH, 45433-7531 Tel: (937) 255-8474 Fax: (937) 656-4945 Email:Narendra [email protected]





Last Updated: 13 December 2001






January 9, 2001

Reno, NV

PRELIMINARIES

Chairman Achille Messac called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Messac, Batill, Baysal, Cramer, Eldred, Finckenor, Gage, Grossman, Guruswamy, Khot, Striz, Stephens, Balabanov, DeLaurentis was represented by Schrage, Krishnamachari, Kodiyalam. Other guests were Zang (NASA), Reuther (NASA), Ali de-Jongh (AIAA staff), Emily Springer (AIAA staff), Pete Wells (Boeing). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Achille in order to maintain a consistent and up to date database. Changes will be passed on to Mike Eldred for inclusion on the web site.

AIAA Activity

Ali de-Jongh from AIAA staff talked about the congressional visit in March 2001 by the members of the TC. Achille will attend this event. Emily Springer from AIAA distributed T-shirts to the members who did not get them during the last meeting. She talked about the newly designed AIAA web page. She also mentioned that AIAA has started a three year program on selecting historical sights. Last year five sights were selected. If any one wants a particular sight to be considered for this selection, a special form has to be filled in and submitted to AIAA.

Long Beach MDO-TC (September 7, 00) MEETING MINUTES REVIEWED AND APPROVED

MA &O 2002 STATUS REPORT

Dan Schrage discussed in detail the current status of the Georgia Tech Planning Team's effort in organizing this conference. The conference will be held at Grand Hyatt Atlanta located at Buckhead , Atlanta's most prestigious and fashionable area. The theme of the conference will be MA&O for Systems Affordability. The general co-chairs for this conference will be Farrokh Mistree and Dan Schrage. The technical programs co-chair will be Dan DeLaurentis from GT School of AE. The



industrial co-technical chair will be announced soon. AIAA will send out announcement for the papers in July 2001.

PRESENTATION BY KRISHNAMACHARI

Dr Krishnamachari from Boeing Corporation made a short presentation on the Multidisciplinary Optimization problems in Air Traffic Management. He will be presenting a paper on this topic during SDM 2001 conference at Seattle.

CPSRS (Collection of Preferred Space Related Standards)

Pete Wells made a short presentation on the development of CPSRS. The present SPACE standards are too many and outdated. He was seeking volunteers to participate in this endeavor. If any one of the members are interested he should visit www.aiaa.org/cpsrs for more information. The budget for this effort is of the order of $1 million. At present 295 reviewers, 15 countries, 155 organizations and 20 universities are going to participate in this effort.

Presentation by Daniel Schrage

Prof. Schrage gave an overview of the Center for Aerospace Systems Analysis (CASA) at Georgia Institute of Technology. This center combines research and education of School of Aerospace Sciences. The full time faculty is 32 with 250 graduate students and 250 undergraduate students. The function and the details about this center can be found at their web page www.asdl.gatech.edu.

SDM meeting at Seattle, April 2001

Fred Stritz represented the TC at this Conference. He mentioned that there will be 7 sessions on MDO at this meeting. Out of 43 abstracts submitted, 42 were accepted.The total number of sessions would be 7. There will be 2 panel discussions: 1)Use of Optimization Tools in Practical Design chaired by Dr. Balabanov, and 2) Multidisciplinary Systems Optimization Using Simulation Models chaired by Dr. Krishnamachari. MDO TC at 2002 SDM will be represented by Srinivas Kodiyalam.

Future Committee Activity

Four Subcommittees: Applications, Awards, Education and Publications will be headed by Tim Purcell, Bob Canfield, Kemper Lewis and Alfred Striz respectively.

MEETING SCHEDULE

NEXT MEETING: Tuesday April 17 2000 at Seattle WA , 7:00 PM-10:00 PM.


http://www.asdl.gatech.edu/






Last Updated: May 16, 2001





MDO TC Meeting Minutes (4 April 2000)


September 7, 2000

Long Beach , CA

PRELIMINARIES

Achille Messac called the meeting to order at 7:30 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Messac, Anderson, Batill, Baysal, Chen, Eldred, Finckenor, Giesing, Grossman, Gurdal, Guruswamy, Khot, Lewis, Purcell, Stephens, Striz, Rodriguez, Suleman, Balabanov, DeLaurentis, Tappeta, Wujek, Ex-officio chairs Barthelemy and Perez represented Renaud. Other guests were Morelle, Blair, Mistree, Grandhi, Giunta, Simpson, Toropov, Schulbach (NASA), Chris Horton (AIAA). Thanks to these and their sponsoring organizations for attending.



PRESENTATION BY DR. CATHY SCHULBACH

Dr.Schulbach who is the project manager at Ames Research Center, made a presentation on the High Performance Computing and Communications Program started by NASA advancing the frontiers of science and technology on earth and in space .NASA's HPCC Program is a critical element of the Federal Information Technology Research and Development effort. The five project areas included in this program are Computational Aerospace Sciences (CAS), Earth and Space Sciences (ESS), Remote Exploration and Experimentation (REE), Learning Technologies (LT) and NASA Research and Education Network (NREN). The detailed information on this NASA's HPCC effort may be found on the following web pages: http://www.hpcc.nasa.gov; http://cas.nasa.gov; http://www.aero-space.nasa.gov and http://nais.msfc.nasa.gov.

AIAA Activity

Chris Horton from AIAA announced that the MAO 2002 conference will be held at Hyatt located at Buckhed, about 40 miles from Atlanta. Mistree who is the technical chair of the conference described in detail the amenities available at this hotel and the city.

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.sept00.html (1 of 3)12/29/2006 12:29:05 PM


Atlanta MDO-TC (April 4, 00) MEETING MINUTES REVIEWED AND APPROVED

MA &O 2000

General Chairman and Technical Chairman for this conference are Jean-Francois Barthelemy and Kumar Bhatia respectively. Barthelemy made a detailed presentation on organizing the conference and compared various aspects with the previous MA & O conferences.The theme of the conference was Multidisciplinary Design--Adding Value. The total number of papers submitted were 264 , and 249 papers were accepted for presentation with the acceptance ratio of 94%. The number of work in progress papers were 22. Most of the papers were accepted based on the single review. Second review was initiated only for papers on the bubble. This was the first AIAA conference where AIAA Conference Management System was used to receive the abstracts and review them.Barthelemy mentioned that the foreign participation in MA &O conferences has been steadily increasing while there has been steady decline of USA participants.

SDM 2001 meeting at Seattle, WA. April 2001

Fred Stritz will represent MDO TC at SDM 2001. He attended SDM 2001 organization committee meeting and mentioned that there will be eight sessions available for MDO related papers including a panel session . 25 abstracts have been received until now. Gurdal who represented MDO TC at the long term planning committee of SDM, mentioned that there was discussion about MDO TC representative chairing the SDM conference sometimes in near future. Stritz proposed to have MA &O conference as add-on specialty conference similar to Dynamic specialty conference. Barthelemy proposed that we should be fully participant and not add-on to the SDM conference. These proposals were put to vote . Majority voted for fully participation in the conference. In order to pursue this matter further, ad-hoc committee was formed with Gurdal, Grossman, Grandhi, Striz and Messac as the members.

Awards Committee (Kolonay)

The MDO Technical Award at the MA &O 2000 conference was awarded to Dr V. B. Venkayya who is the founding member of this committee.

AIAA Design Competition (Anderson)

Anderson mentioned that AIAA has invited MDO to propose student competition in MDO and made a presentation on his tentative proposal. The abstract on the proposal was required to be submitted to AIAA by 10th Oct. 2000. Grossman mentioned that there are already number of undergraduate student



competitions and graduate students do not have enough time to participate in these competitions. Committee decided not to respond to this proposal. Instead it was proposed that TC could hold best paper competition for the undergraduate student participants. The committee was formed with Anderson, Grossman, Kemper, Long and Guruswamy as the members to study the proposal.

Messac discussed the results of his survey.

MEETING SCHEDULE

NEXT MEETING: Reno, Nevada ( date will be announced )





Last Updated: December 21, 2000







April 4, 2000

Atlanta, GA

PRELIMINARIES

Achille Messac called the meeting to order at 7:30 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Renaud, Giesing, Striz, Khot, Balabanov, Baysal, Bhatia, Chen, Finckenor, Gage, Gilje, Grossman, Gurdal, Guruswamy, Kolonay, Lewis, Nagendra, Padula, Stephens, Schrage was represented by DeLaurentis. Other guests were Martinez, Tappeta (GE), Walsh (NASA), Housner (NASA), O'Leary (AIAA), Chris Horton (AIAA), Thanks to these and their sponsoring organizations for attending.



PRESENTATION BY DR. JERRY HOUSNER

Dr.Housner made a presentation on the Intelligent Synthesis Environment (ISE) Initiative started by NASA aimed at making substantial progress toward fulfilling the NASA Administrator's vision for revolutionizing next generation science and engineering capabilities. This initiative will achieve this vision by developing revolutionary ISE related technologies,engineering practices and coordinating related on -going NASA activities, industry activities, other government agency initiatives and university research. The set of view graphs used by Jerry may be found at the web http://ise.larc.nasa.gov.

AIAA Activity

Steve O'Leary from AIAA staff discussed the plan for electronic submission of abstract, reviewing process for the future AIAA meetings. 42nd AIAA/ASME SDM conference will be the first major AIAA conference where electronic submission will be strongly encouraged. Authors having trouble submitting abstracts electronically may e-mail the abstracts . Chris Horton from AIAA discussed the possible locations for holding MAO 2002 conference. He suggested that a city like Buckhed closer to Atlanta would be more suitable than Atlanta because of the size of the hotel and the available room



rates. The MDO TC and MAO Organizing committee members will have to decide on the location soon.


MA &O 2000

Sharon Padula discussed the progress of this conference which will be held at Long Beach , Westin Hotel from 6 - 8 Sept. 2000. Sept. 4 is the Labor Day. General Chairman and Technical Chairman for this conference are Jean-Francois Barthelemy and Kumar Bhatia respectively. The theme of the conference is Multidisciplinary Design--Adding Value. The manuscripts are due by 23rd of June. AIAA's new Conference Management System was used to receive the abstracts, evaluation etc. The total number of papers submitted were 264 , and 249 papers were accepted for presentation The number of work in progress papers were 22. Most of the papers were accepted based on the single review. Second review was initiated only for papers on the bubble. The upgrade of the AIAA Conference Management System is in progress and will facilitate future conferences.

SDM 2001 meeting at Seattle, WA. April 2001

Fred Stritz will represent MDO TC at SDM 2001. He attended SDM 2001 organization committee meeting and mentioned that there will eight sessions available for MDO related papers. Zafer Gurdal will represent MDO TC at the SDM organization committee. He mentioned that there was discussion about MDO TC representative chairing the SDM conference sometimes in near future.

SDM meeting at Atlanta, April. 2000

Khot who represented the TC at this Conference mentioned that there will be 7 sessions on MDO at this meeting. 51 abstracts were received . Out of these 46 papers were included in the seven sessions and 2 papers were rejected. The authors of the 3 papers were suggested to present their results at the postal session. These papers otherwise would have been rejected.

Awards Committee (Kolonay)

Ray Kolonay mentioned that there were two nominees for MDO Technical Award and he is in the process of selecting the committee. The final decision will be taken after consulting with the full TC. Future guide lines for this award will be formulated by the TC during the next meeting.

Application Subcommittee (Gilje)



Gilje informed the TC that this meeting would be his last, and that a replacement for his sub-committee would be needed.

Publication Subcommittee (Messac)

Messac had written the yearly Aerospace America article for the past four years. Hovever, since he will become Chair of the TC, this effort will need to be led by another member.

Messac presented a plaque to John Renaud for his excellent work as the chairman of the MDO TC committee.

MEETING SCHEDULE

NEXT MEETING: Thursday September 7, 2000 at Long Beach CA, 7:00-10:00 PM.












January 11, 2000

Reno, NV

PRELIMINARIES

New Chairman-Elect Achille Messac called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Khot, Balabanov, Baysal, Chen, Cramer, Eldred, Finckenor, Gage, Gilje, Grossman, Gurdal, Guruswamy, Human, Messac, Stephens, Schrage was represented by DeLaurentis, Padula was represented by Walsh. Other guests were Giunta (Sandia), Emily Springer (AIAA), Dave Culpepper (AIAA), Chris Horton (AIAA), Chris Pestak (Analex Corp.), Pete Wells (Boeing), Tom Weeks (J. of Aircraft). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Achille/John in order to maintain a consistent and up to date database. Changes will be passed on to Mike Eldred for inclusion on the web site.

AIAA Activity

Emily Springer from AIAA staff distributed T-shirts to all members who were present. Dave Culpepper, V.P. of TAC, discussed the possibility of TCs getting about $700.00 for the TC projects. Funding will be coming from the voluntary contributions from the AIAA members. At present this project is approved for a period of three years. At the end of three years if the concept is found worthwhile, it will be extended. TAC will then put the required expenditure in their budget. He also mentioned that the TC members will be called upon to help review the papers in the future that will be sent in .pdf format. Chris Horton from AIAA mentioned that there is no possibility of holding MAO 2002 conference in Atlanta because of the restricted number of attendees and the anticipated expenditure (That statement has evolved since then). The MDO TC would probably have to select some other city. In this respect Chris Pestak from Analex, Clevaland made a pitch for holding the MAO2002 at Renaissance Cleveland Hotel, which satisfies most of the MAO conference requirements. The MDO TC and MAO Organizing committee members will have to decide on the location soon.

St. Louis MDO-TC (September 12, 99) MEETING MINUTES REVIEWED AND APPROVED



PRESENTATION BY DR. TOM WEEKS, EDITOR OF JOURNAL OF AIRCRAFT

Dr. Weeks discussed the scope of the Journal . He wants the TC's help in identifying and motivating reviewers. The TC generated reviewer list would be useful. Also, authors who present papers at the conferences need to be identified and motivated to submit the papers to the journals. He also mentioned that there will be a special issue in year 2003 in celebration of the Wright Brothers first flight a hundred years ago. He is seeking suggestions on history papers, topics and possible authors.

MA &O 2000

Joanne Walsh discussed the progress. This conference will be held at Long Beach Westin Hotel from 6 - 8 Sept 2000. Sept 4 is the Labor Day. Session will be finalized by 3rd of March and the manuscripts are due by 23rd of June. Bernie Crossman mentioned about the Meeting held at Air Vehicle Directorate regarding the Energy Methods in Design methodology. Khot will get in contact with Dr. David Moorhouse to check whether he would propose a special session at MA& O conference.

CPRS (Collection of Preferred Space Related Standards)

Pete Wells made a short presentation on the development of CPRS. The present SPACE standards are too many and outdated. He was seeking volunteers to participate in this endeavor. If any one of the members are interested he should visit www.aiaa.org for more information.

Fall Internet Meeting

Mike Eldred briefed on the fall internet meeting. Some members had problems logging into the system or were reluctant to visit the OneList commercial Web site from work computers. Ideas to improve future internet meetings and the MAO2002 contingency plan were discussed.

SDM meeting at Atlanta, April. 2000

Khot who represented the TC at this Conference mentioned that there will be 7 sessions on MDO at this meeting. The total number of papers would be 46 . The conference will be for a period of four days with sessions on Thursday afternoon.

Design Engineering TC

Vladimir Balabanov who is the member of this committee suggested to have joint meeting in order to make this committee aware of the MAO activity. Gurdal and Messac volunteered to attend Design Engineering TC meeting at Atlanta in April 2000 during SDM conference.



Associate Fellows

Following members were selected to be Associate Fellows of AIAA: Jean-Francois Barthelemey, Todd Beltracchi, Christina Bloebaum, Robert Canfield, Farrokh Mistree and John Renaud. Congratulations to these present and past members.

Future Committee Activity

The new chairman of the committee Achille Messac initiated a discussion on the future projects the TC should consider initiating in order to make MDO activity known to the AIAA community. Peter Gage volunteered to write down the points of discussion. He would send the list of items to all the members for their comments.

MEETING SCHEDULE

NEXT MEETING: Tuesday April 4 2000 at Atlanta Ga., 7:00 PM-10:00 PM.












April 12, 1999

St Louis, MO

PRELIMINARIES

Chairman John Renaud called the meeting to order at 7:30 PM. Nelson Wolf who represented Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Renaud, Stritz, Bhatia, Eastep, Gilje, Iqbal, Kodiyalam,Kolonay, Lewis, Livne, Long, Messac,Orozco, Padula, Purcell,Balabanov . Other guests were Wolf,Grandhi,Zailar,Mavris,Venkayya,Guruswamy. Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through John in order to maintain a consistent and up to date database. Changes will be passed on to Mike Eldred for inclusion on the website.

MA&O CONFERENCE 2000

Jean-Francois Barthelemy discussed the details of this meeting which will be held at Long Beach, CA, Sept 6 - 8.He mentioned that the increased focus will be on applications, pursue third party vendors, invite state-of the art papers, special presentations etc. He also discussed the paper review process.

SDM CONFERENCE 99

Srinivas Kodiyalam who represented MDO TC at this conference , mentioned that he had received a total of 33 abstracts. Each paper was reviewed by 3 reviewers. 32 papers were accepted and were divided into 6 sessions. Narendra Khot will be TC representative at SDM 2000 conference.

WAC 99

John mentioned about the call for papers at this conference.

AEROSPACE SCIENCES 2000



Achille Messac represented TC at this conference. He has sent E-mail requests for the papers and he is looking forward for good papers.

ISSMO 1999.

Kemper Lewis gave the details of the conference which will take place in May 1999. The registration fee for this conference will be $260.00. Tennis and golf tournaments are planned after the conference.

MA&O Conference 2002.

John Renaud mentioned that he will send out the call for nominations for chairs on private MDO E-mail, to organize this conference. The procedure was voted acceptable by the members.

TC OPERATION

John Renaud showed TC Operations and TC structure chart. He asked for publication lead to replace Eli Livne. He mentioned that replacements for new officers and co-chairs will be elected by this summer.

BREAKOUT SESSION

The members were asked to break into small groups to come up with the action items for MDO. Following were the tentative action items suggested by the groups:

● Educate people about optimization and MDO.● Survey AIAA members about MDO use.● Educate starting with H. S., Colleges, Work Places, Managers etc.● Optimization demos to show its usefulness.● How to Educate teachers to teach the subject.● MDO articles; how is it helpful.● Sponser design contests.● MDO committee publish benefits -- How much did it improve product.● Put together Web Page on MDO benefits.● Action items : Put some abstracts on the Web-- John will check this out.● Out reach to other TC's about MDO value

John will consider these and suggest a three year plan . He will try to have it ready for the next TC Chairman.



SUBCOMMITTEE REPORTS

Awards Subcommittee (Kolonay) Ray Kolonay had proposal for modifications to MDO award selection procedures. Move to accept the proposal for selecting the MDO award was made. This was passed by votes. He wanted suggestions for AIAA Fellow nominations by the MDO committee.

Application Subcommittee (Gilje) Ron Gilje would put the report on the Web.

Education (Messac) Achille Messac discussed the progress of MDO article in Aerospace America and MDO Course.

Benchmarking Subcommittee (Lewis) Kemper Lewis has set up MDO TEST SUITE on the web (http://fmad-www.larc.nasa.gov/mdob/MDOB/mod.test/index.html). He is planning to add more bench mark problems on the Web.

Strategic Committee John Renaud mentioned that he wants someone to represent MDO TC at this committee.

MEETING SCHEDULE

NEXT MEETING: World Aviation Congress Oct 1999 in San Franscisco.







Last Updated: October 13, 1999






MINUTES OF THE MDO TC MEETINGJanuary 11, 1999

Reno, NV

PRELIMINARIES

Chairman John Renaud called the meeting to order at 7:00 PM. Nelson Wolf who represented Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Renaud, Stritz, Balling, Bayard, Baysal, Eldred, Finckenor, Gage, Ghattas, Gilje, Grossman, Gurdal, Human, Kolonay, Livne, Majumdar, Messac, Mosher, Santangelo, Schrage was represented by DeLaurentis, Lewis was represented by Bloebaum (WCSMO3). Other guests were Riddie, LeGresley, Ovens, Townsend, Sobieski, Artcliff, Emily Davies (AIAA), Tom Weeks (J. of Aircraft). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through John in order to maintain a consistent and up to date database. Changes will be passed on to Mike Eldred for inclusion on the website.

PRESENTATION: Optimization Activities at Sandia National Laboratories

Mike Eldred made a presentation on the activities of Sandia National Labs in the area of optimization software and on the development of parallel optimization methods for the Accelerated Strategic Computing Initiative (ASCI). He overviewed the capabilities of the DAKOTA software under development at Sandia and highlighted the exploitation of multiple levels of parallelism which results in near-linear scaling on massively parallel computers containing thousands of processors.

St. Louis MDO-TC (September 1, 98) MEETING MINUTES REVIEWED AND APPROVED

PRESENTATION BY AIAA STAFF

Emily Davies updated the TC on AIAA activities. AIAA has now over 30,000 members, 19 lifetime members. This year 30 fellows were selected.




Dr. Weeks discussed the scope of the Journal . He wants the TC's help in identifying and motivating reviewers. TC generated reviewer list would be useful. Also, authors who present papers at the conferences need to be identified and motivated to submit the papers to the journals.

WHITE PAPER

John discussed action on the white paper. White paper would be on the web in March 1999. He would like to receive comments.

MA &O 2000

John stated 2000 MA & O meeting-site selection still is on going. He suggested a motion that 3 years prior to the future meetings Conference Chairman be selected. Fred Stritz seconded the motion. Motion was passed.

WCSM03

Bloebaum discussed the progress of the conference. The details of the conference can be found on the web http://www.eng.buffalo.edu/Research/MODEL/wcsmo3/.

CANDIDATES FOR AIAA DIRECTOR-AT-LARGE

Dr. Richard Antcliff and Mr. Anthony Gross, who are standing for election for the positions of director-at-large, distributed handouts giving the summary of their education and experience.


Awards Subcommittee (Ghattas)Omar discussed the procedure he is following on selection of best paper from MAO98. The award will be presented at MAO2K.He discussed the roll of TC in nomination for AIAA fellow award and wants the members to send names of the potential nominees for AIAA fellow. The voting will take place during the committee meeting at SDM99. Congratulations to Rudi Yurkovich past TC member who was selected fellow this year. Ray Kolonay will be taking over the chairmanship of this committee after SDM 99 conference

SDM meeting at St Louis April. 1999John mentioned that there will be 6 sessions on MDO at this meeting. Srinivas Kodiyalam was the TC representative at this conference. Narendra Khot will be the TC representative at SDM 2000 conference.

Publications Subcommittee (Livne)


http://www.eng.buffalo.edu/Research/MODEL/wcsmo3/


Eli Livne discussed the progress on the special issue of Journal of Aircraft on MDO in Aerospace Context.

Membership(Balling)He discussed the status of membership. The maximum number of members is limited to 35.

Benchmarking Subcommittee (Lewis)Kemper Lewis has taken over as chairman of the subcommittee. The subcommittee report was distributed by Kemper. The details are posted on the MDO TC AIAA web page. The test suite of MDO problems developed at the MDO Branch of NASA Langley Research Center can be accessed on the web under URL (http://fmad-www.larc.nasa.gov/mdob/mdo.test/). Kemper mentioned that he has been made aware by other researchers and graduate students that there are errors in some of the problems . Some effort is going into correcting the existing problems.

World Aviation Congress 99

John mentioned that he is organizing two sessions on MDA&O and looking to get ten papers.The abstracts for this conference are due by 15 Jan 1999. He would like to know if anyone would be interested in presenting a paper.

Application Subcommittee(Gilje)

Gilje outlined the objectives of the committee and the planned contents of the report. He had some problems hearing from some of the committee members . The committee progress has been posted on the web page.

Liaison Subcommittee (Grossman)

Grossman will send web site information to Mike Eldred.

MEETING SCHEDULE

NEXT MEETING: Monday April 12 1999 at St. Louis, Mo , 7:00 PM-11:00 PM.


Respectfully submitted on 3/12/99Narendra KhotAFRL/VASD2210 Eighth StreetWright -Patterson AFB ,OH, 45433-7531


http://fmad-www.larc.nasa.gov/mdob/mdo.test/


Tel: (937) 255-8474Fax: (937) 656-4945Email:Narendra [email protected]



Last Updated: April 8, 1999






MINUTES OF THE MDO TC MEETINGSEPTEMBER 1, 1998

St. Louis, MO

PRELIMINARIES

Chairman John Renaud called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Renaud, Giesing, Striz, Khot, Eastep, Majumdar, Kolonay, Suleman, Messac, Morris, Padula, Petiau, Grossman, Gage, Orozco, Finckenor, Smith, Gilje, Eldred, Gurdal, Stephens, Nagendra, Kodiyalam, Karpel and Schrage. Other guests were Grandhi (Past Member), Canfield (Past Member), Brewster King (AIAA), Chenevey (AIAA), Cramer (Boeing), Shimko (Universal Analytics), Platnick (Universal Analytics), Yurkovich (Boeing), Stettner (NASA). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Renaud, in order to maintain a consistent database. Changes will be passed on to Mike Eldred for inclusion on the website

Long Beach MDO-TC (April 98) MEETING MINUTES REVIEWED AND APPROVED

PRESENTATION: ASTROS OPTIMIZATION SOFTWARE

Joseph Platnick made a presentation on the capabilities of ASTROS (Automated STRuctural Optimization System) program.This is a finite element-based software that has been designed to assist, to the maximum practical extent, in the preliminary design of aerospace structures. ASTROS supports the multidisciplinary nature of design by implementing the disciplines in separate modules and by the use of MAPOL (Matrix Analysis Problem Oriented Language), concerning a high level language, to direct the interaction among the modules. This software ASTROS Version 20.1 at presented is marketed by Universal Analytics, Inc.Torrance, California .


Brewster King and Cathy Chenevey discussed the possible locations and the problems in selecting the Hotel for MA&O 2000 meeting.



MA&O CONFERENCE - 1998

Bob Canfield (Technical Chair) discussed his experience in the organization of the MA&O 1998 conference. This was the first AIAA conference where electronic abstracts submission was used. He mentioned that 305 abstracts were received , 279 papers were accepted and 24 papers were withdrawn. There were 50 papers from Europe , 23 work-in -progress . Ramana Grandhi (General Chair)mentioned that MA&O 2000 will be organized by Drs. Jean -Francois Barthelemy and Kumar Bhatia, since they received maximum votes from the committee members. This proposal was put to motion and committee approved it. Grandhi recommended that M&O 2002 be organized by Profs. F. Mistree and D. Schrage which was the second team who had shown interest in organizing MA&O 2000. However,the committee recommended that this should be finalized by the future TC and not the present one.

MDO WHITE PAPER PROPOSAL FOR 1998

Joe Giesing discussed the papers to be presented in the two special sessions on Industry MDO Applications and Needs. There would be 11 papers and one summary paper which was authored by Giesing and Barthelemy. All the papers presented in the two sessions will constitute AIAA MDO TC white paper. The summary paper will be put on the web. Renaud suggested that TC members should submit their comments on all the papers to Giesing so that he can get them finalized. Renaud will get in touch with Livne to check whether the summary paper can be published in the Journal of Aircraft in his special issue.


Awards Subcommittee (Ghattas)Omar Ghattas did not attend the meeting . Renaud had received E-MAIL from Omar discussing the progress of his committee.The TC- supported AIAA Fellow nomination package for Chris Borland. The 1998 MDO Award will be presented to Prof. Raphael Haftka.The 6th MAO Symposium Best Paper Award will be presented to :V. Balabanov, M Kaufma, D Knill, D. Haim, O. Golovidov, a. Giunta, R. Haftka, B.Grossman, W, Mason, and L. Watson, for their paper titled , "Dependence of Optimal Structural Weight on Aerodynamic Shape for a High Speed Civil Transport", AIAA Paper 96-4046.

Conference Support Subcommittee (Chair Vacant at Present)Srinivas Kodiyalam , TC representative at 1999 SDM conference mentioned that he had received 33 papers and their will be 6 sessions at the conference on MDO related topics. Narendra Khot has agreed to be the TC representative at 2000 SDM conference.

ISSMO (Padula)Padula discussed the progress on organizing this conference, which will be held at Niagara Falls on May 17-21, 1999. The abstracts for this conference are due by 30 Oct 1998.



Publication Subcommittee (Livne)

Livne was not present. Renaud received E-MAIL discussing the progress of his subcommittee. 28 papers for the special issue of J of Aircraft have been reviewed and are sent to publication. The estimated date of publication is Jan/Feb 1999. Messac will be writing a MDO review article for the Aerospace America.

Education Subcommittee (Renaud)John Renaud put the motion to the floor to approve the new charter which was put on the web. The motion was passed.

Internet Subcommittee (Eldred)Mike talked about the updates to the web page . He mentioned that new member roster is on the web. Mike will get in contact with new members to get their Bio data.

Liaison Subcommittee (Grossman)Bernie discussed the need for members to represent MDO TC at other TC committees.

Application Subcommittee (Gilje)

Gilje outlined the progress of the committee .

Benchmarking Subcommittee (Striz)Fred Striz discussed the progress of his subcommittee on selecting the benchmark problems.

ACTION ITEM:All subcommittee chairs are requested to put their progress reports on the web so that the interested members can get the details..

MEETING SCHEDULE

NEXT MEETING: Jan 11, Monday (7:00PM-11:00PM)at Reno, NV.

Renaud presented a plaque to Jean-Francois Barthelemy for his excellent work as the chairman of the MDO TC committee. The plaque was received by Padula since Barthelemy was not present

Respectfully submitted on 10/13/98,Narendra KhotAFRL/VASD2130 Eighth Street, Ste 1Wright-Patterson AFB. OH , 45433-7541Tel: (937)255-6992Fax: (937)255-3740



Business email: [email protected]









MINUTES OF THE MDO TC MEETINGApril 20, 1998

Long Beach, CA

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Barthelemy, Giesing, Eldred, Eastep, Briggs, Striz, Messac, Rais-Rohani, Canfield, Grandhi, Iqbal, Renaud, Majumdar, Haftka, Gurdal, Khot, Livne, Bhatia, Karpel, Bolognese, Human, Gilje Mazzaway and Rocha. Other guests were Sharon Padula, Jonathan Bishop, Pramod Rao Bangarpet ,Tom Zielen, Mike Long, Resende, Brewster King (AIAA). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Renaud, in order to maintain a consistent database. Changes will be passed on to Mike Eldred for inclusion on the website.

Reno MDO-TC (Jan 98) MEETING MINUTES REVIEWED AND APPROVED

PRESENTATION: OPTIMIZATION AT VMA ENGINEERING

Gary Vanderplatts made a presentation concerning the optimization services and software offered by VMA Engineering, Inc. He gave historical background on the developement of GENESIS program , its new analysis and design capabilities. He also mentioned that GENESIS PC version is now available He discussed the capabilities of the general purpose numerical optimization software DOC/DOT. The company has a web site (www.vma.com).


Brewester King discussed the possible locations and the problems in selecting the Hotel for MA&O 2000 meeting. The final decision will be made after the selection of the Chair for this conference.



http://www.vma.com/


Bob Canfield (Technical Chair) discussed in detail the progress in the organization of the conference. He mentioned that 257 papers have been accepted for presentation at the conference. The organization plan is on schedule. The detailed program will be published in June 1998 Aerospace America . Ramana Grandhi gave general overall picture of the conference mentioning the names of the keynote speakers and award luncheon speaker.


Joe Giesing discussed the proposed white paper. This paper will emphasize real-world problems, mainly taken from industrial projects. Nine draft papers have been received. The summary paper will be written by Giesing and Barthelemy. The invited papers will be presented in two sessions in St. Louis at the MA&O conference.


Awards Subcommittee (Ghattas)Omar Ghattas did not attend the meeting. The TC nomination for AIAA Fellow was discussed.

ACTION ITEM: Omar will send everyone emails with information on the nominees, and voting will be done electronically.

Conference Support Subcommittee (Chair Vacant at Present)Srinivas Kodiyalam will be TC representative at 1999 SDM conference. Narendra Khot has agreed to be the TC representative at 2000 SDM conference.

Publications Subcommittee (Livne)Eli Livne is working with Tom Weeks to put together a Journal of Aircraft issue on MDO. He mentioned that 13 papers have been received in final form and 6 papers are due within a month. It will be about two weeks before the work will be completed.

Education Subcommittee (Renaud)MDO TC committee charter has been placed on the web. Sharon Padula will take over the chairmanship of this committee since Renaud will be the new TC Chairman.

ACTION ITEM: Each subcommittee chair should review the subcommittee function discussion in the charter and send their suggestions and comments to Renaud. It is suggested that the chairman of the committees should consider holding meetings with their members using phone system to discuss the agenda and the progress.

Internet Subcommittee (Eldred)Mike talked about the updates to the web page, MDO related web sites and discussed the MDO Electronic Mailing Lists details as posted on the web.



Membership(Balling)Richard mentioned the names of the new members. He will get in contact with the new members and suggest that they get involved in the subcommittee tasks of their interest.

Application Subcommittee (Gilje)Gilje outlined the objectives of the committee. He mentioned that even though he has retired from TRW he would like to continue his work with the committee.

Benchmarking Subcommittee (Striz)Fred Striz discussed the progress of his subcommittee on selecting the benchmark problems.

MEETING SCHEDULE

NEXT MEETING: Sep 1, Tuesday (7:00PM-11:00PM) and Sep 3, Thursday (7:30PM-9:30 PM) at HYATT REGENCY UNION STATION, St. Louis, MO.

This was the last day for Jean-Francois as the chairman of the committee. All members present at the conference applauded before the meeting was adjourned at 11:15 hours.

Respectfully submitted on 5/13/98,Narendra KhotAFRL/VASD2130 Eighth Street, Ste 1Wright-Patterson AFB. OH , 45433-7541Tel: (937)255-6992Fax: (937)255-3740Business email: [email protected]










Reno, NV

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 7:00 PM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Barthelemy, Giesing, Renaud, Khot, Grandhi, Grossman, Livne, Majumdar, Gilje, Kolonay, Gage, Agrawal was represented by Unger, Balling, Canfield, Eastep, Fadel, Ghattas, Mazzawy, Peraire, Santangelo, Schrage was represented by Olds and Mosher. Other guests were Brewster King (AIAA), Ray Chatman (AIAA), Tom Weeks (J. of Aircraft), C. L. Bloebaum (ISSMO 99). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Jean-Francois, in order to maintain a consistent and up to date database. Changes will be passed on to Mike Eldred for inclusion on the website.

St. Louis MDO-TC (October 9, 97) MEETING MINUTES REVIEWED AND APPROVED


Brewster King discussed the opportunities the TC has to organize short courses (2-5 days) or tutorials ( 1 day) and the time table for implementation of these professional development courses. Ray Chatman discussed the scholarship programs developed by AIAA for undergraduate and graduate students. At present 20 undergraduate scholarships ($1000 each) and 5 graduate scholarships ($5000 each) are being awarded by AIAA. The information regarding this program has been sent to the universities. This program will be expanded as new endowments will be coming.


Dr. Weeks discussed the scope of the Journal and mentioned that he is primarily interested in papers on applied aircraft technology. He wants TC's help in identifying and motivating reviewers. Also, authors who present papers at the conferences need to be identified and motivated to submit the papers to the journals. He also mentioned that the page limit on survey papers has been removed. In addition, he



requested from the TC on how the publication process can be improved.

ACTION ITEM: Messac will follow-up on his idea of last year to ping the members for their preference as reviewer and communicate the information to the appropriate Associate Editors.


Bob Canfield (Technical Chair) discussed the format, the procedure he is going to follow to get the papers reviewed and the time table the super chairs should adhere to, in order to get the summary rating sheets back to him either by mail or internet by 28 Feb 1998. He mentioned that the MAO planning is on schedule up until now.

ACTION ITEM: Todd Mosher will get in contact with Bob Canfield to discuss the procedure followed by IEEE to receive the abstracts and check the status regarding acceptance of paper.

ACTION ITEM: Ghattas will provide Grandhi a point of contact in the Dept of Energy for inviting a speaker on the subject of High Speed Computing.


Joe Giesing discussed the proposed white paper. This paper will emphasize real-world problems, mainly taken from industrial projects. The format has been changed from a single report to a series of invited papers accompanied by a summary paper. The invited papers will be presented in two MA&O sessions. Papers will stress the use of MDO, bring up industrial needs, or point out areas where MDO could have been used to enhance the design process. Eight abstracts have been received, one is eminent and three are in progress. There will be two full sessions covering these invited papers and the summary paper.

ACTION ITEM: Joe Giesing and Jean-Francois Barthelemy will write the summary paper for presentation at the conference and subsequent editing by the TC.

MA &O 2000

1998 Chair/Technical Chair team will send a call for nomination between ASM and SDM to past and current TC members for proposals to organize the conference and select the sight. Between SDM and MDO Chair/Technical Chair will review and prioritize the proposals. The General Chair, Tech Chair and the location will be announced at 1998 MA&O 1998 meeting at St. Louis.

ACTION ITEM: Barthelemy will communicate to Grandhi the procedure for selection of the 2000 MA&O team.




Awards Subcommittee (Ghattas)Omar Ghattas discussed the roll of TC in nomination for AIAA fellows award. He will put on the web the procedure TC plans to follow. Nominations for MDO Award are due by Jan 20 1998.

Aerospace Sciences Meeting Jan. 1999Three sessions are planned at this meeting related to MDO.

Publications Subcommittee (Livne)Eli Livne discussed the progress on the special issue of Journal of Aircraft on MDO in Aerospace Context. There will be about 16-17 papers.

Education Subcommittee (Renaud)Renaud distributed draft copy of the charter for MDO TC operating plan. He will put this on the web. TC members should take a look at the draft and send their comments to Renaud.

Membership(Balling)16 new applicants have been received to become TC members. The recommendations for these applicants are as follows: 9 members, 3 associate members and 4 will be denied memberships at present.

Benchmarking Subcommittee (Striz)The subcommittee report was distributed by Renaud. At present the members of this committee are Haftka, Majumdar, Messac, Renaud and Morris. If any one is interested in becoming the member of this subcommittee should get in contact with Striz. The test suite of MDO problems developed at the MDO Branch of NASA Langley Research Center by Sharon Padula and her associates can be accessed on the web under URL (http://fmad-www.larc.nasa.gov/mdob/mdo.test/)

World Aviation Congress 98

Bill Bickard from Boeing Long Beach was looking for a TC member to organize a session on MDO taking into consideration needs of the practical designers. He needs information regarding the title of the papers etc. by May 1998. The information will be put on the web.

Dan Schrage will represent MDO TC at the Aircraft Technology Integration and Operation meeting at the WAC.

Application Subcommittee(Gilje)

Gilje outlined the objectives of the committee, the planned contents of the report and plans for publication of the report. The main objective of this committee is to communicate, educate internally and externally within the general public MDO concepts.


http://fmad-www.larc.nasa.gov/mdob/mdo.test/


MEETING SCHEDULE

NEXT MEETING: Monday April 20 1998 at Long Beach, CA, 7:00 PM-11:00 PM.


Respectfully submitted on 2/2/98,Narendra KhotWL/FIBD2130 Eighth Street, Ste 1Wright -Patterson AFB ,OH, 45433-7541Tel: (937) 255-6992Fax: (937) 255-3740Email: [email protected]



Last Updated: February 17, 1998





MDO TC Meeting Minutes (9 October 1997)

MINUTES OF THE MDO TC MEETINGOctober 9 , 1997St. Louis , MO

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 8:10 AM. Narendra Khot recorded the minutes. There was an introduction of members and guests. Members present were: Barthelemy, Giesing, Yurkovich, Eldred, Briggs, Striz, Messac, Rais-Rohani, Iqbal, Kodiyalam, Renaud, Haftka, Khot, Bhatia, Ghattas, Gilje, Grossman, Schrage, Mazzaway, Human and Bolognese. Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Jean-Francois, in order to maintain a consistent database. Changes will be passed on to Mike Eldred for inclusion on the website.

KISSIMMEE MDO-TC (APRIL 97) MEETING MINUTES REVIEWED AND APPROVED

PRESENTATION: iSIGHT at ENGINEOUS SOFTWARE Inc.

Srinivas Kodiyalam made a presentation concerning the Compuer Aided Optimization (CAO) software which is under developement at Engineous Software Inc.The software would be used for Design Automation, Design Integration and Design Optimization with least effort to switch from one to other commericially or in-house developed software. The program is being marketed and at present is being used by the industry and the govermental agencies. The software uses MDOL language. Members who attended this meeting are requested to send their comments to Jean-Francois Barthelemy.


Ramana Grandhi (General Chair) and Bob Canfield (Technical Chair) were unable to attend the meeting due to the last minute unforseen difficulties. Rudy Yurkovitch discussed the location of the conference and the support he would be getting from his company regarding permiting interested engineers from the company to attend the conference.He mentioned that about 94 persons attending the conference would be able to visit Boeing to see the PROLOG Room and military aircraft assembly lines.

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.oct97.html (1 of 4)12/29/2006 12:29:13 PM



Joe Giesing discussed the proposed white paper. This paper will emphasize real-world problems, mainly taken from industrial projects. The format has been changed from a single report to a series of invited papers accompanied by a summary paper. The invited papers will be presented in two MA&O sessions. Papers will stress the use of MDO, bring up industrial needs, or point out areas where MDO could have been used to enhance the design process.Five abstracts have been received and three are in progress. There will be two full sessions covering these invited papers

ACTION ITEM: Joe Giesing and Jean -Francois Barthelemy will write the two page summary for distribution during the sessions at the MA&O Conference.

ACTION ITEM: Joe Geising ,Bob Canfield and Ramana Grandhi will get in contact to discuss the session organization ,time etc.

1998 SDM AIAA CONFERENCE

Bhatia discussed the session organization of the papers under MDO at this conference. 49 abstracts were received and out of these 38 papers were accepted. Six sessions have been organized. Chair and co-chair persons have been selected.Bhatia will get in contact with these persons to confirm their acceptance.


Awards Subcommittee (Ghattas)Omar Ghattas discussed the roll of TC in nomination for AIAA fellows award. He will put on the web the procedure TC plans to follow . This will be voted by the TC during the Reno meeting.

Conference Support Subcommittee (Chair Vacant at Present)Bhatia suggested that TC should select representative for 1999 AIAA SDM meeting. Srinivas Kodiyalam has agreed to take this responsibility. MDO TC is not in the current SDM conference chair rotation. It was recommened that TC should send a formal letter to AIAA, stating that the committee is not in favor of becoming the conference chair.

ACTION ITEM : Bhatia and Barthelemy will prepare the formal letter to be sent to AIAA regarding TC's unwillingness to become member of SDM conference chair rotations

Publications Subcommittee (Livne)Eli Livne could not attend the meeting. However he communicated the message that he is making good progress on putting a special issue of Journal of Aircraft on MDO.



Achille Messac will start and initiate a discussion on what is MDO and which is the proper terminology.He seeks contributions from the members. He plans to organize and condense the response.

Education Subcommittee (Renaud)There is no formal charter for MDO TC with AIAA.

ACTION ITEM :Barthelemy and Renaud will draft a formal charter for MDO TC for sending to AIAA. The draft will be ready for Reno meeting.

Internet Subcommittee (Eldred)Mike talked about updates to the web page.Call for papers for MA&O meeting is on the web. MDO and optimization related web sites have been listed on the web. The list contains information from the universities, industry and the goverment agencies.

Liaison Subcommittee (Fadel)Clark Briggs mentioned that the Design Engineering TC report has been put on the web.

Benchmarking Subcommittee (Striz)Fred Striz reported his progress on selecting the problems for this task. He is of the opinion that these bench mark problems should be put on the web for easy access. He will have more on this to say during the Reno meeting.

World Aviation Congress

Dan Schrage will represent MDO TC at the Aircraft Technology Integration and Operation meeting at the WAC.

Special Thanks to Rudy Yurkovitch for arranging the meeting at the Boeing Company, and providing the bus from the hotel and to the airport.He also gave attending TC members tour of the PROLOG room, a virtual reality dome and the military aircraft final assembly building.

MEETING SCHEDULE

NEXT MEETING: Monday Jan. 12, 1998 at Reno Hilton in Reno, NE, 7:00 PM-11:00 PM.

The meeting was adjourned at 12:00 PM..

Respectfully submitted on 10/22/97,Narendra KhotWL/FIBD2130 Eighth Street, Ste 1



Wright -Patterson AFB ,OH, 45433-7541Tel: (937) 255-6992Fax: (937) 255-3740Email: [email protected]









MINUTES OF THE MDO TC MEETINGApril 4, 1997

Kissimmee, FL

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 7:05 PM. Allan Goforth recorded the minutes. There was an introduction of members and guests. Members present were: Barthelemy, Goforth, Giesing, Yurkovich, Eldred, Fadel, Thomas, Eastep, Briggs, Striz, Messac, Rais-Rohani, Kolonay, Martin, Canfield, Grandhi, Iqbal, Renaud, Majumdar, Haftka, Gurdal, Khot, Livne, Bhatia, Karpel, Ghattas, and Rocha. Member Gilje was represented by Dean Waldie (TRW). Other guests were Prabhat Hajela (RPI), Jaroslaw Sobieski (NASA LaRC), Keram Nazari (Altair Computing), Larry Pinson (MRJ Tech. Solutions), Emily Davies (AIAA), Brewster King (AIAA), and Joanna Spar (AIAA). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. It was noted that corrections should come through Jean-Francois, in order to maintain a consistent database. Changes will be passed on to Mike Eldred for inclusion on the website.

RENO MDO-TC (JAN 97) MEETING MINUTES REVIEWED AND APPROVED

PRESENTATION: OPTIMIZATION AT ALTAIR COMPUTING

Harold Thomas made a presentation concerning the optimization services and software offered by Altair Computing, Inc. Altair does structural optimization work under contract for the Big Three automobile makers using commercial software packages such as Genesis and MSC/NASTRAN. In addition, Altair has it's own software products. These are Hypermesh, a pre/post processor similar to PATRAN, and Optistruct, which does topology optimization. The company has a website (www.altair.com).

PRESENTATION: AIAA BUSINESS DEVELOPMENT

Emily Davies outlined the restructured (flat) organization of AIAA and the new emphasis on business development. Results from last year's survey of the membership by market analysts were discussed. Response to the calendar survey was excellent. Brewster King talked about business development. He works with the TC's to improve the conferences and define new technical areas that AIAA should get



involved with. He is also interesting in developing new ideas for continuing education courses.

PRESENTATION: CONGRESSIONAL VISIT DAY

Larry Pinson discussed the AIAA Congressional Visit Day held March 12, 1997. The AIAA wanted to make three main points: (1) AIAA represents both business and individuals in an aerospace industry that was responsible for a $24 billion positive balance of trade last year. (2) AIAA is located locally and should be the source of information for the aerospace community. (3) AIAA is very concerned about the well being of research in the United States. The members of congress visited seemed quite receptive to our input. Congressman Dana Rohrbacher asked AIAA to take a position on proposed fast-track patent legislation, which he is afraid may harm small businesses.


Ramana Grandhi (General Chair) announced that the1998 MA&O Conference will be held Sept. 2-4 (Wed, Thur, Fri before Labor Day), 1998, at the Hyatt Regency in St. Louis. The theme of the conference will be real-world applications of MDO. It was mentioned that the St. Louis area has gambling on riverboats, and that a $1 Metrolink ride to the hotel is available. Bob Canfield (Technical Chair) announced that he is seeking volunteers for Superchairs. These people basically will arrange to have reviewers for the abstracts in their area, distribute abstracts to reviewers, collect them after review, consolidate scores, and pass these on to the organizing committee. A call for papers will be sent to AIAA around 9/1/97. The November Issue of Aerospace America will include the call for papers. Abstracts to Superchairs will be due 1/31/98. Reviews of these abstracts are due 2/21/98. There is a possibility of using electronic submittal for the abstracts - a website may be established for this purpose.


Joe Giesing discussed the proposed white paper. This paper will emphasize real-world problems, mainly taken from industrial projects. The format has been changed from a single report to a series of invited papers accompanied by a summary paper. The invited papers will be presented in two MA&O sessions. Papers will stress the use of MDO, bring up industrial needs, or point out areas where MDO could have been used to enhance the design process. Several suggestions were made concerning subject matter, and a list of possible authors was presented by Joe. The call for papers and guidelines has been drawn up and a tentative schedule established. The papers will be due before the normal deadline for papers, in order for the summary paper to be written. Some prospective authors have been contacted. It was suggested that the list was dominated by aircraft, and additional authors from the space community will be contacted.

ACTION ITEM: Fred Striz will contact people at Loral to see if they can participate.

ACTION ITEM: Dean Waldie will make contact at TRW regarding their participation.



ELECTIONS

Elections will be held electronically on the web before the next TC meeting. This should take place during July/August and be complete by September 1.


Awards Subcommittee (Thomas)Harold Thomas is rolling off the TC, and Omar Ghattas will be replacing him as chair. The process for selecting the best paper award at the MA&O was discussed. At present, the process is to review papers which are chosen from both the highly-rated abstracts and the best-of-session papers, as chosen by the session chairs. A motion was made and seconded to continue this process in the future for best paper. MOTION APPROVED.

The question of including MDO papers from other conferences (SDM, ASM, WAC, etc.) as candidates for best paper was brought up. A motion was made to limit the candidate papers to only those from the MA&O Conference. MOTION APPROVED.

A discussion of when to present the award ensued. A motion was made to notify the recipient, announce the award in Aerospace America, and to wait for the next MA&O Conference for the award presentation. MOTION APPROVED.

A motion was made to recognize the ten finalists by sending a letter from the TC to each one. MOTION APPROVED.

The TC nomination for AIAA Fellow was discussed.

ACTION ITEM: Harold will send everyone emails with information on the nominees, and voting will be done electronically.

Conference Support Subcommittee (Chair Vacant at Present)Kumar Bhatia reported on the 97 SDM conference. There were 6 invited one-hour papers, and these were very successful (the rooms were full). About 2/3 of submitted abstracts were accepted. Kumar is also the representative for the 98 SDM conference. He wants to maintain the invited papers, but focus on one or two areas - possibly from industry. He would also like to limit these presentations to 45 minutes, leaving 15 minutes for questions.

A suggestion was made that the MA&O replace the Adaptive Structures Conference as a two-day event every other year at the SDM Conference. This proposal was voted on by the TC. MOTION FAILED.

Kumar asked for any suggestions concerning the invited papers for MDO at the 1998 SDM Conference



to be emailed to him.

Publications Subcommittee (Livne)Eli Livne is working with Tom Weeks to put together a Journal of Aircraft issue on MDO. He has about 15 papers which are candidates for inclusion, but is looking for more. He has asked people to come up with MDO papers more general in scope with good bibliographies, instead of using previously published papers. This makes it more difficult, but Eli is now more hopeful of success than a few months ago.

Achille Messac has been asked for help in identifying knowledgeable reviewers for several AIAA journals.

ACTION ITEM: Achille will send out emails asking for information in order to form a database of potential reviewers.

Education Subcommittee (Renaud)Because the education panel session at the MDO meeting in Seattle was so successful, this will be done again at the next MA&O Conference.

Internet Subcommittee (Eldred)Mike talked about updates to the web page. The membership list has been updated. New additions to the website include number of hits and FAQ. Mike discussed ideas on implementing Sobieski's idea of having an MDO Electronic Forum. The recommended approach is a combination of email and Web based formats.

Liaison Subcommittee (Thomas)Georges Fadel will be replacing Harold as subcommittee chair. Shreekant Agrawal will replace Rudy Yurkovich as the liaison for the Applied Aero TC. Clark Briggs reported that the Design Engineering TC had two sessions on knowledge-based design at the WAC which were very good. They are putting out the third edition of their design handbook, this time an electronic version. Fred Striz reported that the Structural Dynamics TC is producing a structural dynamics video.

Benchmarking Subcommittee (Striz)Fred Striz reported that the test suite of MDO problems being developed at NASA Langley can be accessed at the following website:

http://fmad-www.larc.nasa.gov/mdob/mdo.test/start.html

This includes a new CASCADE code (Complex Application Simulator for the Creation of Analytical Design) developed at SUNY/Buffalo by Christina Bloebaum. Fred is working on the report of the panel discussion and paper session on benchmarking and testing at the last MA&O Conference. He would like to include a session on benchmarking and testing in MDO at next year's MA&O Symposium and if there is enough interest, at the next SDM. If you are interested in writing a paper in this area, regard this as the




first Call for Papers, and contact Fred. Fred reported that he has the MBB fin benchmarking problem available to anyone interested. It consists of structural model data on disk with papers and reports covering previous comparisons. You must supply the aerodynamics. Contact Fred Striz if you are interested.

MEETING SCHEDULE

NEXT MEETING: Thursday, Oct. 9, 1997 at Henry VIII Hotel & Conference Center in St. Louis, MO, 8:00 AM-12:00 NOON. McDonnell-Douglas facility tour 1-5 PM.

The meeting was adjourned at 23:15 hours.

Respectfully submitted on 8/22/97,Allan GoforthLockheed Martin Skunk WorksDept. 25-45, Bldg. 6111011 Lockheed Way Palmdale, Ca. 93599-2545Tel: (805) 572-4997Fax: (805) 572-6327Home email: [email protected] email: [email protected]










Reno, NV

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 7:00 PM. Allan Goforth recorded the minutes. There was an introduction of members and guests. Members present were: Barthelemy, Goforth, Giesing, Yurkovich, Drela, Agrawal, Eldred, Gelhausen, Bhatia, McIntosh, Olds, and Fadel. Members Eastep, Rais-Rohani, and Raj were represented by V. B. Venkayya, Abdollah Arabshahi, and Brian Goble, respectively. Other guests were Andrew Santangelo, Jerry Heffner (AIAA), Bob Bell (AIAA Business Dev.), Tom Weeks (J. of Aircraft), Karl Bradshaw (AIAA), Emily Davis (AIAA), and Cathy Chenevey (AIAA). Thanks to these and their sponsoring organizations for attending.


Current membership information, i.e., name, email address, subcommittee membership status, was passed around for review and update by members present. All members should check this information on the web and send corrections to Mike Eldred.

PRESENTATION OF AIAA STATUS

Emily Davis made a presentation in which the following points were made:

● AIAA now has 29,931 members, a slight increase over last year. ● 29 AIAA Fellows were picked this year, and will be honored in May at the Awards Banquet. ● There are 3 new Corporate members (a total of 29 Domestic and 19 International Corporate

Members). ● AIAA published 6 new books. ● A mid-March congressional visit is planned. ● The first year of providing AIAA Journal and Meeting Papers on Disk was successful. ● There have been over 200,000 hits since March on the AIAA website.

Jerry Heffner from TAC (Technical Activities Committee) discussed how the MDOTC fits in with the basic AIAA organization. It was pointed out that virtually all the personnel making up the TAC organization are volunteers. In August, an activity called TAC Self-Examination was initiated. The purpose of this is to develop new initiatives which will make AIAA participation attractive to potential members. They want to make it clear to sponsoring organizations that there is a payoff in value returned



to that organization when there is participation in conferences, symposiums, etc.

Some of the possible initiatives are:

● Assessment and evaluation of TAC with regard to meeting members needs in 2005 ● Development of Virtual Symposia ● Enhance International cooperation and collaboration ● Assessment of TAC effectiveness ● Assessment of AIAA Conferences

These initiatives may lead to recommendations for TAC reorganization.

BELLEVUE MDOTC MEETING MINUTES REVIEWED AND APPROVED


Joe Giesing presented slides of his plans for producing a TC white paper on MDO in 1998. Joe mentioned that the old white paper, from 1991, would be put on the web by the end of the month. Joe made a strawman outline for the white paper and sent it to the 16 people who expressed interest. He asked for comments and suggestions. Joe said the response was helpful, but not large. This led Joe to conclude the following: (1) that the outline may be too ambitious, (2) we have enough survey papers, and (3) we should present real-world MDO needs. Discussion lead to the idea of having "White Paper Sessions" at the 1998 MA&O Conference. These papers would present needs - not solutions in search of a problem. This idea will be developed further and discussed at the next TC meeting.

1997 MEMBERSHIP SELECTION FOR MDO TC

John Olds presented a review of the current roster for the TC, and discussed membership goals. It was noted that 50% of members will roll off the TC in 1998, and we will need critical replacements from NASA-Ames, Boeing, and McDonnell Douglas-St. Louis. The 1997 applicant pool includes 9 full members, 4 associates, and 2 international potential members. The TC membership committee recommended a "yea" vote on all the applicants. This motion was put to the TC and approved. The new TC members are Balling(BYU), Bolognese(NASA-Goddard), Chang(Lockheed-Martin Skunk Works), Gilje(TRW), Grossman(Virg. Tech), Human(NC A&T St.), Khot(AF-Wright Lab), Kolonay(AF-Wright Lab), Martin(GE Aircraft Eng), Mazzawy(UT Pratt&Whitney), Mosher(Aerospace Corp), Peraire(MIT), Petiau(Dassault, France), Schrage(Ga. Tech), and Suleman(IST, Portugal).

AIAA SHORT COURSES

Bob Bell, from AIAA, reported on AIAA short courses offered in 1998. He is trying to generate interest. He mentioned that they would be meeting later in the week at Reno to finalize plans. Any TC member



interested in developing a short course or tutorial should contact Bob.


Dr. Weeks presented slides on the process of publication at the Journal. They keep close watch on the backlog of publications, since this is a measure of the stability of the Journal. Two types of backlog are tracked: accepted and pre-accepted. Accepted backlog consists of papers that have undergone review and revision, and are awaiting publication. Pre-accepted backlog is made up of papers which have been submitted, but are undergoing review and revision. Their problem is that, while pre-accepted backlog is holding fairly constant, accepted backlog is declining rapidly. This is mainly due to the fact that reviews and revisions are taking much longer to complete. Consequently, Dr. Weeks is asking the TC for help in identifying and motivating reviewers. Also, authors need to be motivated to provide timely revisions to their papers. In addition, he requested feedback from the TC on how the publication process could be improved. It was also mentioned that Frank Eastep will be an Editorial Board Advisor, as well as Associate Editor for the Journal.

AIAA CONGRESSIONAL VISIT DAY

Andrew Santangelo discussed the upcoming Congressional Visit Day this March. Andrew attended a planning meeting Monday in Reno. He will canvass the TC membership for suggestions by email. Mike Eldred will put a link on our website to Andrew's to facilitate this.


Applications Subcommittee (Radovcich-not present)No report. It was decided that this subcommittee needs a replacement for Radovcich, since he is unable to devote time to the task.

ACTION ITEM: Joe Giesing will search for a new chairman.

Awards Subcommittee (Thomas- not present)Jean-Francois reported. Harold Thomas is rolling off the TC in April, and Omar Ghattas will be replacing him as chair. Omar missed Reno because of illness. The process for selecting the best paper award at the MA&O was discussed. Now the process is to issue the award two years later at the next MA&O conference. There seemed to be general agreement that the award should be based on the paper, not the abstract. There was also agreement that the award should be presented as soon as possible after the MA&O meeting. Recommendations will be passed on to Omar, and the process will be formally defined.

Rudy Yurkovich pointed out that he was a member of a group that arrived at criteria for evaluating awards, and sent a letter to Virgil Smith with recommendations. Rudy noted that the MDO award fails



most of the criteria.

ACTION ITEM: Rudy will mail a copy of this letter to Omar.

Congratulations to Vipperla Venkayya and Rafi Haftka, who were both named AIAA Fellows. The subcommittee will decide at the SDM meeting who to nominate for Fellow in 1997. Call for nominations for 1998 MDO Award will come out in Oct. 1997.

Conference Support Subcommittee (Chair Vacant at Present)

● Shreekant Agrawal reported on the current ASM conference. Of the MDO papers, 8 came from academia, 4 from research centers, and 1 from industry. There was a 30% rejection rate for MDO papers.

● Kumar Bhatia reported on the 97 SDM conference. Kumar received 41 papers, of which 27 were accepted (66% acceptance rate). There were 6 invited papers. We need a representative for the 98 SDM conference. Kumar will consider it - it was a lot of work. He mentioned there were 3 reviews of each paper.

● Paul Gelhausen reported that the 2nd World Aviation Conference will be held in October 25, 1997 in Anaheim. He wants to set up 2 MDO sessions. Papers can be full papers or not - slides may be presented alone. Paul is looking for help, as session chairs, etc. He will send email to TC members on the subject.

Education Subcommittee (Renaud - not present)John Olds mentioned that he is no longer required to prepare a question for the PE's exam, because the NSPE is dropping the Aerospace Professional designation, and the test will not be offered after this year.

The education panel session at the MDO meeting in Seattle was well attended and enthusiastically received. Panelists included Bill Mason of VPI, Ilan Kroo of Stanford, Dan Schrage from Georgia Tech, Kumar Bhatia of Boeing, and Rudy Yurkovich of McDonnell Douglas. A paper session on engineering education was also a success. The education subcommittee would like to volunteer to organize a paper and a panel session at the 1998 MDO meeting.

Internet Subcommittee (Eldred)Mike talked about changes to the web page. He has added an alternate membership list with no pictures which is much faster. Subcommittee reports are now being included, and subcommittee chairs are urged to send this information directly to Mike. He also noted that the old MDO white paper is being added.

Liaison Subcommittee (Thomas - not present)Georges Fadel will be replacing Harold as subcommittee chair. Georges reported that Bob Canfield sent a report on the AI TC. They would like to participate in the next MA&O conference by organizing a special AI session. At their next meeting, the chairman will solicit a potential chairman for an MAO session on AI Applications for Design.



Rudy Yurkovich discussed the Applied Aero TC meeting which took place on the previous evening in Reno. Rudy mentioned that they had name tents at their TC meeting - this would be a great convenience, since most people on a TC are not acquainted with everyone else. The Applied Aero Chair is encouraging his TC members to branch out and support other technical areas - so there may be a chance to recruit some aerodynamic people for our TC.

Publications Subcommittee (Livne - not present)Eli Livne is working with Tom Weeks to put together a Journal of Aircraft issue on MDO. He has some papers which are candidates for inclusion, but is looking for more.

Benchmarking Subcommittee (Striz - not present)Jean Francois reported from Fred's written submittal. The MA&O Symposium in Bellevue featured a paper session on benchmarking, as well as a panel session moderated by Professor Haftka.

ACTION ITEM: The subcommittee is compiling the ideas received on benchmarking during those sessions, and will have a full report at the next TC meeting in Florida.

AIAA EVENT PLANNING REORGANIZATION

Karl Bradshaw introduced Cathy Chenevey, who will be the events planner for the next MA&O Symposium 98. Karl said they had reorganized at headquarters. They have hired some new people and are trying to get more people involved in event planning so the body of knowledge is not lost if they decide to leave AIAA. Karl has done some research on possible locations for the next MA&O Symposium, and he will pass the information on to Ramana Grandhi.

MEETING SCHEDULE

NEXT MEETING: Monday, Apr.7, 1997 in Orlando, Florida 7:00-11:00 PM. Dinner at 6:00.












MDO TC Meeting Minutes(5 September 1996)

MINUTES OF THE MDO TC MEETINGSeptember 5, 1996

Bellevue, Wa.

PRELIMINARIES

Chairman Jean-François Barthelemy called the meeting to order at 7:15 PM. Allan Goforth recorded the minutes. There was an intorduction of members and guests. Members present were: Barthelemy, Goforth, Rais-Rohani, Thomas, Bloebaum, Mistree, Striz, Messac, Haftka, Giesing, Yurkovich, Eldred, Seidel, Bhatia, Livne, Renaud, Grandhi, Ghattas, Canfield, Ewing, McIntosh, Karpel, Gurdal, Morris, Briggs, Olds, Majumdar, Gelhausen, Eastep, Fadel, and Lawrence. Guests attending were Chris Borland (Past Chariman), Jaroslaw Sobieski (NASA LaRC), Kemper Lewis (SUNY-Buffalo), Wei Chen (Clemson), Evin Cramer (Boeing), Prabhat Hajela (RPI), Johnsoo Lee (RPI), Mike Long (Cray Research/SGI), and Johannes Schweiger (Daimler-Benz). Thanks to these and their sponsoring organizations for attending.

SALT LAKE CITY MINUTES REVIEWED AND APPROVED

DISCUSSION OF CURRENT MA&O CONFERENCE

Chris Borland thanked everyone for their help in organizing the current symposium. Chris also introduced the team to head the next MA&O Conference: Ramana Grandhi from Wright State University will be the overall Chairman, with Bob Canfield (AFIT) as the Technical Chair. Christina Bloebaum mentioned that there were 360 attendees at the conference - this compares favorably to the 274 at the previous one. The general feeling was that the conference had gone very well. The panel discussions, in particular, seemed very popular.

The discussion of the current symposium evolved into a general discussion of the approval process for papers. Rafe Haftka thought we should try harder to eliminate "bad" papers from the conference. Moti Karpel suggested that a date should be set by which papers could be withdrawn if the work wasn't complete. However, it was pointed out there seemed to be no correlation between late papers and "bad" ones. Another suggestion was to review briefly papers as they are submitted, but there was general agreement that this extra step would be time consuming. Fred Stritz suggested moving the papers to a "work in progress" session, if they were incomplete. Several TC members thought that the quality of the papers was generally very high at our conference, and that we should not be overly concerned with the few that didn't measure up. It was pointed out by Jean-François that any change to the selection process could be worked out by the new Chairs for the next conference.



Chris mentioned that we were forced into a short week because of hotel availability and Labor Day. He thought we might need to add a day to the next conference. This conference included two social events, and the general feeling among the TC was that this was about the right number. Jean-François brought up the issue of annual MA&O conferences and a discussion of this ensued. It was decided to delay any decision on annual meetings, and wait to see if attendance holds up. Joe Giesing suggested we tighten up on approval of papers a little and see how it goes. Location of the next conference was discussed, and an informal show of hands showed that Montreal and San Antonio were popular with the TC. Chris pointed out that this was a highly constrained problem, however, due to hotel availability, time of year, etc.

WHITE PAPER ON MDO

Jean-François brought up the possibility of writing a white paper on the state of MDO, similar to one produced in 1991. A discussion of this took place, resulting in a motion to produce such a white paper by the time of the next MA&O Conference - the format and content to be determined by the authors. The motion passed.

Action Item: Joe Giesing will try to put together an outline by the Reno meeting.

Jarek Sobieski initiated a discussion concerning the conflict between industry - MDO users, and Universities - MDO methodology developers. Industry typically wants software checked out with real-life problems, and the developers don't always do this. Jarek suggested setting up a clearinghouse for MDO problems on the Web, which would act as an interface between industry and academia. It might be included with the NASA problem suite, or with the TC Home page.

Action Item: Jarek and Rafe Haftka will investigate to see what resources are required.


Applications Subcommittee (Radovcich-not present)Nick had the idea of putting together a brochure on MDO, but it is not clear if he is able to work on it.

Action Item: Joe Giesing will follow up with Nick Radovcich to see what he will be able to do for the subcommittee.

Awards Subcommittee (Thomas)The best paper award at the MA&O was discussed. Harold got 5 volunteers to review approximately 20 papers. Frank Eastep said that the Journal of Aircraft would like to institute a "fast track" for publication of the better papers as chosen by the TC. He will work with Harold to pick the top 25 papers from the current conference. Harold also mentioned that he will be dropping off the TC after the SDM Conference, so a replacement will be needed.

Action Item: Harold will decide on the criteria for judging best paper, and come up with a selection



process by the Reno meeting.

Conference Support Subcommittee (Chair Vacant at Present)Joe Giesing reported Agrawal's notes from the Reno ASM conference. There were two MDO sessions with 13 papers presented - five were rejected. Paul Gelhausen reported that the 2nd World Aviation Conference will be held in October 25, 1997 in Los Angeles. This group has reduced the standards for their papers in order to get more up-to-date information. Kumar Bhatia, our SDM representative, inquired to the SDM TC about our participation in the rotating chair for the SDM Conference. This met with a negative response. He will continue to pursue this, at a low level, since our TC didn't express strong feelings on the matter. Kumar has received 39 papers for SDM 97. Last time there were 55 submitted, with 45 accepted. He would like to have an invited lecture at the start of each MDO session. This would be a 20 minute talk with 20 minutes for discussion. The TC generally supported this idea.

Education Subcommittee (Renaud)John Olds has volunteered to come up with a test problem for Conrad Newberry's Professional Engineer's exam. John Renaud reported that the education panel discussion at the current conference was very successful.

Internet Subcommittee (Eldred)Mike talked about changes to the web page. Membership list includes AIAA membership level and picture. Check to see if it's correct. If you have not sent a picture, please send either a GIF or JPEG format that is already shrunk to about the proper size. Subcommittee reports are now included, and subcommittee chairs are urged to send this information to Mike. Mike mentioned that the last issue of the newsletter has not gotten on the website. It was also brought up that we should register our web page with the major search engines.

Action Item: Georges Fadel will take action to register our web page.

Liaison Subcommittee (Thomas)Harold presented notes from Srinivas Kodiyalam, the MDO liaison to the Structures TC. Items included the following: (1) AIAA focus on standards for coverage of space launches. (2) Report on the 38th SDM-theme is "Vision 2000". (3) Upcoming Conferences - World Aviation Congress, International Adaptive Structures Conf., 3rd International Conf. On Composite Eng., 11th International Conf. On Composite Materials, Society of Engineering Science: 33rd Annual Meeting. (4) August 26-27 Planning meeting of Structures TC. - Location: Wright Patt AFB.

Rudy Yurkovich discussed the Applied Aero TC meeting which took place last June in New Orleans. They have received 116 abstracts and accepted 104 of these for Reno meeting. They have made some progress on developing a "fast track" process for some of the papers. Rudy mentioned that some editors are not in favor of this- they want peer review.

Membership(Olds)



John has been active in trying to recruit new members. The response looks good. Applications were available at the current conference, and many were picked up by the attendees.

Newsletter (Bloebaum)The last issue of the newsletter has been published.

Publications Subcommittee (Livne )Achille Messac put together MDO highlights for the December issue of Aerospace America. He thanked the TC membership for providing him with a lot of input.

Benchmarking Subcommittee (Striz)Fred reported that the set of test problems developed at NASA-Langley is being readied for its official opening at or before the MA&O meeting. It has grown to three problem classes and a total of 17 cases. Sample test pages can be accessed under the following URL:


The current MA&O Conference featured a full paper session on benchmarking and a panel discussion devoted to this topic. Fred also reported that he will be receiving a test case for the MBB fin benchmarking problem soon. There has been no progress on obtaining the Airbus wing model.

MEETING SCHEDULE

NEXT MEETING: Monday, Jan. 6, 1997 in Reno. 7:00-11:00 PM - Dinner at 6:30

SPRING MEETING: Monday, Apr.7, 1997 in Orlando 7:00-11:00 PM













MDO TC Meeting Minutes(15 April 1996)

MINUTES OF THE MDO TC MEETING15 April 1996

Salt Lake City, Utah

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 7:10 PM. Allan Goforth recorded the minutes. Members present were: Barthelemy, Goforth, Rais-Rohani, Thomas, Sepulveda, Bloebaum, Mistree, Striz, Messac, Haftka, Giesing, Kodiyalam, Yurkovich, Eldred, Seidel, Bhatia, Livne, Renaud, Grandhi, Radovcich, Ghattas, Canfield, Ewing, and McIntosh. Guests attending were Mike Long (Cray Research/SGI), Zafer Gurdal (Va. Tech), Moti Karpel (Technion-Israel), Miseille Gerard (AIAA), Emily Davies (AIAA), Carlos Orozco (U. of Va.), Christos Chamis (NASA Lewis), Doug Sagehorn (Raytheon), Ashok Belegundu (Penn. St.), and Martin Stettner (Daimler-Benz, representing Johann Krammer). Thanks to these and their sponsoring organizations for attending.

MEMBERSHIP CHANGES

Jean-Francois announced the selection of five new members for the TC. They are Dr. Frank Eastep from U. of Dayton, Dr. Mike Eldred from Sandia, Dr. Moti Karpel from Technion-Israel, Dr. Alok Majumdar from Sverdrup (MSFC), and Prof. Allan Morris from Cranfield Univ. (UK). The TC extends a cordial welcome to these very capable individuals. In addition, Past-Chairman Chris Borland has decided to retire from the TC, after seven years of perfect attendance. Thanks to Chris for all the great work he did for the TC during that long timespan.

NEW WEB SITE

The TC web site has been moved to http://endo.sandia.gov/AIAA_MDOTC/main.html

RENO MINUTES REVIEWED, CORRECTED, AND APPROVED

PLANNED OPTIMIZATION CONFERENCE

Farrokh Mistree presented ideas for an optimization conference which is to be held in Florida during March of 1997. The focus of this conference is Engineering Design and Optimization in Industry, and Dr. Ashok Belegundu of Penn. St. will be the conference chairman. There will be substantial industry participation with the purpose of establishing a dialog among academia and industry. There will also be international participation. They want to establish a true dialog - what works, what doesn't, what are industry's problems, and where do we go from here. They are seeking the following things from our TC:



(1) Co-sponsorship by AIAA, (2) Inclusion of a flier in MDOTC newsletter, (3) Links from AIAA and TC websites to their website, and (4) participation of AIAA members in the conference. The conference is planning on 11 or 12 full-length papers. The rest of the participants will come with ideas - not complete, but evolving, expressed on two-page position papers. The four-day conference is expecting to attract about 65 people, with perhaps 15 or so invited. Poster-board sessions are also planned. They are trying to keep expenses to a minimum - approximately $500.

In the following discussion, it was pointed out by Rudy Yurkovich that getting AIAA sponsorship will require considerable effort. A more informal type of support was discussed, and the conclusion was that the TC could add something to the newsletter, put links on our website, and perhaps even write a letter of support, but that formal sponsorship by AIAA was not possible at this time. Anyone interested in participating in the conference should contact Farrokh.

AIAA REORGANIZATION

Emily Davies, Staff Liaison for Technical Activities, gave a presentation on the new organizational structure of the AIAA. Headquarters has moved from Washington to Virginia, basically for financial reasons. A study concluded two years ago that the AIAA should be reorganized. Many layers of management have been eliminated, and the AIAA is now organized around projects. The new organization makes it easier to find help, and a new 800 telephone line has been added. The technical committees, consisting of 2500 volunteers, make up the largest part of the organization. AIAA membership is now at about 30,000. AIAA now recognizes that TC's have a much broader function than just to host conferences.

Miseille Gerard, who was brought in for business development, then spoke about the future of AIAA and how we could help. AIAA was doing no marketing until recently. She wants to work with the TC to establish long-range goals, and is eager to work on improving processes to achieve these goals. All conference papers will now be put on CD-ROM, and will also be printed on paper, because surveys showed that libraries wanted hard copy also. Abstracts from as far back as 1960 have been added to the CD-ROM database. Eventually, they will probably go to all electronic publishing, but surveys showed that their customers are not ready for that yet. The process of making a conference CD-ROM now takes about 12 weeks, which is too long a time before the conference. Authors often have trouble getting papers in six weeks before a conference. Even though authors generally submit disks as well as hardcopy, the disks must be scanned and copied to CD.

ELECTION OF VICE-CHAIRS

Mr. Joe Giesing of McDonnell Douglas and Dr. John Renaud of Notre Dame were nominated and elected by acclamation to the positions of Vice-Chairman/Technical and Vice-Chairman/Communications, respectively.

INDUSTRY VISION FOR MDO



Jean-Francois reported the results of combined subcommittees at NASA who were tasked with reviewing MDO progress. One of their conclusions was that Industry needed to make clear what it expected and required from MDO, possibly including areas where research is needed. Much discussion took place concerning the scope and purpose of this task, which might be taken up by industry members of the TC. It was decided to assemble the members from industry and any other interested parties in a separate meeting the next day to continue the discussion.


Applications Subcommittee (Radovcich)Harold Thomas sent Nick an example of a project which failed because MDO was not used. Nick is investigating the idea of putting together a kind of sales brochure for MDO, which may include items such as this. This would show examples of optimization problems and indicate what kind of payoffs are obtained by using MDO. The applications subcommittee has not met since Reno.

Awards Subcommittee (Thomas)The MDO Award was discussed by Harold. It will be given at the MA&O Conference, and the recipient is chosen by a selection committee of five from our TC. Nominations come from the AIAA, and these have been received. A recipient has not been chosen yet because only two of the selection committee members are attending this SDM conference. The possibility of giving a best paper award at the MA&O was discussed. There is a problem with giving the award based on the abstract or on reviews by session chairmen. It seems to be impossible to read through all the papers. In addition, the subcommittee has nominated Dr. Venkayya for AIAA Fellow.

Conference Support Subcommittee (Santangelo - not present)Jean-Francois presented the report. Three conferences are being supported at this time. Ramana Grandhi reported on the current SDM conference - 11% of papers were rejected this year. There were many MDO papers at this year's SDM, and 7 sessions were devoted to MDO. Kumar Bhatia will be our representative at the SDM conference in 1997. Christina mentioned that at last year's SDM, some papers were chosen for poster sessions without the author's consent; she is opposed to this practice. Shreekant Agrawal wanted to remind everyone that the abstract deadline for next year's Reno conference is May 17. The other conference we support is the Global Aviation Conference, held in the fall, which is a joint conference with the SAE. Paul Gelhausen is our representative for this.

Education Subcommittee (Renaud)A report was provided by John. The following items were mentioned: The MDO TC review procedure for short courses has been finalized. A copy of this procedure is being forwarded to Prof. Murthy at Purdue. The MA&O Conference in Seattle will feature one paper session dealing with MDO education and one panel session which will address this subject. The panel will be composed of representatives from industry, government, and academia. John is looking over the material required for Conrad Newberry's test problem for the Professional Engineer's exam. He will put something on the web when



he has it ready and anyone interested can participate in coming up with the problems and required documentation.

Internet Subcommittee (Thomas)Harold talked about changes to the web page. Membership list includes AIAA membership level and picture. Check to see if it's correct. If you have not sent a picture, please send either a GIF or JPEG format that is already shrunk to about the proper size. Subcommittee reports are now included, and subcommittee chairs are urged to send this information to Harold. Related items of interest are now linked to our website.

Liaison Subcommittee (Thomas)Harold reported that there were no new reports, since the last meeting was just three months ago. They still need liaisons for Materials TC, GNC TC, Standards TC and AI TC. Harold said that the Liaison job is getting easier because more TC's are establishing web pages. It is not really necessary to attend the meetings. The task is to filter the MDO content out of their minutes. Bob Canfield volunteered for the AI slot. Rafi Haftka passed out a flyer on an upcoming design competition which uses MDO to design and build the smallest airplane which is capable of performing a specific mission. It must fly a specific course and photograph four letters marked on the ground. Kumar Bhatia is seeking suggestions for speakers for the 1997 SDM Conference, for the plenary session and Awards Luncheon. Farrokh Mistree reported that he is the General Chair for the 18-23 September 1998 ASME Conference in Atlanta. He wants to put this on the record, since there have been conflicts in the past.

MA&O Conference Subcommittee (Bloebaum)Christina had several handouts, including the newsletter. A list of session chairs was also passed out - the authors and chairs should be getting packages from the AIAA soon. The conference generally has five parallel sessions, with six sessions occurring twice. Also work-in-progress sessions of fifteen minutes each will be held. These facts give us a large total number of papers ( ~ 210). Work-in-progress papers will be given an AIAA number, and the author's will be asked to write a two page overview which will appear in the proceedings. There are 38 sessions, four of these being panels. The people organizing the panel sessions are Fred Striz (Benchmarking in MDO), Jim Rogers (Managing the Design Process), John Renaud (Issues in Engineering Education), and Christina is temporarily heading up the Industry panel session. She is looking for an Industry person to take this over, however. These panel session chairs are themselves organizing the sessions and will contact other TC members for help.

Christina proposed a plan for electing the new General Chair for the MA&O Symposium, and she passed out a handout which included the responsibilities of the General Chair and the Technical Chair (who is appointed by the General Chair). Also included was the process of selection for the new General Chair. Briefly stated this process was the following:

Between the AS and SDM Conferences, an announcement will be made that the present committee (General and Technical Chair) is accepting nominations. This announcement should be made to all present and past TC members, since these people are all eligible. If the nomination is not a self-



nomination, there must be assurance that the nominee is willing to serve. Nominations should be written and should outline the resources that the individual can commit to the task and any experience/background which might make him/her appropriate. Between SDM and MDO, the committee will review nominations and prioritize selections. This will be sent to the MDO TC Chair for comments, which he will relate to the General Chair. The present General Chair will then make the selection for the next General Chair prior to the MA&O Symposium. The new General Chair is encouraged to select a new Technical Chair before the MDO.

A discussion of the responsibilities and the selection process ensued. It was decided to add two modifications to the proposal: (1) If available from AIAA, a profile of desirable attributes for the Chairs will be added to the proposal, and (2) the announcement for nominations will be sent to members of ISSMO. This proposal was seconded and approved by voice vote of the TC.

Publications Subcommittee (Livne)Eli reported that the subcommittee is planning an issue of Journal of Aircraft devoted to MDO. He hopes to select papers from the MA&O Conference for this. Eli would like to get the papers from AIAA, as early as possible, to facilitate this process. He will also need help in reviewing the papers, to see which ones are suitable. TC members may be asked to review one paper for this purpose. Eli mentioned also that he would like to get someone else to put together the annual optimization article for the December issue of Aerospace America. Achille Messac volunteered for this task.

Benchmarking Subcommittee (Striz)Fred reported that the set of test problems developed at NASA-Langley is being readied for its official opening at or before the MA&O meeting. It has grown to three problem classes and a total of 17 cases. Sample test pages can be accessed under the following URL:


A tech transfer agreement between USAF and Daimler-Benz for an Airbus wing model is still in the works. The subcommittee is attempting to get disk space on AIAA machines to store some of the larger benchmarking cases. The MA&O Conference will feature a session on benchmarking and a panel discussion devoted to this topic.

ACTION: Fred will circulate an email request asking for input as to what questions will be addressed by the panel.

MEETING SCHEDULE

NEXT MEETING: Thursday (Plenary Meeting) & Friday (Subcommittee Meetings), Sept. 5&6, 1996, 7:00-10:00 PM, MA&O Conference in Bellevue

WINTER MEETING: Monday, Jan. 6, 1997 in Reno, 7:00-11:00 PM




SPRING MEETING: Monday, Apr. 7, 1997 in Orlando, 7:00-11:00 PM





Last Updated: June 25, 1996





Latest MDO TC Meeting Minutes( 18 January 1996)

MDO TC Meeting Minutes18 January 1996

Reno

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 8:35 hours. Allan Goforth recorded the minutes. Other members present were: Past Chairman Borland, Radovcich, Agrawal, Bhatia, Bloebaum, Drela, Fadel, Gelhausen, Ghattas, Giesing, Haftka, McIntosh, Mistree, Raj, Santangelo, Thomas, and Yurkovich. Otto Sensberg attended representing Johann Krammer. Special thanks to those and their sponsoring organizations for consecutive attendance.

LOS ANGELES MINUTES REVIEWED, CORRECTED, AND APPROVED


Conference Support Subcommittee (Santangelo)Andrew Santangelo is stepping down from the chair of this subcommittee, and is looking for a replacement. Andrew reported that the MDO committee hosted two full sessions at this year's Aerospace Sciences Meeting and Exhibit (ASM). The sessions were well attended. Shreekant Agrawal will be the MDO representative for the 1997 ASM conference. At this time, the ASM is planning for two MDO sessions, Monday morning and afternoon, for 1997 in Reno. Our TC meeting schedule will have to take this into account. Kumar Bhatia volunteered to be the MDO representative for the 1997 SDM Conference.

Education Subcommittee (Renaud - not present)A report was provided by John and presented by Christina. The following items were mentioned:

1. A panel session on engineering education at the 1996 MA&O will be held. John will be leading efforts to organize this. Contact him if your are interested in participating or have any input.

2. NASA Langley has put together an MDO Test Problem Suite which is available at the following web site: http://fmad-www.larc.nasa.gov/mdob/mdo.test/start.html. Sharron Padula was the lead on this project at Langley.

3. A formal procedure for approving sponsorship of AIAA Short Courses was proposed.

Awards Subcommittee (Thomas)The purpose of the subcommittee and the process of upgrading membership status for the AIAA was discussed. This is explained in the AIAA Handbook, which is connected to our Web Page. Nominations



for AIAA Fellow must be in by July 15 - five written recommendations are necessary (AIAA encourages them to be from other AIAA Fellows). The subcommittee will decide on a list of candidates to be voted on by the entire TC at the April SDM meeting. The MDO Award will be given at the MA&O Conference, and the recipient is chosen by our TC. A discussion of our participation in the SDM Award selection was brought up, since there are now more MDO papers at the SDM Conference than Materials papers. Chris said there was once discussion of making the MDO a first-tier participant in the SDM Conference but that was not approved.

ACTION: Jean-Francois will investigate the status of our involvement in the SDM Award selection.

Liaison Subcommittee (Thomas)ISSMO - The society held its first world congress on structural and multidisciplinary optimization (WCSMO) in Goslar Germany in May. Elections for officers and an executive committee were held at the Congress. Rafi Haftka was elected new President of ISSMO. Jarek Sobieski is on the executive committee, and Fred Stritz is on their Benchmarking Committee. ISSMO will cosponsor the next MA&O Conference at Bellevue, and will organize the review of abstracts from Europe. Rafi is preparing for the ISSMO 2nd World Congress to be held in Zakopane Poland June 16-20, 1997. ISSMO would like to see more MDO presence at this congress, and hopes for help from the AIAA MDO TC.. He needs help in promoting the conference, and in making the transition from Structures to MDO. Abstracts are due in Fall 1996. Contact Rafi if you are interested - he especially needs non-structures people.

SDM TC- This TC met in October. Mody Karpel will teach an Aeroelasticity Short Course stressing current applications at the next SDM. The education subcommittee is putting together a high school science fair booklet, citing demonstration projects. The TC is trying to move to Web site.

Applied Aero TC - (June 95 Meeting) Discussed no paper - no podium rule. They decided that the author will be allowed to present if he brings copies of his paper to the meeting. It was reported that the goal of publishing AIAA papers on CDROM and saving approximately $100 per volume looks feasible. (Jan 96 Meeting) Bob Bell AIAA Staff Business Dev. gave a report on the reorganization of AIAA HQ. This was completed six months ago. AIAA is now split into seven teams. TC's are now part of member services. It was noted that attendance was down at Reno this year. The Applied Aero TC has a web site: http://www.eng.ksu.edu/MEDEPT/aerocom/aerocom.html.

ACTION: Jean-Francois will contact AIAA to see if we can receive a presentation on the new organizational structure at a future meeting.

Aircraft Design TC - AIAA sponsors two design competitions each year, one individual and the other a team competition. The design TC is responsible for judging the competition. Problems are picked by the AIAA. The suggestion was made that the MDO TC should contribute a design problem to the competition.

ACTION: John Renaud, chair of the Education Subcommittee, will contact the Design TC to see how



we can help them.

We still need liaison representatives for five other TC's: 1) Materials, 2) Guidance & Control, 3) Standards, 4) Propulsion, and 5) Artificial Intelligence.

Newsletter Subcommittee (Bloebaum)There was a discussion of whether the newsletter should be continued in printed form, since we are trying to use the web as much as possible for communications. Christina would like to find a replacement as editor. It was decided that there will be at least two more printed issues, possibly three, but that there will be a transition to the web. It will still be necessary to have an editor for the web page. This person will be a member of the new Internet Subcommittee

ACTION: Christina will continue the newsletter for two more issues until we convert completely to the web.

Applications Subcommittee (Radovcich)Nick is investigating the idea of putting together a video illustrating MDO applications. This would show many examples of actual optimization problems and the solution process. Perhaps this will help define the meaning of MDO by example, if not formally, to the engineering community at large.

Benchmarking Subcommittee (Striz - not present, presented by Harold Thomas)As mentioned in the Education Subcommittee report, a set of test problems is being developed by NASA-Langley. This now consists of five cases. Venkaya is working to get an MDO model of the Airbus wing which will be donated to the MDO TC by Daimler-Benz. Dr. Krammer of Daimler-Benz has volunteered to donate to us an MDO model of an MBB fin in NASTRAN format. Fred Striz will be coordinating benchmarking activities with ISSMO. There is a session on benchmarking at the MA&O Conference - contact Fred if you can review abstracts or want to participate in a panel discussion.

Publications Subcommittee (Livne - not present, presented by Jean-Francois)Eli has been talking with one of the new TC members about the possibility of publishing a design challenge in the Journal of Aircraft. This has been done before by guidance and control. If you have any suggestions for this contact Eli.

AIAA PROPOSAL - TEST PROBLEM DEVELOPMENT

Conrad Newberry presented a proposal to the TC regarding problem development for the AIAA sponsored Aerospace Engineering Registration Examination. The people who put together this examination need help in coming up with problems/solutions. Conrad proposed that the TC consider development of 1 or 2 problems a year which would be included on the test. The problems developed by the TC should take approximately 30-45 minutes to solve, and are open-book type problems. Conrad will send a package detailing what is required. He estimates the paperwork necessary to document the problem will take about 4-6 hours to put together. The first problem would be due at the end of this year.



AIAA would like for us to commit to one or two problems (with solution) per year. This proposal was referred to the Education Subcommittee for consideration and recommendations.

MA&O CONFERENCE - BELLEVUE

Chris Borland reported on efforts to secure speakers for the conference. He is trying to get Air Force Chief Scientist Dr. Edward Feigenbaum, one of the country's leading experts in artificial intelligence, to speak at the plenary session. Dr. John McMaster of Boeing has agreed to be the luncheon speaker on Thursday. He is an expert in biological fluid mechanics and is famous for his "bird and bee lecture". A Wednesday reception and tour of the Museum of Flight has been set up. Chris is looking for suggestions for the Spouses Program. Chris has brochures available to anyone interested in finding out about the areas attractions. A tour of the Boeing plant is still a possibility, but it seems that accommodating large tour groups is probably not possible. It may be possible to arrange for a tour for TC members only - this would probably take place Saturday after the Conference.

Christina summarized her written planning report for the TC. The Call for Papers went out in the November issue of Aerospace America. In addition, 14,000 copies of the Call were mailed out to members of AIAA, ASME, and previous attendees. AIAA has made arrangements for 6 parallel sessions if we need them - last conference had five. A layout of the rooms for the sessions was included in the report. There was also a list of superchairs, who have responsibility for getting papers reviewed in their areas of expertise A tentative Conference Schedule was also part of the handout. This schedule allows for 228 papers - last time we had 176. Christina also talked about the industry panel sessions planned for the conference. This has worked very well at other conferences when it has been tried. She would like it to cover all industries, not just aerospace. She is soliciting names of people who would make good participants on the panel - if you have ideas contact Christina.

TC OPERATIONS REVIEW

Jean-Francois proposed a restructuring of the TC organization, based on the TC Review which was begun at the last meeting. The TC structure would be divided into three tiers - Planning, Technical, and Communications. Subcommittees would be arranged under these general headings. The TC Chair would head Planning, and the Technical and Communications Branches would be headed by Vice-Chairs. The TC Chair and Vice-Chairs would be elected for two year terms by the TC. After a lively discussion of the proposals, and some modifications, the TC voted on and approved the new organizational structure. Details of the new organization will be available on the TC web site.

TC MEETINGS

Jean-Francois proposed to continue meeting for day-long sessions at each of our three yearly meetings (ASM, SDM, AETOC/MA&O). There was strong resistance from the membership, primarily because this interferes with technical sessions. It was agreed to maintain one full-day (or the equivalent) meeting every other year at the MA&O Conference, and otherwise, to revert to the standard 4 hour evening



session. Each subcommittee is to meet at least once (physically or by teleconference) in-between or before the full committee meetings

MDO MANUAL ON THE WEB

Harold Thomas reported on the status of the TC manual on the web. Included information on the Web site is the operations manual, membership list, subcommittee list, latest newsletter, and latest minutes. These are all hyperlinked together, so you only need to get to one site. The address is http://www.silcom.com/~thomas/MDOTC . Please check out the information on these sites and send any corrections to Harold Thomas at [email protected] and Jean-Francois at [email protected] . Also a picture of all TC members is to be included in the membership list. You can send a picture electronically to Harold (jpg or gif file), or if you do not have scanning capability, send a photograph to Achilles Messac at Northeastern University, who has volunteered to digitize it. In sending material to Harold for the web site, html format is preferred, with ASCII text a second choice. Rafi Haftka has volunteered to maintain a mirror site for the TC files on the web.

MEMBERSHIP ISSUES

Chris Borland reported on membership issues. Ten members are in their last year of the 3-year term on the TC. With only four new US applicants for the TC, it appears that we will not have a problem with too many members, and will remain at the 38-40 level. Although a notice is placed in the June/July Aerospace America issue, not much else has been done about recruiting. The suggestion was made to distribute an announcement of membership availability electronically. The overall makeup of the TC was discussed, with the idea that we should recruit members from other industries, such as automobile manufacturers. Lack of members representing general aviation, which may be coming back to life, was discussed, and Sam McIntosh volunteered to make some contacts in that area. We will make an effort to advertise the TC at the MA&O Conference.

ADDITIONAL TECHNICAL ACTIVITIES

Pradeep Raj was asked by Jean-Francois to serve as the focal point for additional technical activities. Traditionally, we have engaged in some technical activities - benchmarking is one example. Pradeep did not receive any e-mail suggestions before the meeting, but would like to hear suggestions from TC members. He did talk with Venkaya, who is presenting a paper which may be apropos at the SDM Conference. People seem to have their own idea of what MDO is, and this makes it difficult to focus on specific technical issues. Christina suggested that a discussion of what MDO is could be held on the Tuesday night preceding the Bellevue MA&O Conference. We already have a room reserved.

ACTION: Christina will work to set this up.

Another useful technical activity would be to somehow extract a knowledge base from technical presentations at the conferences. AGARD does this, but they have approximately 10% of the number of



papers which we do at our meetings.

The question arose as to whether we should have technical presentations at the TC meetings, and this generated some discussion. It was decided that this would be an appropriate matter to be taken up at the next meeting, after everyone has time to think about it.

ACTION: At the SDM Conference we will address what we should do technically as a TC.

NEW BUSINESS

Farrokh Mistree mentioned that he, Sobieski, and others are putting together a "Gordon" Conference. These conferences are like workshops conducted in a relaxed, reflective manner with time to think about and discuss the presentations which are made. This should take place in Florida during March or April, 1997. The theme is "How Optimization is Used in Engineering". Farrokh will have more information available at the April SDM meeting.

Nick Radovcich posed the question, " Can we identify a project that failed because of lack of MDO?" Jean-Francois stated that this seemed like a perfect topic for the MA&O discussion session.

MEETING SCHEDULE

NOTE THAT THE NEXT MEETING TIME HAS BEEN CHANGED !NEXT MEETING: Monday, Apr.15, 1996 in Salt Lake City (1st day of SDM Conf.) 7:00-10:00 PMFALL MEETING: Thursday & Friday, Sept. 5&6, 1996, 7:00-10:00 PM, MA&O Conference in BellevueWINTER MEETING: Monday, Jan. 6, 1997 in Reno. 7:00-10:00 PM












Latest MDO TC Meeting Minutes( 18 September 1995)

MDO TC Meeting Minutes18 September 1995

Los Angeles

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 18:45 hours after an elegant gourmet meal. Allan Goforth recorded the minutes. Members present were: Sepulveda, Thomas, Livne, Gelhausen, Messac, Briggs, Goforth, Agrawal, and Giesing. Past Chairman Borland was also present. Special thanks to those and their sponsoring organizations for consecutive attendance. The guests and visitors were Todd Beltracchi of Aerospace Corp., Otto Sensberg of Daimler-Benz Aerospace - Germany, Dudley Smith of the University of Oklahoma, Doug Neil of Universal Analytics, and Karl Bradshaw of the AIAA.

UPDATED ROSTER AND SUB-COMMITTEE MEMBERSHIP

An updated roster for the TC and sub-committee membership list was passed out by Jean-Francois. Please review this for accuracy and send corrections to Jean-Francois.

NEW ORLEANS MINUTES REVIEWED, CORRECTED, AND APPROVED

TC PROCESS REVIEW

Jean-Francois presented slides outlining the TC operations, functions, and responsibilities. Some items relating to TC operations are currently being reviewed and were broken down into the following areas:

1. TC REACH - The TC's primary function is to interact with the AIAA by providing input (membership, highlights, awards, ...), by participating in conferences, and by engaging in technical activities (benchmarking, propulsion, applications, ...). The success of the technical activities has been limited. Jean-Francois wants to solicit ideas and opinions about whether we should engage in more technical activities, and if so, what those activities should be. If you have ideas along these lines, please contact Pradeep Raj, who will be the focal point in this area.

2. TC OPERATIONS - Most work is done by individuals in intense work periods before meetings. Should we change our mode of operation? How many times should we meet, and for how long? Can we maximize effectiveness during meetings and minimize travel requirements? Jean-Francois will act as the focal point in this area, so please contact him with your ideas.

3. TC MANUAL - The TC manual needs to be rewritten. Current manual is a collection of disjointed parts.

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_min.sep95.html (1 of 5)12/29/2006 12:29:20 PM


4. TC MEMBERSHIP - Process of selecting members is being reviewed. Should we change our approach to selecting/rotating members? This process should broaden committee diversity, including representation by discipline, and representation by industry/academia/government. Please contact Chris Borland with any ideas on this. The makeup of the committee was further discussed - committee has been dominated by structural specialists in the past. It was suggested that the discipline of costing has an important place in MDO and possibly should have two sessions at the next MA&O conference.

MA&O CONFERENCE - SEATTLE

Karl Bradshaw from the AIAA is the new Meeting Director for the biennial Seattle MA&O Conference (replacing Julie Walker) , so we will be working closely with him in order to bring the meeting together. Karl's phone number is (202) 646-7528, and his email address is [email protected] . A conference planning report was passed out by Chris. Call for papers should be coming out in the November (possibly October) Aerospace America. AIAA has made arrangements for 6 parallel sessions if we need them - last conference had five. Included in the report was a list of superchairs, who have responsibility for getting papers reviewed in their areas of expertise. Anyone wanting to be added to this list, contact Christina Bloebaum ([email protected]). Christina is the Technical Chair of the Conference. Anyone with ideas concerning speakers at the conference, please contact Chris Borland.


Planning Subcommittee(Barthelemy)The inclusion, on the planning subcommittee, of the chairs of the various subcommittees was discussed as a new approach to facilitating communication within the TC. Chairs will participate in the planning committee's activities with the intent of representing the membership of their subcommittees. Proposals for activities or policies may be drafted in the planning subcommittee for further consideration, modification or reframing by the whole TC. The planning subcommittee considered the review of four areas of TC activities. For each area, one focal point was identified. The focal point is to develop a draft proposal for his area of review to be available to the membership, both on the Web and by E-mail, by the middle of November. Full membership involvement begins now, by sending suggestions to the focal points. Once the proposals are available, the membership then has another month to comment. The focal points are then to finalize their proposals which they will present for approval at our Reno meeting and submit for an up or down vote. Proposed modifications will then be documented in our TC manual. While each review must consider all aspects of the activity considered, this is not an exercise in change for the sake of change, rather it is an opportunity to look at what we are doing, how we are doing it, and to determine if there is any value added in changing some aspects of the way we do business. To avoid devoting too much of our energy to this exercise, we will essentially conclude it at our next meeting

The four areas for review are as follows:

1. TC OPERATIONS: Barthelemy will be the lead for this review. The objective is to address TC



organization, TC meeting schedule and format and any other issue that could affect the productivity of the TC.

2. TC TECHNICAL ACTIVITY INVOLVEMENT: Raj will be the lead for this review. The TC has had several attempts at carrying out technical activities. At this point we have benchmarking which looks at standard test problems, as well as education which puts together short technical courses, publications which deals with highlights and special issues of journals and conference support where we organize technical sessions or review papers. The last three activities are traditional TC involvement and there is a sense from the membership that more technical activities are warranted. Yet previous attempts have not always succeeded.

3. TC MEMBERSHIP: Borland will be lead for this review. Memberships begins with inviting qualified technical people to apply, selecting new members so that the TC's membership becomes more diverse, represents better the different disciplines at play in aerospace design and balances properly representation by industry, academia and government.

4. TC OPERATIONS MANUAL: Thomas will be lead for this review. Our current manual, now available on the Web, is quite incomplete and sometimes quite inaccurate. We will review the outline of the manual, its content, as well as the process by which we keep it current.

Awards Subcommittee (Thomas)The purpose of the subcommittee and the process of upgrading membership status for the AIAA was discussed. The call for nominations for the MDO award will be in the October Aerospace America issue.

Benchmarking Subcommittee (Striz - not present presented by Dudley Smith)A set of test problems is being developed by NASA-Langley. These should be available for benchtesting on the World Wide Web around the time of the 1996 SDM Conference. An MDO model of the Airbus wing will be donated to the MDO TC. Discussions with the AIAA are being held to see if we can locate some of the larger benchmarking models on AIAA computers, so individual companies will not have to supply these resources ( disk space, etc.). The question of the format necessary and the standards to be applied for benchmark testing was discussed. ACTION: Todd Beltracchi of Aerospace Corp. will get a copy of his paper on this subject to Fred Striz.

Conference Support Subcommittee (Santangelo - not present)A representative for the 1997 ASM conference, who must also attend the planning session at the 1996 meeting, was needed. Shreekant Agrawal volunteered for this task. In addition, ________ volunteered to provide support at the 1996 Design Conference in Los Angeles. ACTION: Jean-Francois will contact Andrew Santangelo concerning these conference support representatives.

Liaison Subcommittee (Thomas)Representatives are needed for Materials, GNC, Standards, Propulsion, and AI. Reports were not available from the representatives to other TC's. ACTION: Thomas will contact liaison members and assemble reports.



Publications Subcommittee (Livne)Annual Aerospace America article on MDO has been done. Discuss ensued about possibility of pushing for a dedicated issue of AIAA Journal or Journal of Aircraft dedicated to MDO. A joint collaboration of the Publications and Benchmarking Subcommittees was proposed to establish and publish a standard set of problems and solutions for MDO problems. ACTION: Eli will contact Fred Striz. They will look into what's involved, and Eli will report on conclusions at Reno. Farrokh Mistree has written a paper which has been reviewed by several TC members. There was a discussion about how to publish this - should it be an MDO TC paper, or should Farrokh publish it? ACTION: Jean-Francois will contact Farrokh and discuss the issue.

Newsletter Subcommittee (Bloebaum - not present)We want to make an effort to get the newsletter out within a month of each TC meeting.

Education Subcommittee (Renaud - not present)A report was provided by John. The following items were mentioned: 1) A session/panel on engineering education at the 1996 MA&O will be held. John and Christina will be coordinating. Contact them if your are interested in participating. 2) Planning is underway for an AIAA workshop on work-flow management to be given at the MA&O Conference. 3) ISSMO Education Subcommittee wants to coordinate activities with our Education Subcommittee.

Applications Subcommittee (Radovcich - not present)Nick is probably going to be the new chairman.ACTION: Jean-Francois will contact Nick for report.

MDO TC INFORMATION ON THE WORLD WIDE WEB

Harold Thomas distributed information concerning Web access for TC members. All TC members should obtain a copy of this (perhaps it will appear elsewhere in the newsletter). Included information on the Web site is the operations manual, membership list, subcommittee list, latest newsletter, and latest minutes. These are all hyperlinked together, so you only need to get to one site. The operations manual address is http://www.neptune.net/~thomas/MDOTC_man.html . Please check out the information on these sites and send any corrections to Harold Thomas at [email protected] and Jean-Francois at [email protected] . Since Harold is providing this service to the TC, he asked to be allowed to include a message saying that his consulting firm is sponsoring this page. A motion to this effect was made, seconded, and approved by the TC.

NEW BUSINESS

There is a meeting on MDO September 26, at NASA-Langley which should be of interest to our TC.

MEETING SCHEDULE


http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_man.html


NOTE THAT THERE IS A CHANGE IN OUR USUAL SCHEDULE !NEXT MEETING: Thursday, Jan. 18, 1996 in Reno (Last day of ASM Conference) - meeting tentatively scheduled for the whole day Thursday.SPRING MEETING: Wednesday, April 17, 1996 in Salt Lake City - meeting tentatively scheduled for the whole day Wednesday.FALL MEETING: Sometime during the week of the MA&O Conference, September 4-6, 1996 in Seattle.

The meeting was adjourned at 2305 hours.









MDO TC Subcommittee Reports

MDO TC Subcommittee Reports

● Applications Subcommittee Reports ● Awards Subcommittee Reports ● Education Subcommittee Report: January 2002 ( PowerPoint File ) ● Publications Subcommittee Reports

● MA&O Symposium Support Subcommittee Reports


Last Updated: 02 January 2002

Tony Giunta [email protected]

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_rep.html12/29/2006 12:29:23 PM

http://endo.sandia.gov/AIAA_MDOTC/communications/mdotc_edu_subcom_jan02.ppt


MDO TC Newletters

MDO TC Newletters

NOTE: The June 1996 newsletter is the final edition. The newsletter effort has been discontinued in favor of disseminating information directly from the MDO TC Web site.

● MDO TC Newletter No. 22, June 1996

● MDO TC Newletter No. 21, April 1996

● MDO TC Newletter No. 20, Dec. 1995

● MDO TC Newletter No. 19, July 1995

● MDO TC Newletter No. 18, April 1995


Last Updated: January 10, 1997


http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_new.html12/29/2006 12:29:23 PM



MDO TC Newsletter No. 20: Dec 1995


FROM THE EDITOR: Christina L. Bloebaum

Happy Holidays Everyone! I hope this finds you all enjoying your holiday season and preparing your abstracts for the next MA&O Conference! Please remember that the deadline for full paper abstract submittals is January 19th and the deadline for WIP abstracts is March 1st. Also remember that while electronic submissions are possible (as described in the call), only text format is permitted. Also remember that George Rozvany has kindly agreed to serve as European Paper Coordinator which means those who are closer to Germany can submit to his address rather than mine in Buffalo. Look forward to seeing you in Reno and remember to submit interesting items to the newsletter.

CHAIRMAN'S CORNER: Jean-Francois Barthelemy

Our TC has been in operation for over six years and has produced a string of impressive accomplishments. From the running of successful conferences to special publications and short courses. It is time for us to take stock and assess where we have been, where we are and where we are going.

We are reviewing our operations to better serve the MDO community at large. You will read the details in the minutes further in this newsletter. Essentially, Chris Borland ([email protected]) is reviewing our membership process, Harold Thomas ([email protected]) our TC manual and its distribution, Pradheep Raj ([email protected]) the TC technical involvement and I am reviewing the TC operations. Each reviewer is to provide a proposal for comments in advance of the Reno meeting. The proposals will be distributed electronically for comments by all. The comments are to be factored in modified proposals to be presented at our upcoming meeting in Reno. I strongly encourage you to review the proposals and to offer your comments to the reviewers. Should you want to offer inputs on any of those subjects, please do contact the lead reviewer directly.

Our next committee meeting is Thursday 18 January 1996, the last day of the Aerospace Sciences Meeting in Reno, NV. Check in Aerospace America for information about hotels. We will need to spend most of the day in plenary or subcommittee meetings. While I intend to distribute an agenda closer to the meeting, we will be devoting one plenary meeting to reviewing the proposals to modify our operations as a TC and another one to conduct normal TC business. The remaining of the day is set aside for subcommittee meetings. I am exploring with AIAA opportunities for a social event at the end of the day.

I am looking forward to seeing you in Reno.

Jean-Francois Barthelemy 804/864-2809 (phone)

http://endo.sandia.gov/AIAA_MDOTC/communications/MDOTC_new.20.html (1 of 7)12/29/2006 12:29:25 PM


804/864-9713 (FAX) [email protected]

MINUTES OF THE MDO TC MEETING: LA, California, September 18, 1995

PRELIMINARIES

Chairman Jean-Francois Barthelemy called the meeting to order at 18:45 hours after an elegant gourmet meal. Allan Goforth recorded the minutes. Members present were: Sepulveda, Thomas, Livne, Gelhausen, Messac, Briggs, Goforth, Agrawal, and Giesing. Past Chairman Borland was also present. Special thanks to those and their sponsoring organizations for consecutive attendance. The guests and visitors were Todd Beltracchi of Aerospace Corp., Otto Sensberg of Daimler-Benz Aerospace - Germany, Dudley Smith of the University of Oklahoma, Doug Neil of Universal Analytics, and Karl Bradshaw of the AIAA.

UPDATED ROSTER AND SUB- COMMITTEE MEMBERSHIP

An updated roster for the TC and subcommittee membership list was passed out by Jean-Francois. Please review this for accuracy and send corrections to Jean-Francois.

NEW ORLEANS MINUTES REVIEWED, CORRECTED, AND APPROVED

TC PROCESS REVIEW

Jean-Francois presented slides outlining the TC operations, functions, and responsibilities. Some items relating to TC operations are currently being reviewed and were broken down into the following areas:

1. TC REACH - The TC's primary function is to interact with the AIAA by providing input (membership, highlights, awards, ...), by participating in conferences, and by engaging in technical activities (benchmarking, propulsion, applications, ...). The success of the technical activities has been limited. Jean-Francois wants to solicit ideas and opinions about whether we should engage in more technical activities, and if so, what those activities should be. If you have ideas along these lines, please contact Pradeep Raj, who will be the focal point in this area.

2. TC OPERATIONS - Most work is done by individuals in intense work periods before meetings. Should we change our mode of operation? How many times should we meet, and for how long? Can we maximize effectiveness during meetings and minimize travel requirements? Jean- Francois will act as the focal point in this area, so please contact him with your ideas.

3. TC MANUAL - The TC manual needs to be rewritten. Current manual is a collection of disjointed parts.

4. TC MEMBERSHIP - Process of selecting members is being reviewed. Should we change our



approach to selecting/rotating members? This process should broaden committee diversity, including representation by discipline, and representation by industry/academia/government. Please contact Chris Borland with any ideas on this. The makeup of the committee was further discussed - committee has been dominated by structural specialists in the past. It was suggested that the discipline of costing has an important place in MDO and possibly should have two sessions at the next MA&O conference.

MA&O CONFERENCE - SEATTLE

Karl Bradshaw from the AIAA is the new Meeting Director for the biennial Seattle MA&O Conference (replacing Julie Walker) , so we will be working closely with him in order to bring the meeting together. Karl's phone number is (202) 646-7528, and his email address is [email protected] . A conference planning report by Bloebaum was passed out by Chris. Call for papers should be coming out in the November (possibly October) Aerospace America. AIAA has made arrangements for 6 parallel sessions if we need them - last conference had five. Included in the report was a list of superchairs, who have responsibility for getting papers reviewed in their areas of expertise. Anyone wanting to be added to this list, contact Christina Bloebaum ([email protected]). Christina is the Technical Chair of the Conference. Anyone with ideas concerning speakers at the conference, please contact Chris Borland.


Planning Subcommittee (Barthelemy)The inclusion, on the planning subcommittee, of the chairs of the various subcommittees was discussed as a new approach to facilitating communication within the TC. Chairs will participate in the planning committee's activities with the intent of representing the membership of their subcommittees. Proposals for activities or policies may be drafted in the planning subcommittee for further consideration, modification or reframing by the whole TC. The planning subcommittee considered the review of four areas of TC activities. For each area, one focal point was identified. The focal point is to develop a draft proposal for his area of review to be available to the membership, both on the Web and by E-mail, by the middle of November. Full membership involvement begins now, by sending suggestions to the focal points. Once the proposals are available, the membership then has another month to comment. The focal points are then to finalize their proposals which they will present for approval at our Reno meeting and submit for an up or down vote. Proposed modifications will then be documented in our TC manual. While each review must consider all aspects of the activity considered, this is not an exercise in change for the sake of change, rather it is an opportunity to look at what we are doing, how we are doing it, and to determine if there is any value added in changing some aspects of the way we do business. To a! void devoting too much of our energy to this exercise, we will essentially conclude it at our next meeting.

The four areas for review are as follows:

1. TC OPERATIONS: Barthelemy will be the lead for this review. The objective is to address TC organization, TC meeting schedule and format and any other issue that could affect the



productivity of the TC. 2. TC TECHNICAL ACTIVITY INVOLVEMENT: Raj will be the lead for this review. The TC

has had several attempts at carrying out technical activities. At this point we have benchmarking which looks at standard test problems, as well as education which puts together short technical courses, publications which deals with highlights and special issues of journals and conference support where we organize technical sessions or review papers. The last three activities are traditional TC involvement and there is a sense from the membership that more technical activities are warranted. Yet previous attempts have not always succeeded.

3. TC MEMBERSHIP: Borland will be lead for this review. Memberships begins with inviting qualified technical people to apply, selecting new members so that the TC's membership becomes more diverse, represents better the different disciplines at play in aerospace design and balances properly representation by industry, academia and government.

4. TC OPERATIONS MANUAL: Thomas will be lead for this review. Our current manual, now available on the Web, is quite incomplete and sometimes quite inaccurate. We will review the outline of the manual, its content, as well as the process by which we keep it current. Awards Subcommittee (Thomas) The purpose of the subcommittee and the process of upgrading membership status for the AIAA was discussed. The call for nominations for the MDO award will be in the October Aerospace America issue.

Benchmarking Subcommittee (Striz - not present presented by Dudley Smith)A set of test problems is being developed by NASA-Langley. These should be available for benchtesting on the World Wide Web around the time of the 1996 SDM Conference. An MDO model of the Airbus wing will be donated to the MDO TC. Discussions with the AIAA are being held to see if we can locate some of the larger benchmarking models on AIAA computers, so individual companies will not have to supply these resources ( disk space, etc.). The question of the format necessary and the standards to be applied for benchmark testing was discussed. ACTION: Todd Beltracchi of Aerospace Corp. will get a copy of his paper on this subject to Fred Striz.

Conference Support Subcommittee (Santangelo - not present) A representative for the 1997 ASM conference, who must also attend the planning session at the 1996 meeting, was needed. Shreekant Agrawal volunteered for this task. In addition, Paul Gelhausen volunteered to provide support at the 1996 Design Conference in Los Angeles. ACTION: Jean-Francois will contact Andrew Santangelo concerning these conference support representatives.

Liaison Subcommittee (Thomas)Representatives are needed for Materials, GNC, Standards, Propulsion, and AI. Reports were not available from the representatives to other TC's. ACTION: Thomas will contact liaison members and assemble reports.

Publications Subcommittee (Livne)Annual Aerospace America article on MDO has been done. Discuss ensued about possibility of pushing for a dedicated issue of AIAA Journal or Journal of Aircraft dedicated to MDO. A joint collaboration of the Publications and Benchmarking Subcommittees was proposed to establish and publish a standard set



of problems and solutions for MDO problems. ACTION: Eli will contact Fred Striz. They will look into what's involved, and Eli will report on conclusions at Reno. Farrokh Mistree has written a paper which has been reviewed by several TC members. There was a discussion about how to publish this - should it be an MDO TC paper, or should Farrokh publish it? ACTION: Jean-Francois will contact Farrokh and discuss the issue.

Newsletter Subcommittee (Bloebaum - not present) We want to make an effort to get the newsletter out within a month of each TC meeting.

Education Subcommittee (Renaud - not present) A report was provided by John. The following items were mentioned: 1) A session/panel on engineering education at the 1996 MA&O will be held. John and Christina will be coordinating. Contact them if your are interested in participating. 2) Planning is underway for an AIAA workshop on work-flow management to be given at the MA&O Conference. 3) ISSMO Education Subcommittee wants to coordinate activities with our Education Subcommittee.

Applications Subcommittee (Radovcich - not present)Nick is probably going to be the new chairman. ACTION: Jean-Francois will contact Nick for report.

MDO TC INFORMATION ON THE WORLD WIDE WEB

Harold Thomas distributed information concerning Web access for TC members. All TC members should obtain a copy of this (perhaps it will appear elsewhere in the newsletter). Included information on the Web site is the operations manual, membership list, subcommittee list, latest newsletter, and latest minutes. These are all hyperlinked together, so you only need to get to one site. The operations manual address is:

http://www.neptune.net/~thomas/MDOTC.html

Please check out the information on these sites and send any corrections to Harold Thomas at [email protected] and Jean-Francois at [email protected]. Since Harold is providing this service to the TC, he asked to be allowed to include a message saying that his consulting firm is sponsoring this page. A motion to this effect was made, seconded, and approved by the TC.

NEW BUSINESS

There is a meeting on MDO September 26, at NASA-Langley which should be of interest to our TC.

MEETING SCHEDULE

NOTE THAT THERE IS A CHANGE IN OUR USUAL SCHEDULE!


http://www.neptune.net/~thomas/MDOTC.html




NEXT MEETING: Thursday, Jan. 18, 1996 in Reno (Last day of ASM Conference) - meeting tentatively scheduled for the whole day Thursday.

SPRING MEETING: Wednesday, April 17, 1996 in Salt Lake City - meeting tentatively scheduled for the whole day Wednesday.

FALL MEETING: Sometime during the week of the MA&O Conference, September 4-6, 1996 in Seattle.

The meeting was adjourned at 2305 hours.

Respectfully submitted on 10/25/95,

Allan Goforth Lockheed Martin Skunk Works Dept. 25-45, Bldg. 611 1011 Lockheed Way Palmdale, Ca. 93599-2545 Tel: (805) 572-7205 Fax: (805) 572-7679 Home email: [email protected] email: [email protected]

Please send your comments and contributions for the next MDO TC Newsletter to: Professor C. L. BloebaumDepartment of Mechanical and Aerospace Engineering1009 Furnas Hall State University of New York at BuffaloBuffalo, New York 14260 email address is: [email protected] phone is: (716) 645-2593 x2231departmental phone is: (716) 645-2593FAX is: (716) 645-3875

AIAA MULTIDISCIPLINARY DESIGN OPTIMIZATION TECHNICAL COMMITTEE 1994/95

Chairman: Jean-Francois Barthelemy



Members: Shreekant Agrawal Kumar G. Bhatia Christina L. Bloebaum Carl M. Bosch Clark Briggs Robert Canfield Peter C. Coen Mark Drela Mark S. Ewing George Fadel Paul Gelhausen Omar Ghattas Joseph Giesing Edward A. Goforth Ramana V. Grandhi Raphael Haftka Srinivas Kodiyalam Johann Krammer Charles Lawrence Eli Livne Samuel C. McIntosh Achilles Messac Farrokh Mistree S. N. B. Murthy Henry Neimeir Jerry Newsom John R. Olds Paul Pierson Nick Radovchich Masoud Rais-Rohani Pradeep Raj David C. Redding John E. Renaud Gerald Seidel Abdon E. Sepulveda Gary Stanley Alfred G. Striz Harold Thomas Rudy Yurkovich

Assoc. Member: Andrew Santangelo

Past Chairman: Christopher Borland

Back to Newsletter list







MDO TC Newsletter No. 19: July 1995


FROM THE EDITOR

Christina L. BloebaumHello All! This newsletter is a bit bare of miscellany but packed with minutes. Please remember that you can email anything at all to me that you think might be of interest to the MDO community and I will put it in the newsletter. You can also send it to me via snail mail (by post office, that is) or FAX. Remember that the next Multidisciplinary Analysis and Optimization Symposium is only a little more than a year away. The Call for Papers will be sent out within the next few weeks so that you can start preparing those abstracts. Be sure you read the instructions VERY carefully, since any abstracts that are not responsive to the Call will be rejected.

CHAIRMAN'S CORNER

Jean-Francois BarthelemyThis first `Chairperson's Corner' of my tenure as TC Chair must begin with a heartfelt Thank You for your confidence. I am honored to have been selected to chair the committee and you have my pledge that I will do so with all my energy. As I indicated in my brief remarks in New-Orleans, I believe this committee to be yours, I know that the membership will make the difference into what the committee achieves and contributes. I therefore see my responsibility as one of coordination and facilitation as well as communication with AIAA.

The committee has had impressive accomplishments since its creation in 1989. Under Jarek Sobieski's pioneering leadership, it sprang from an idea in a few people's minds to a coherent organization. It produced a number of important publications, helped run very successful biennial meetings in San Francisco and Cleveland and participated in several other respected forums. The membership has progressively evolved from a dedicated group of mostly structural optimization specialists to an eclectic organization of multidisciplinary and disciplinary experts. Chris Borland's tenure has seen expansion in these areas, particularly the very successful Panama City MA&O meeting and the planning for the upcoming Bellevue MA&O meeting. The MDO community is getting more coverage in the technical press and our second AIAA highlight is being written.

In the words of Chris Borland, we are indeed in interesting times as the aerospace industry that we mostly serve is undergoing tremendous reshaping. There is clear consensus that continued profitability is keyed to a multidisciplinary approach to problem solving, whether in performance design, life-cycle cost engineering or in process management. Our committee is poised to be a major contributor in this



reshaping. I believe that it is timely that we reflect on where we have been, where we are and where we are going and I intend to devote our next meeting to beginning this process. I believe that we must review the mission of the committee as well as its operations. It is an appropriate time for us to look at all our activities, decide which have served their purpose, which need to continue as planned, which need to be carried on differently and which need to be started. While I want the Planning Subcommittee to facilitate this exercise, I want it to be clear that every committee member is expected to participate. I urge all of you to attend our next meeting prepared with your reflections and suggestions.

To enhance this process, I am enlarging the Planning Subcommittee membership to include its current members and all subcommittee chairs. I see this as a means to providing a better communication mechanism within the committee as each subcommittee carries out the work in this review while the Planning Subcommittee proposes a process and integrates the suggestions. The results of this review are likely to require changes in our operations and I want those documented in our operations manual. Harold Thomas is working on getting an updated version of the operations manual available on WWW. I consider this version a draft that will be modified by the result of our review and should be finalized as soon as possible, no later than when our membership is renewed in the Spring of 1996. Many thanks to Harold for taking the initiative on this. While our committee reviews its mission and operations, normal business will continue and planning is well underway for the next MA&O conference in Bellevue, WA 4-6 September 1996. Chris Borland will chair and Christina Bloebaum will serve as Technical Program Chair. The call for paper will be available shortly and mailed by AIAA. I am grateful that chairs from previous MA&O have volunteered their expertise and agreed to serve on an Organizational Support Committee while Prof. George Rozvany has offered to coordinate European contributions.

Our next committee meeting is Monday 18 September 1995, the evening before the beginning of the Aircraft Technology and Operations Congress in the Sheraton Hotel (check AIAA announcements for details). The Planning Subcommittee is to meet between 16:00 and 18:00, while the full membership meeting will be 18:00-22:00. I am looking forward to seeing you there.

Jean-Francois Barthelemy 804/864-2809 (phone) 804/864-9713 (FAX) [email protected]


Reno, Nevada, January 9, 1995 The meeting was called to order at 1837 hours by chairman Christopher Borland. Nick Radovcich recorded the minutes. For member, guests and visitors present, see attached list. Special Thanks to those and their sponsoring organizations for consecutive attendance.

MEMBERSHIP CHANGES AND ELECTIONS

Join a Sub-committee.



PRESENTATION

David Riley of the TAC Finance Committee gave a presentation on "Everything you always wanted to know about AIAA Finances (and were glad you didn't ask) or "Where does all the money go?". The bottom line issues revolved around the base and the cost added to the base. The base budget has cyclic highs and lows. David suggested that much could be accomplished if the lows could be filled in with other functions which will attract support from the members. The cyclic nature of the attendance are not well understood. The even year and location of a conference are just some of the factors. The added cost (to the base) comes from the organizing committee requested services. Costs like slide projectors, VTC's, etc. are supplied by a third party supplier or by the facilities where the conference is being held. A simple request like refreshments in the afternoon sessions comes to 2-4 $ per serving. The added costs are increasing.

INTRODUCTION OF MEMBERS AND VISITORS

PANAMA CITY MINUTES REVIEWED, CORRECTED, AND APPROVED ROSTER AND SUB-COMMITTEE MEMBERSHIP WILL BE UPDATED

Chris passed out AIAA MDO TC Subcommittees brochure. He asked for input for corrections.


Conference Support (Santangelo)Aerospace Sciences 95 - Santangelo, MDO first half in the morning was well attended. The Wed session needs help because four people bailed out. Much discussion on the form of paper versus informal presentations. No conclusions except to note that AIAA policy is no paper no podium.

37th SDM April 10-12, 1995 - New Orleans - Bloebaum (not present; no report).

1996 Design Technical Conference - Minneapolis (? not listed in the AIAA Meeting Schedule) - John Renaud will be the liaison.

1995 1st Aircraft Engineering, Technology, and Operations Congress in Los Angeles Mariott Hotel Airport Sept 19-21 - This could be the site of our Fall meeting.

Current MDO Symposium; Sept 7-9, 1994 in Panama City, Florida - 5th AIAA/AF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization. Sobieski General Chair, Berke, Borland, Rozvany, AA). 110 papers were from academia.

MDO Symposium 1996 - Borland reporting: Christina Bloebaum is the Technical Chairperson. Received much material from Sharon Padula's Panama City team. The Jackson Hole site will not work.



Working with Julia Walker on Bellevue Hyatt near Seattle. Conflict with ASME 9/23-27. Next target is the 9/16-20 with a possible switch with ASME date. Working on Boeing supporting a reception at Museum of Flight and tour. This will be in stead of a boat ride. The next option is a tour of the Boeing Everett Plant. This is a major undertaking and would add another day especially for those from the East coast.

Liaison-StraubNeeds a lot of help- many vacant posts; Straub reported that he will be leaving after the SDM meeting.

Structural Dynamics TC - no report GNC: no report Air Breathing Propulsion: no report Propulsion TC: no report Structures TC: no report HSCT: ?

Education- Renaud (not present - Todd reported)Remind everyone of the call for papers due 2/1/95. Chris said he did not know of any offerings.

Publications - Salama (not present Borland reported)Many thanks to Eli Livne for his Highlights Article in the Aerospace America. A discussion about distributing material via CD's . Two issues - software and practical considerations for updates required in iterations.

Newsletter - Bloebaum (not present)All MDO newsletters will be on the world wide web aero.stanford.edu/mdotc.html.

Awards- Neill (not present) Thomas reported.Associate fellow nominations are not as competitive as first assumed. However, fellow nominations are very competitive. Senior members elections are almost automatic; Some discussion as to what that really means. Anyway, the process is proceeding.

Planning- BorlandApplications for Membership (May 95 - April 98) are in. 14 Received with a good mix of industry, government and academia. However, the applications are heavy on Structures and light on other disciplines. There are 14 members leaving (unless request to continue). If we accept all the applications, MDO will be over the 35 limit. However, this does not appear to be a problem at AIAA.

Benchmarking- Venkayya (not present)Beltracci says that real progress in bench marking has be slow at best. The direction of this activity is not clear.



Reporting Guidelines (sub-committee of one , and after leaving - zero) - integrating the comments of three reviewers. Good samples at a case level is required. Todd's background is in Mechanical Engineering and feels that he could cover that. He has little understanding of thermodynamics and structure optimization.

Guidelines for engineering results - we need people with aircraft design experience. There is a lack of background to establish standards here. However, some has accepted Todd's paper as the guideline for students to report results. Chris- is there room to have an educational guideline? Todd - yes, he has an example with source code. The main focus is accessibility. Discussed possibility of working this form into MDO papers. This issue will be further explored in the next call for MDO papers. WE NEED MEMBERS OF THIS TC TO CARRY ON THIS WORK AFTER TODD LEAVES.

Applications- Coen (not present)no report

Propulsion - new chair Chriswith limited travel budgets, electronic communication between sub-committee members will probably be required in the planning of their activities.

PROPOSAL(Raj Pradeep)

● New Membership "Care Package": Outline of TC/TAC Framework History, mission, vision ● Sub-committee structure with charters ● Roster (proposal for care package accepted; include with acceptance letter) ● New Membership Selection Committee Review Application, Maintain and Update Roster (in

place) ● Planning Sub-committee redefined Future activities - conferences, etc. (in place; as is) ● Present and review 3-year plan once/year ( no action items) ● Recommend and investigate Short Courses ● Publications of Proceedings on CD-ROM ( already discussed) ● Liaison and Subcommittee reports mailed ahead of time ● Discussed - another candidate for World Wide Web - need a directory and how to use. The

liaison person would review newsletters from other TC and produce a one page summary to be included in the MDO newsletter. This could make the time at TC meeting more productive.

CHAIRMAN ANNOUNCEMENTS

none

NEW BUSINESS



none

TC next meeting; time and place-The next MDO TC meeting was set for 1930 hours Monday April 10, 1995 at New Orleans- SDM.

The meeting was adjourned 2207 hours.

Respectfully submitted on 3/30/95Nick Radovcich Lockheed Aeronautical Systems Co. Dept 73-HA, Zone 0988 Marietta, GA 30063 Tel: (404) 793-0008 Fax: (404) 793-0149


New Orleans, LA, April 10, 1995

The meeting was called to order at 1930 hours by chairman Christopher Borland. Nick Radovcich recorded the minutes. For member, guests and visitors present, see attached list. Special Thanks to those and their sponsoring organizations for consecutive attendance.

MEMBERSHIP CHANGES AND ELECTIONS

Chris Borland, following the steps of Jarek Sobieski, passed the gravel to Jean Barthelemy. Chris assumes the Past-Chairman. Many thanks for a job well done and making the TC better. Many Thanks to the members ( and their supporting organizations) for their service on the TC: Sobieski (NASA), Beltracchi (Aerospace Corp), Berke (NASA), Gallman (NASA), Harry (TRW), Hill (NASA), Kroo (Stanford), Neill (Universal Analytics), Salama (JPL), Straub (McDD), Taylor (Old Dom. U.), Venkayya (AF Wright L), Wainfan (Mc DD); Associates: Bell (Aerospace Corp), Sikes ( PDA Eng). Welcome to the new Members: Agrawal (McDD), Briggs (JPL), Canfield (AFIT/ENY), Fadel (Clemson U.), Ghattas (CMU), Giesing (McDD), Haftka (UofF), Kodoyalam (GE), Lawremce (NASA), Messac (NorthE. U.), Pierson (LockMart), Rais-Rohani (MissStU), Seidel (NASA), Stanley (LockMart), Thomas (StrOptSpec), Santangelo (MichTechCorp). Harold Thomas became the new chair for Liaison Subcommittee, replacing Straub and also Awards replacing Neill. Clark Briggs became the new chair for Publications replacing Salama. Alfred Striz with Raphael Haftka became co-chairs for Benchmarking replacing Venkayya.

PRESENTATIONS

none



INTRODUCTION OF MEMBERS AND VISITORS

RENO MINUTES REVIEWED, CORRECTED, AND APPROVED

Minutes were distributed at the meeting. They were not in the newsletter.

ROSTER AND SUB-COMMITTEE MEMBERSHIP WILL BE UPDATEDS

Chris passed out Roster for update.


6th MA&O (1996) - Bloebaum (Tech-chair)/Borland (General Chair) Bloebaum reports

● Around middle of January all the abstracts will be collected, sorted out, and sent out to the super chairs. The super chairs will then distribute them for reviews in twenty categories. A sign-up sheet for super chairs (10-15) was distributed. Christina will contact the super chairs in September.

● Concerned about drop off in attendance during the last days of technical presentations and the attraction of one paper which drains off the attendance from the other papers. Wants inputs about selecting two or three papers to be presented without competition, and about work in progress. The projected growth of the conference requires the consideration of expanding multiple sessions, providing for shortened presentations, or extending the length of the conference. There was extensive discussions on the introduction of 20 and 30 minute presentations; increasing the rejection rates and keeping the numbers the same as last time, having a poster session (no), and a 1-2 hour overview from industry (yes). There was no significant discussion on lengthening the conference. The members are urged to consider these options and reply back to Chistina before next meeting.

● Chris - a tour of the Boeing plant at Everett is planned for Friday afternoon or maybe Saturday. Airline schedules for going East from Seattle Friday night are not many except for some red-eyes.

Conference Support - Santangelo (not present - Borland reporting)

● 1995 1st Aircraft Engineering, Technology, and Operations Congress in Los Angeles Mariott Hotel Airport Sept 19-21 - Borland. Had one MDO abstract in the High Speed Civil Transport to give a flavor of the MDO process.

● 1996 34th AIAA Aerospace Sciences Meeting in Reno Jan 15-18 - Santangelo rep. ● 1996 37th Structures/SDM Conference in Salt Lake City Apr 15-17 ; Adapt Structures and

Dynamics Specialist Apr 18-19 - Mistree/Grandhi rep. attending the organizing meeting at this time. Christina reported on the interactive plenary sessions of which MDO had 7 papers. These were selected on basis of their high quality which were germane to the subject. There were first



and second place awards of $500 and some software.

Liaison-Thomas

● Currently has a list of 5 members on this sub-committee. Will report on Structure Dynamics Oct 13 meeting. They were unhappy with the poster sessions. Christina had reports that these were the best of the papers. No resolution of the different field of visions.

● Assignments - Fred Straub for SDM Paul Gelhausen for A/C John Renaud for GNC Rudy Yurkovich for Applied Aero Open - Structures and Propulsion, John Renaud - Roster update

Education- Renaud (not present)Bloebaum reported, written report distributed.

1. Guidelines for Reporting Engineering Optimization Results (T. Beltracchi) - Todd reports that Kroo will put a copy of this guideline on the WWW at Stanford and it was submitted as a paper to next fall's ASME design Automation Conference.

2. MDO Test Problem - Larry Green ([email protected]) and Natalia Alexandrov ([email protected]) are compiling a suite of MDO test problems in a problem description form (equations, analysis tools used, etc.) Notre Dame has submitted 10. This project is in the early stage. Contact the principles for access or for making contributions.

3. 1996 MDO Conference will include a session/panel on Engineering Education. (See Christina's Comments in the MDO newsletter) There are five programs funded by NASA for industry interaction.

4. Thanks to those committee members who submitted MDO papers to the 1995 ASME Design Automation Conference, Boston next September and who participated in the review process.

Publications- Salama (not present Borland reported)Repeated the many thanks to Eli Livne for his Highlights Article in the Aerospace America. A discussion about distributing material via CD's. Two issues - software and practical considerations for updates required in iterations.

Newsletter - BloebaumAll MDO newsletters will be on the world wide web aero.stanford.edu/mdotc.html. Need more timely secretary inputs. Chris complemented Christina on the newsletter quality and its timely publication.

Awards- Neill (not present Thomas reporting)There were a number of nominations from the last time. Nomination deadline is June 15.

Planning- BorlandThanked the outgoing members for their service and welcomed the new members (see Elections).

Bench marking- Venkayya



A status of work done and what needs to be done. Discussion was far ranging. Revisited what makes a good test case. Some discussion as to direction of the subcommittee. One question - "Who is the Customer". No converts. Striz and Haftka became new co-chairs.

Applications- Coen ( not present) no report.

Propulsion - new chair

LIST OF ACTION ITEMS

1. Comments on Christina questions posed under 1996 MDO. 2. Yes on industry overview - Christina 3. Work in process; How many paper slots? - to Christina 4. Exchange of newsletters between TCs

CHAIRMAN ANNOUNCEMENTS

Words of advice from past-past-chairman - Sobieski

NEW BUSINESS

none

TC next meeting;time and place-The next MDO TC meeting was set for 1800 hours Monday, September 18, 1995 at AETOC, Los Angeles. Buffet will be served.

The meeting was adjourned 2017 hours.

Respectfully submitted on 7/17/95Nick Radovcich Lockheed Martin AeronauticalSystems Dept 73-HA, Zone 0988 Marietta, GA 30063 Tel: (404) 793-0008 Fax: (404) 793-0149




Chairman: Jean-Francois BarthelemyMembers: Shreekant Agrawal Kumar G. Bhatia Christina L. Bloebaum Carl M. Bosch Clark Briggs Robert Canfield Peter C. Coen Mark Drela Mark S. Ewing George Fadel Paul Gelhausen Omar Ghattas Joseph Giesing Edward A. Goforth Ramana V. Grandhi Raphael Haftka Srinivas Kodiyalam Johann Krammer Charles Lawrence Eli Livne Samuel C. McIntosh Achilles Messac Farrokh Mistree S. N. B. Murthy Henry Neimeir Jerry Newsom John R. Olds Paul Pierson Nick Radovchich Masoud Rais-Rohani Pradeep Raj David C. Redding John E. Renaud Gerald Seidel Abdon E. Sepulveda Gary Stanley Alfred G. Striz Harold Thomas Rudy Yurkovich Assoc. Member: Andrew Santangelo Past Chairman: Christopher Borland

Past Chairman: Christopher Borland








MDO TC Newsletter No. 18: April 1995


FROM THE EDITOR: Christina L. Bloebaum

Hello everyone. Looking forward to seeing you all at SDM, which is right around the corner.

I am announcing a slightly different approach to the newsletter, which will be implemented immediately following the SDM Conference, as long as no one has any strong objections. Instead of sending the newsletter out right before a conference, I propose it be sent out right after. This should make it much easier for everyone to get submissions to me, since most TC members will be in attendance at the meetings. You can either hand your submissions to me directly at the conference or email/send them to me immediately following. We can discuss this at the TC meeting on Monday, April 10, at the SDM Conference.

I have included an editorial that came across the internet on Cray and his recent filing for Chapter 11. I found it to be quite interesting and hope you do as well. Good reading!

CHAIRMAN'S CORNER:Chris Borland

Dear Fellow TC Members,The old Chinese curse says "May you live in interesting times," and we certainly do. Most of you are aware of recent upheavals in the Aerospace Industry, and although the developing situations (which seem to change minute by minute) may be disheartening and disorienting to many, I hope that we can look at as an opportunity for further insight and growth. Many of the reorganizations that we hear about may actually give some opportunity to consider the way business is done, and the chance to establish some new paradigms. If we try to think about what is really needed, a more enlightened approach to cross-functional and interdisciplinary activities is clearly a prime candidate. The excellent ground work that the MDO community has established over the past several years should be a starting place for even more advancement in the future. However, I think we still have a selling job to do as to MDO's long term worth and how it will affect the business's bottom line. The more good examples of practical applications and results that can be shown the easier this job will be. Please keep this in mind as we press on into the ever-exciting future. (Off the soap box).

The 6th Multidisciplinary Analysis and Optimization Symposium will be held at the Hyatt Regency Bellevue Hotel in Bellevue, Washington, on September 4-6, 1996. Dr. Christina Bloebaum of the State University of New York at Buffalo has agreed to serve as Technical Program Chair, and I will serve as General Chair. Julie Walker of AIAA and I have visited the facility and think it will be well suited to our meeting. It is located in downtown Bellevue, about 20 minutes from downtown Seattle and 25-30 minutes from the Airport by Shuttle. There are many interesting shops and eating places located both within the hotel complex and nearby, and for your significant others the world-famous Bellevue Square



Shopping Mall and Art Museum is across the street. This is the optimum time of the year to visit the Pacific Northwest, and we hope to offer a variety of interesting activities to supplement the usual excellent technical program. Since this is the week following Labor Day weekend, some of you may wish to spend a few extra days enjoying our expected fabulous weather and scenery (no guarantees!) More details will be announced at upcoming TC meetings and in the newsletter.

Our next TC meeting will be held in conjunction with the Structures, Structural Dynamics, and Materials Conference in New Orleans. The meeting will be Monday, April 10, at 7:30 PM, following a reception at 6:30 PM. This is our "no meal" meeting, so hit the reception and eat all the free goodies you can! This will be my last meeting as Chairman, and at its conclusion the virtual gavel will pass into the capable hands of Jean-Francois Barthelemy. Hope to see you all there.

Regards, Chris Borland

AGENDA FOR UPCOMING MDO MEETING

Monday, April 10 at the AIAA 36th Structures, Structural Dynamics and Materials.Conference in New Orleans, LA. The meeting will start at 7:30 PM. There is a reception scheduled for 6:30 PM, so we will not be serving food. Room will be posted or in the TC meeting list available at the registration desk.

1. Meeting called to order2. Introduction of members and visitors3. Minutes Review, Correction, and Approval4. Update of roster5. Subcommittee reports: Conference Support - Santangelo MA&O (1996) - Borland Liaison - Straub Education - Renaud Publications - Salama Newsletter - Bloebaum Awards - Neill Planning - Borland Benchmarking - Venkayya Applications - Coen Propulsion - ?6. Chairman's announcements 7. Business from the floor8. Introduction of New Chairman9. Upcoming TC meeting information10. Adjourn

9037 CASH-STARVED CRAY COMPUTER CLOSES, SEEKS CHAPTER 11



27 NEWS FLASH by Norris Parker Smith, Editor at Large, HPCwire

Why did Seymour Cray's company, Cray Computer Corporation (CCC), descend into bankruptcy despite his unmatched contributions to computing, his remarkable skills at computer architecture and manufacturing, and his exceptional persuasiveness?

How will the absence of CCC affect the high-performance market -- and the high-performance computing community?

The answer to the first question consists of a few simple propositions:

● The world changed; ● Seymour Cray tried hard to keep up with those changes without compromising his own vision of

high-performance computing; ● Toward the end, he tried very hard indeed. Adjusting prices downward and yielding to the

demands of modern marketing, he accepted that the grace, ingenuity and redoubtable performance of his solutions might not be enough to close sales;

● Despite these efforts, he could not obtain the customers he needed so desperately; ● If solid revenue had been in reach, he might have been able to raise more funds; ● Neither hope came true in time.

Just a few weeks ago, stockholders approved an additional stock issue. Placement was attempted outside the United States in order to avoid the delay required for SEC filing. The foreign investors evidently decided that the risk was too great. Lack of cash forced CCC to close and most of CCC's 350 workers were dismissed.

THE IRONY OF ORDINARINESS

Two clauses in these statements require further examination -- "The world changed" and "his own vision of high-performance computing."

How has the world changed? Some commentaries on CCC's decision to seek the protection of Chapter 11 have attributed it to a decline of supercomputing brought about by the loss of easy defense money and the proliferation of killer microprocessors.

This is mostly to the point, but supercomputing has not declined. It is, in fact, thriving to a degree never approached while Seymour Cray and supercomputer were identical in the public mind, like Einstein and relativity.

Supercomputing is thriving because it has changed so much that the old term -- always half-description, half-slogan -- has acquired misleading implications. The days of expensive, delicate machines, presided over by specialists and used mostly by people with exceptional requirements, are passing.



Supercomputing has become ordinary; just another way to do computing.

In this perspective of his long and productive career, this outcome means that Seymour Cray's lifework has been a remarkable success. Ironically, as Mr. Cray defined the objectives of CCC and the character of its products, it was unable to find a place in a world of ordinariness. Seymour Cray has always accomplished the extraordinary and he wanted to do it one more time.

GOLDEN YEARS

In the golden years, supercomputing supported one profitable medium-sized company, Cray Research, three Japanese emulators that probably rarely made much genuine profit doing it, and a few small companies pursuing different paths, like Thinking Machines (TMC) and Intel's supercomputing (now scalable) computer division.

Both TMC and Intel were dependent to some degree on outside funds -- DARPA subsidies for Thinking Machines and support for Intel's parallel-processing venture from the central treasury at Intel headquarters where the money is made.

Dozens of would-be Craylets rushed to occupy a narrow niche called mini- supercomputers. Only Convex survived, doing well yet significantly smaller than Cray Research.

Thus, there was only one influential, self-sustaining operation in the industry, Cray Research. It was shaped to a remarkable degree by Seymour Cray's personal creativity, persistence and skill.

If one includes the earlier years at Control Data when Seymour Cray was beginning to establish his way to do heavy-duty computations, the span extends to almost two decades. In modern times, few individual creative persons have shaped so much of a significant technology for so long. FOUR CHANGES In recent years, however, supercomputing has changed in four ways:

1. A raid-the-junkyard philosophy has been adopted. Microprocessors, other components and, in some cases, whole modules are borrowed from workstations or other technologies -- thus reducing costs and making it easier to raise reliability.

2. It has become yet another corporate phenomenon. Big companies like IBM, Digital, AT&T and Hewlett-Packard are now playing important roles. Silicon Graphics, which started a number of years after Cray Research concentrating upon specialized graphics workstations, now has more than twice the revenues of Cray Research. It has, with considerable success, gone into the lower end of supercomputing -- almost as a sideline.

3. Sales are increasing greatly. At the lower end (prices under $2 million), sales in the scientific/technical market alone are expected to reach about $1.5 billion this year, 50 percent more than the highest figure reached during the golden age. For years, despite intermittent effort, supercomputing made few inroads into commercial markets -- perhaps three times as large as the scientific market. Although much skepticism remains among commercial customers, significant



progress is now being made. 4. Ironically, sales are growing primarily in the lower levels, in systems with a few dozen -- or

fewer -- processors and various strategies to share a central memory. In principle, Seymour Cray would be at home with this kind of architecture.

Indeed, his efforts at CCC -- the CRAY-3 and CRAY-4 -- were in the most basic terms shared-memory multiprocessor systems with similar numbers of processors.

The big difference nowadays: each multi-megaflop (soon to be gigaflop) processor of a typical shared-memory multiprocessor sells for a few tens of thousands of dollars, not a couple of million.

In late 1994, CCC brought its price per GFLOPS for the CRAY-4 down toward the levels of its competitors, but the change came too late.

ADAPTATION

This is not simply a reenactment of that great drama of American nostalgia, the worthy individualist whose high-quality small business or delightful country store is squeezed out by faceless, homogenizing corporate giants. The troubles of CCC can be illuminated by considering Cray Research which retained the imprint of Mr. Cray for some years after he left it, and Seymour Cray's historical baseline customers: the code-breaking/intelligence agencies and the big federal laboratories.

Since the late 1980s, Cray Research has been seeking to adapt to the realities of a changing world.

It has been a protracted struggle marked by the extrusion of two of the main protagonists: Steve Chen and Seymour Cray himself. Mr. Cray left in 1989 to, in effect, set up his own country store, operated according to his distinctive principles.

Cray Research has gone on to expand its product line to a degree that would have been high heresy in earlier decades. It has now embarked upon a reconsolidation based on a multi-layer configuration, also pursued by Convex and other makers, that may now become a prevailing solution, especially for large-scale problems and the ascent of Mt. Teraflop.

Some new models even include memory caches, another heresy. Various euphemisms are applied by Cray Research and others, but "cluster" is the simplest description of the top layer of these proposed hierarchies, where nodes consisting of shared-memory systems are aggregated into a distributed supersystem.

ANAGRAM OF LATENCY

For a purist like Seymour Cray, cluster is just an anagram of latency. Bandwidths may be rising sharply



and cleverness can disguise the effects of high latency, but it is inescapable.

Seymour Cray has always preferred straightforward, simple solutions, combined with a willingness to invest much money and originality in the computational infrastructure that makes simplicity possible.

His favorite mathematical statement is, in effect: "the best connection between two components is the shortest possible line." He rejected fancy ideas like caches that, in his view, introduced self-defeating complexities. Mr. Cray also emphasized fundamental virtues: plenty of bandwidth, especially to memory, careful attention to memory management, and a sound balance among processor power, memory and I/O. He quietly criticized, for example, parallel systems that claimed huge aggregate performance but were so imbalanced that actual capabilities were a small percentage of the claimed performance.

HIGH OVERHEAD

CCC was established in the hopes of achieving exceptional performance while pursuing these pure goals. In doing so, Mr. Cray largely avoided the borrow-parts-don't-make-them philosophy that now dominates most of high-performance computing.

His strategy was to fit very fast processors into innovative, compact packages. The chips would be fast because they would be based on gallium arsenide (GaAs) rather than silicon. Mr. Cray chose, moreover, the most demanding of two basic ways to use GaAs.

This led to much frustration and delay. Finally CCC acquired its own GaAs foundry. More time was spent achieving acceptable yields with this brittle, unfamiliar material. The intricate design and compact dimensions of the processor nodes required special light-fingered robots.

Both were remarkable accomplishments, and Seymour Cray relished as always his ability to reach goals that others had scoffed at. Nevertheless, these accomplishments were also causes of added expense and delay.

CCC thus ended up with exceptionally high manufacturing overhead while other vendors of high-performance systems were increasing modularity to keep costs down. Furthermore, many competitors were moving toward production rates in the hundreds of units per year while CCC's costs would be spread over a dozen or so units at best.

None of these burdens would have been crippling if CCC's first product, the CRAY-3, had offered distinctively higher performance compared with preceding models and could sell into an unoccupied marketing window.

Instead, Cray Research's C-90 reached the market at about the same time with similar performance. As CCC spokesmen later acknowledged, the timing was bad. CCC did make one tentative sale, to Lawrence



Livermore National Laboratory (LLNL) for support of the Department of Energy's nationwide research program. In December 1991 when CCC was unable to meet delivery/performance goals, the order was cancelled.

A small CRAY-3 was later placed at the National Center for Atmospheric Research (NCAR) where, after some time and effort, it became ready for production work. This was not followed, however, by an actual sale.

DEDICATED, LOYAL CUSTOMERS

From his early days at CDC, Seymour Cray depended upon a foundation of dedicated, loyal customers that could be counted upon to buy the single-digit serial numbers -- and thus provide cash flow for expenses required by later manufacturing and sales.

This roster included national laboratories like LLNL, National Science Foundation-supported sites like NCAR and, later, the national supercomputing centers. Above all, Mr. Cray looked toward the cryptanalysis/image analysis/signal analysis requirements of federal intelligence agencies.

Once again, these were the customers that Seymour Cray counted upon to put CCC on track toward customer acceptance and adequate cash flow.

It never worked out. These agencies chose to stay with Cray Research which offered similar performance and much better prospects of financial stability.

Even worse, these agencies became deeply interested in clustering -- of classical C90 vector supercomputers and, later, of Cray Research's new J90 scaled-down, modernized, shared-memory multiprocessors.

Seymour Cray pressed ahead with the CRAY-4, an even-more-so machine with tighter design parameters and roughly twice the per-node performance of the CRAY-3.

If CCC had not already spent all the money it could get, the CRAY-4 might have won some customers away from Cray Research's counterpart, the T90. Now, no one will ever know for sure.

WIDER EFFECTS

Most customers and competitors have undoubtedly anticipated -- and discounted -- the departure of CCC for some time. That would have been considered an inescapable outcome if it were not for the unique aura of Seymour Cray's name and personality.

For most observers, timing was the only question. Now that uncertainty has ended, Cray Research may



find it somewhat easier to sell T90s, due for volume shipment during the latter half of this year.

The T90 will need all the luck it can get. The market for upper-end vector machines has shrunken, due to loss of market share to scalable parallel systems and to shared-memory devices like Cray's own J-series as well as decline in the supply of defense money.

It would be a shame if CCC's technological achievements in taming GaAs and robotic manufacture would be allowed to gather dust.

Seymour Cray himself suffers from his status (despite his own best efforts) as a mysterious celebrity as well as a highly-talented computer designer. This could be damaging: people have very short memories about celebrities and even shorter attention spans.

Despite the unfortunate course of Cray Computer, however, Seymour Cray should -- and, it is to be hoped, will -- continue to receive respect for his persistence as well as his imagination and the contributions made during a long and unique career.

Above all, his emphasis on the fundamentals on well-balanced systems deserves continuing reiteration. If Mr. Cray wishes to moderate somewhat his habit of silence, his comments could be valuable contributions to the ongoing dialogue of high-performance computing.

Seymour Cray, like many quiet men, combines a gentle humor with a well-developed sense of irony. He was probably amused to hear that a hasty reporter in Minnesota misunderstood the initial news of CCC's filing for bankruptcy and told the world that Cray Research was seeking Chapter 11.

This led to much fuss and a distinct outflow of resumes before the truth was re-established. Supercomputing may have become ordinary, yet the name Cray has not. It still commands respect -- and, at times, can cause confusion. HPCwire has released all copyright restrictions for this item. Please feel free to distribute this article to your friends and colleagues..............

HUMOR

Excerpt from the Buffalo News - Ann Landers

(Message from the Editor: I hope that no one will find this offensive. It is certainly not meant to be so in any way, shape, or form.)

Why God never received tenure at any university

● He had only one major publication. ● It was in Hebrew.



● He had no references. ● It wasn't published in a refereed journal. ● Some doubt He wrote it Himself. ● He may have created the world, but what has He done since? ● The scientific community can't replicate His results. ● He never got permission from the ethics board to use human subjects. ● When one experiment went awry, He tried to cover it up by drowning the subjects. ● He rarely came to class and just told students, "Read the Book". ● Some say He had His Son teach the class. ● He expelled His first two students. ● His office hours were irregular and sometimes held on a mountain. ● Although there were only 10 requirements, most students failed.

DESIGN EDUCATION INNOVATIONS

C. L. Bloebaum

In March of 1993, ASME held a conference entitled Innovations in Engineering Design Education in Orlando, Florida. The papers from the conference have been published by ASME as the Innovations in Engineering Design Education: Resource Guide (ISBN 0-7918-0678-2).

An awards program was established to identify the most noteworthy of these papers and were published in a separate cover entitled "Innovations in Mechanical Engineering Curricula for the 1990's". The overall award winner was a paper entitled "Reverse Engineering and Re-Engineering in Capstone Design", by G.A. Gabriele, L.N. Myrabo, J. Pegna, H.J. Sneck, and B.L. Swersey, from Rensselaer Polytechnic Institute.

These are wonderful papers that I suggest everyone read - whether in academia or not. Remember, we will be having a session on Engineering Education at the next MDO conference so start thinking now!

Please send your comments and contributions for the next MDO TC Newsletter to:

Professor C. L. Bloebaum email address is:Department of Mechanical and Aerospace Engineering [email protected] Furnas Hall departmental phone is: (716) 645-2593 State University of New York at Buffalowork phone is: (716) 645-2593 x2231Buffalo, New York 14260 Fax is: (716) 645-3875




Chairman: Christopher BorlandMembers:J.-F. Barthelemy Todd J. Beltracchi Laszlo Berke Kumar G. Bhatia Christina L. Bloebaum Carl M. Bosch Peter C. Coen Mark Drela Mark S. Ewing John W. Gallman Paul Gelhausen Edward A. Goforth Ramana V. Grandhi David P. Harry III Gary C. Hill Johann Krammer Ilan M. Kroo Eli Livne Samuel C. McIntosh Farrokh Mistree S. N. B. Murthy Douglas J. Neill Henry Neimeir Jerry Newsom John R. Olds Pradeep Raj David C. Redding John E. Renaud Mokhtar Salama Abdon E. Sepulveda Friedrich Straub Alfred G. Striz Arthur C. Taylor III Vipperla Venkayya Barnaby Wainfan Rudy YurkovichAssociate Kevin D. Bell Andrew Santangelo Gregory D. SikesMembers: Harold ThomasPast Chairman: Jaroslaw Sobieski








Download TC files

Download TC files

UNDER CONSTRUCTION

This page is intended for the easy dissemination of electronic documents to TC members and Web site visitors. Currently downloadable documents include:

● Postscript version of operations slides: operations.ps




http://endo.sandia.gov/AIAA_MDOTC/communications/download.html12/29/2006 12:29:29 PM

ftp://endo.sandia.gov/pub/docs/mseldre/operations.ps



Conferences Supported by the MDO TC

Conferences Supported by the MDO TC

● World Aviation Congress 99, October 19-21, 1999, San Francisco, CA, USA.

● 38th AIAA Aerospace Sciences Meeting and Exhibit, January 10-13, 2000, Reno, NV, USA.

● 41st AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, April 3-6 2000, Westin Peachtree Plaza Hotel, Atlanta, GA, USA.

● 8th AIAA/USAF/NASA/ISSMO Symposium on Multidisciplinary Analysis and Optimization, September 6-8 2000, Long Beach, CA, USA.

Other conferences with MDO content (not sponsored by AIAA).

AIAA's calendar of conference and short course events.




http://endo.sandia.gov/AIAA_MDOTC/sponsored/MDOTC_con.html12/29/2006 12:29:29 PM

http://www.sae.org/CALENDAR/wac99tag.htm

http://www.aiaa.org/calendar/asm00cfp.html

http://www.aiaa.org/calendar/ssdmc00cfp.html

http://www.aiaa.org/calendar/ssdmc00cfp.html

http://www.aiaa.org/calendar/index.html



Conferences/Events with MDO Content

Conferences/Events with MDO Content

● INFORMS Fall 1999 Meeting, November 7-10, 1999, Philadelphia Marriott Hotel, Philadelphia, PA, USA.

● ASME 2000 International Design Engineering Technical Conferences and the Computers and Information in Engineering Conference, September 10-13, 2000, Omni Inner Harbor Hotel, Baltimore, Maryland, USA.

Conferences sponsored by the MDO TC.




http://endo.sandia.gov/AIAA_MDOTC/related/mdoconf.html12/29/2006 12:29:30 PM

http://www.informs.org/Conf/Philadelphia99/

http://www.enme.umd.edu/asme2000/

http://www.enme.umd.edu/asme2000/



Short Courses Supported by the MDO TC

Short Courses Supported by the MDO TC

● Optimal Design in Multidisciplinary Systems, AIAA Professional Development Short Course, St. Louis, MO, August 31-September 1, 1998 .

● Software Tools and Techniques for Reducing Time and Cost in the Design Cycle , AIAA Professional Development Short Course, St. Louis, MO, August 31-September 1, 1998 .

AIAA's calendar of conference and short course events.


Last Updated: July 15, 1998


http://endo.sandia.gov/AIAA_MDOTC/sponsored/MDOTC_short.html12/29/2006 12:29:30 PM

http://www.aiaa.org/calendar/pd_aug98-5.html

http://www.aiaa.org/calendar/pd_aug98-6.html

http://www.aiaa.org/calendar/index.html



White Papers and Publications Prepared by the MDO TC

White Papers and Publications Prepared by the MDO TC

● A new White Paper on Industrial Experience with MDO consists of several invited papers and a summary report from the 1998 Symposium on Multidisciplinary Analysis and Optimization.

● Current State of the Art On Multidisciplinary Design Optimization (MDO), An AIAA White Paper, ISBN 1-56347-021-7, September 1991.

● Multidisciplinary design optimization highlights article for annual "Year In Review" issue of Aerospace America (each December).


Last Updated: June 1, 1999


http://endo.sandia.gov/AIAA_MDOTC/sponsored/MDOTC_white.html12/29/2006 12:29:31 PM

http://www.aiaa.org/publications/aa-magazine.html



AIAA Technical Committee

AIAA Technical Committee on Multidisciplinary Design Optimization (MDO)

White Paper on Current State of the Art

January 15, 1991

© 1991, American Institute of Aeronautics and Astronautics, Inc., posted on the Internet by permission

PREAMBLE

AIAA has established a Technical Committee for Multidisciplinary Design Optimization (TC-MDO) with the following charter:

"To provide an AIAA Forum for those active in development, application, and teaching of a formal design methodology based on the integration of disciplinary analyses and sensitivity analyses, optimization, and artificial intelligence, applicable at all stages of the multidisciplinary design of aerospace systems".

One of the functions the TC-MDO established for itself is to provide the aerospace community with a periodic assessment of the state-of-the-art in its field beginning with this White Paper.

The task of developing this initial White Paper was led by Daniel Schrage assisted by Todd Beltracchi, Laszlo Berke, Alan Dodd, Larry Niedling, and Jaroslaw Sobieski.

All members of the TC/MDO reviewed several drafts of the White Paper in its editorial process. A list of the TC-MDO members is included as Appendix II.

FOREWORD

This White Paper's purpose is threefold. First, it explores the need for bringing the diverse disciplinary design technologies involved in development of aerospace vehicles and expounded upon in the other chapters in this volume into a concerted action. This approach is necessary to create advanced aerospace vehicles that must be competitive not only in terms of performance, but also in terms of

http://endo.sandia.gov/AIAA_MDOTC/sponsored/aiaa_paper.html (1 of 41)12/29/2006 12:29:35 PM


manufacturability, serviceability and overall life-cycle cost effectiveness. Second, it reviews some of the recently evolved means by which such concerted action may be implemented in a systematic and mathematically-based manner referred to as the Multidisciplinary Design Optimization (MDO) technology. Third, it points out major directions for research and development.

The discourse is divided into six sections. The first section presents the need for the MDO technology in the historical context of progress in aerospace. In the second section, the emphasis is on the multidisciplinary nature of the aerospace design process. The human element in that process is discussed in the next section as the key component in any design-oriented technology. The fourth section is devoted to computing as the essential part of the design infrastructure. In the fifth section, the attention shifts to sensitivity analysis and optimization methods that form the core of the MDO technology. Finally, the concluding section identifies the development directions for realization of the MDO benefits.

TABLE OF CONTENTS

I Introduction and Background

A. History of Aerospace Systems Design

B. The Need for MDO

II Multidisciplinary Aspects of Design

A. Engineering Design Disciplines

B. Concurrent Engineering Disciplines

C. Supporting Disciplines

III Human Interface Aspects of Design

A. Design Decision Making

B. Meta Design

IV Computing Aspects of Design

A. Information Architecture



B. High Performance Computing

V Optimization Aspects of Design

A. System Level Optimization

B. Decomposition and Sensitivity Analysis

C. Concluding Remarks on Optimization

VI Transitioning to the MDO Environment

VII Conclusions

VIII The Role of the AIAA MDO TC

References

Appendix I: Survey of the Industry MDO Practices

Appendix II: AIAA TC MDO Membership Roster

I. INTRODUCTION AND BACKGROUND

A. History of Aerospace Systems Design

During the pioneering years of aviation, the aircraft designer frequently was the central figure and the jack-of-all-trades -- designer as well as main resource person in aerodynamics, structures, materials, propulsion, and manufacturing, often also test pilot, entrepreneur and founder of great enterprises. The Wright Brothers, Glen L. Martin, Breguet, DeHavilland, Fokker, Heinkel and Sikorsky are just a few of the names which come readily to mind. Creative spirit, clear grasp of essentials, and confidence-inspiring, self-assured personality were their characteristic traits. The knowledge necessary to design an airplane was of a practical kind and for many years it was no more than could be stored in the mind of a capable individual.

This first period came to an end in the early 1930s. Evaluation of wind tunnel tests in aerodynamics, thin shell analysis in structures, thermodynamic efficiencies in propulsion, processing and forming techniques in production - each of them developed into a field of specialization. The design engineer could not possibly keep abreast of all developments and had difficulty coordinating the different inputs coming from various specialists. Yet the solid engineering background and the long experience of the



typical design engineer provided the know-how and the balanced judgment to translate new theoretical knowledge into flying hardware. Thus the senior design engineer had to evolve into what would today be called the systems engineer. This period lasted from the years of exciting technical progress in the 1930s, through the years of mass production during World War II, to the expansion of air transportation in the 1950s. A few prominent names during this period are Johnson, Northrop, McDonnell, Douglas, and Hughes. This period of time also produced rocket pioneers, such as Goddard, Oberth, Korolev and Von Braun.

In the late 1950s a slow change in attitude occurred throughout aircraft design. Partly due to the impetus given by missiles, rockets, and spacecraft which are one of a kind single use systems that used a new set of design guidelines, and partly due to the demands of the military who were striving for maximum performance, the importance and prestige of analytical specialists soared. Specialists were needed to expand the limits of scientific knowledge and to reach for ever higher performance. The best minds were attracted by the challenges of research and development which usually meant estrangement from design. As a result, the design engineer's prestige declined. The analytical specialist was often the originator of novel ideas and the design engineer became the implementor as he translated these ideas into practice.

Then, around 1970, began the big slump in the aircraft industry coupled with a decline in the civilian and military space programs which led to a reduction of the engineering force by about 25%. Simultaneously, two developments of great potential impact and far-reaching effect on aircraft design began to take place. First. computer-aided design came of age and has now relieved the design engineer of much of the earlier drudgery regarding the menial aspects of design. Second, the procurement policy of the military underwent a thorough change. The earlier drive of maximum performance had been superseded by a new quest for balance among performance, life-cycle cost, reliability, maintainability, vulnerability, and other "-ilities". This trend is reflected in the design requirements growth for advanced aeronautical vehicles in Figure 1. A major reason for this emphasis was the control of life cycle costs which are determined by the design concept and thus are very difficult to change significantly past this stage as illustrated in Figure 2. The experience of the 1960s had shown that for military aircraft the cost of the final increment of performance usually is excessive in terms of other characteristics and that the overall system must be optimized, not just performance. The same lesson had been learned earlier by the airlines when meticulous cost accounting had pointed toward possible savings due to improved reliability and maintainability [1]. Cost- effectiveness for an airliner is mostly economic. The aircraft must generate sufficient revenue in excess of operating costs that the purchase investment is more profitable than investing the same amount of money elsewhere. A similar shift of concern toward cost, supportability, launch availability, and reliability in orbit began to occur for similar reasons more than a decade earlier in the space launch vehicles and spacecraft.

The 1980's brought about a number of thrusts both in government and industry to improve U.S. productivity and the quality of products. There has been an on-going quiet revolution in industry for the past ten years to make the necessary corporate, organizational and technical changes to compete successfully in an increasingly competitive global marketplace. These changes occurred first in the automotive and electronics industries, which were receiving intense competition for their products from



Japan, but in the late 1980's had spread to the Aerospace industry. Many of the initiatives in government, particularly the Department of Defense (DoD), can be traced to recommendations from President Reagan's Blue Ribbon Commission on Defense Management (Packard Commission) for improving the weapon system acquisition process. Policy formulation from these recommendations has come in the form of general acquisition streamlining and the Total Quality Management (TQM) Program. Other initiatives can be traced to the DoD's desire to take advantage of emerging information and computing technologies and the environment they provide. The DoD - initiated Computer-Aided Acquisition and Logistics Support (CALS) Program is one example.

As these initiatives have been implemented, there has been increased realization that in engineering, especially design, lies the greatest opportunity to improve product quality and provide concurrency of product and process phases to reduce development time. This realization has resulted in the recent emphasis on concurrent engineering (CE). CE has been defined as a systematic approach to the integrated, concurrent design of products and related processes, including manufacturing and supportability [2]. This definition is intended to emphasize from the outset consideration of all elements of the product life cycle from concept through disposal, including quality, cost, and schedule with traceability to user requirements. In most cases CE is envisioned as a modem application of systems engineering in an integrated computing environment. To date the CE emphasis has been on concurrent consideration of the life cycle phases, as illustrated in the top half of Figure 3, for the two-fold goal of improving quality by allowing the natural coupling among these phases influence the design decisions, and compressing the overall design process timetable.

Close examination of the Design Phase of the CE process reveals potential benefits from rearranging the traditional disciplinary tasks from the conventional sequential order into concurrent activities shown in the bottom half of Figure 3. The designer can exploit the synergism of the interdisciplinary couplings provided that effective mathematical tools and methodologies are available. Thus, the Multidisciplinary Design Optimization (MDO) methodology that combines analyses and optimizations in the individual disciplines with those of the entire system is a technology that enables extension of the CE concept to the Design Phase.

B. The Need for Multidisciplinary Design Optimization (MDO)

Design consists of a hierarchical sequence of steps. It begins with ideas, missions and concepts, takes successively firmer shape until the configuration can be frozen, continues with the practical considerations about hardware, and leads to a set of manufacturing instructions and airworthiness documentation. This evolutionary process usually is depicted as phases from conceptual to preliminary to detail design and then manufacturing and production, as illustrated in Figure 4. As this process evolves design freedom decays rapidly while knowledge about the object of design is increasing as illustrated in Figure 5. As the design process goes forward designers gain knowledge but lose freedom to act on that knowledge. It was demonstrated mathematically in [4] that this natural evolution may lead to suboptimal designs.



Traditionally, for aircraft and most other aerospace systems, design synthesis and optimization of the overall conceptual system has been based on achieving a fuel balance and a minimum weight configuration through parametric variation of a few critical design parameters i.e. wing loading, aspect ratio, etc. This aerospace approach to design synthesis is illustrated in Figure 6. Since aerodynamics and propulsion are the critical disciplines to achieving a fuel balance and vehicle performance, they are emphasized and the greatest level of effort is expended in these areas as illustrated in Figure 5. As the system design moves into the preliminary design phase and the initial configuration is frozen, hardware design considerations begin to dominate and the structures discipline begins to play a more dominant role. In the detailed design phase the controls discipline plays an increasing role as flight dynamics and handling quality improvements usually are necessary to achieve an acceptable flightworthy system. Also, the transition to production places a much bigger emphasis on manufacturing, cost, and to some extent supportability. The obvious problem with this traditional approach is the short conceptual design phase with an unequal distribution of disciplines which does not allow use of design freedom to improve quality and integrate disciplines for optimization. Also, the balanced design sought by the requirements growth in Figure 1 cannot be achieved. This was also a major conclusion from a recent industry survey conducted by the MDO technical committee. The results of this survey have been included as Appendix I.

In recent years there has been an increased emphasis on integrating the structures and controls disciplines into the design at an earlier time. For the structures discipline the increased use of advanced materials with their flexibility and reliability based structural design philosophies has been one force for this emphasis. Another force is the use of composite materials for aeroelastic tailoring, as it couples a structural detail (using skin fiber orientation angle) with the flexible wing aerodynamics and, ultimately, the aircraft performance. The controls discipline has really become an upfront partner. Control configured vehicles offer significant opportunities for expanded flight envelopes and enhanced performance through relaxation of inherent stability margins. Flight control state of the art is perhaps best epitomized by the space shuttle digital fly-by-wire control system which provides control of the vehicle from on-orbit maneuvering, through atmospheric entry, from Mach 25 to a horizontal landing using blended reaction and aerodynamic controls. Full authority digital fly-by-wire flight control has been incorporated in operational military aircraft such as the F/A-18. Application to civil aircraft, prompted by potential performance advantages in aerodynamics, structures, and operations has been initiated. However, concerns over reliability, maintainability, cost, and integrity of such systems has delayed its application in the U.S. although the A-320 AirBus has a digital fly by wire system for use throughout normal flight. Control configured vehicles offer significant opportunities for expanded flight envelopes and enhanced performance though relation of inherent stability margins. In addition, ultra-light-weight actively controlled space structures offer a weight reduction over conventional space structures. The ultimate goal of control integration is to maximize total aircraft performance. This goal can only be achieved by a balanced multidisciplinary design as portrayed in Figure 7 [5].

Aerospace vehicles are engineering systems whose performance depends on interaction of many disciplines and parts and whose behavior is governed by a very large set of coupled equations. In practice, engineers deal with these equations by partitioning them into subsets corresponding to the



major disciplines, such as aerodynamics, structures, flight controls, etc. In this process of pragmatic partitioning, the couplings among the subsets tend to be reduced in number because it is burdensome to account strictly for them all. Couplings are retained or neglected judgmentally on the basis of what is known or assumed about their strength in a particular vehicle category. Generally speaking, the more advanced the vehicle, the more such couplings should be accounted for.

Rotary wing aircraft or rotorcrafts are an excellent example of a highly coupled aerospace system. The multidisciplinary complexity of a rotorcraft, such as a helicopter is illustrated in Figure 8. Unsteady aerodynamics and vortex interaction cause excitation of complex structural dynamics to form a unique aeroelastic phenomenon which is further complicated by a direct coupling with the flight control system to trim the aircraft. The interaction that takes place among the disciplines of aerodynamics, aeroelasticity, structures and materials, and flight mechanics and controls in a typical flight condition is a series of feedback loops as illustrated schematically in Figure 9. The coupling of these disciplines is illustrated in matrix form in Figure 10 by referring back to the feedback loops of Figure 9. Principal and supporting disciplines are identified for each loop. If this off-diagonal coupling was not present, a linear superposition of research conducted by individual researchers at different locations could be combined. However, the coupling is strong, requires an interdisciplinary approach, and is one reason why progress in advancing rotary wing aircraft technology has been difficult. A similar coupling problem is evident on other advanced aerospace systems, although the interaction of disciplines would be different, such as the aerodynamics - propulsion - structures - controls coupling in hypersonic vehicles. The design synthesis flow chart using fuel balance for the Aerospace Plane is illustrated in Figure 11 [6].

While multidisciplinary integration can be associated with the traditional aerospace disciplines aerodynamics, propulsion, structures, and controls there are also the life cycle areas of manufacturability, supportability, and cost which require integration. After all, it is the balanced design with equal or weighted treatment of performance, cost, manufacturability and supportability which has to be the ultimate goal of multidisciplinary integration. Therefore, the multidisciplinary integration aspects of aerospace system design include the traditional disciplines of aerodynamics, propulsion, structures, and controls, as well as the life cycle disciplines of manufacturability, supportability and cost. The goal of this total multidisciplinary integration is illustrated in Figure 12. The changes in Figure 12 from Figure 5 are that the conceptual designer's time has been doubled to capture more knowledge and use more design freedom; the detail design time has been reduced by one third based on the use of more upfront design, and a more evenly distributed effort of disciplines is provided in the conceptual and preliminary design phase. The dashed line projection from the "Knowledge about Design" curve reflects the requirement that more knowledge will have to be brought forward to the conceptual and preliminary design phases. The dashed line projection from the "Design Freedom" curve reflects the need to retain more design freedom later into the process in order to act on the new knowledge gained by analysis, experimentation, and human reasoning. The change in the shapes of the two curves would alleviate the paradox that was discussed in conjunction with Figure 5. That change might be achieved through better integration of multi -and interdisciplinary design, analysis, and optimization. Obviously, another goal is to reduce the design time in order either to shorten the process duration or to develop a broader selection of optimized alternative designs in the constant elapsed time.



A clearly defined objective and sufficient budget to accomplish it is also required for multidisciplinary integration to work. The space station is an example of a system where much upfront design has been performed, but no flight hardware has been built as the funding has been in a continuous state of flux leading to one costly redesign after another.

Of course, an aerospace vehicle constitutes an integrated system by virtue of its physics, thus integration is a physical fact and hardly needs any advocacy for its existence. Therefore, when we postulate integration, we advocate research and development of means to help engineers master the interdisciplinary couplings and to enable them to exploit the associated synergism, toward improved efficiency and effectiveness of the design process and better quality of the final product.

Consistent with the above, an integrated design process may be defined as one in which:

(1) Any new information originated anywhere (in any discipline) in the design organization is communicated promptly to all recipients to whom it matters:

(2) When a change of any design variable is proposed, the effects of that change on the system as a whole, on its parts, and on all the disciplines are evaluated expeditiously and used to guide the system synthesis.

It is evident that (1) relies on the technologies for data management and graphic visualization, while (2) is based on synthesis, analysis and sensitivity analysis. Together, the above attributes form a capability for design optimization to be executed in a symbiosis of the human mind and the computer.

Since the technologies of (1) are well cared for by other AIAA TC's and thrive on the marketplace, it is logical for AIAA TC-MDO to focus its efforts on the technologies underlying (2) which are much less known and, therefore, underutilized: design synthesis, sensitivity analysis, optimization methods, melding the human mind and computer capabilities, and effective organization of engineering to exploit these technologies.

II. MULTIDISCIPLINARY ASPECTS OF DESIGN

A. Engineering Design Disciplines

The traditional engineering disciplines for aerospace vehicles include aerodynamics, propulsion, structures and controls. While these individual disciplines are considered fairly mature for many aircraft applications, there are advances in each discipline, due to theoretical, computational and methodology breakthroughs, that foster substantial payoffs and additional research. Emphasis in recent years, however, has been on the advances that can be achieved with research of the interaction between two or more of the disciplines. Also, new disciplines, such as electromagnetics, for low observability, without a statistical database need to be addressed. For advanced and particularly complex aerospace vehicles this



interdisciplinary approach is often essential owing to the strong couplings among the disciplines and subsystems and, again, the lack of statistical data and human experience.

B. Concurrent Engineering Disciplines

While the engineering design disciplines, their interdisciplinary interaction, and optimization of the product are the primary focus for this technical committee it would be remiss if it didn't address their incorporation in the broader set of Concurrent Engineering (CE) disciplines. As depicted in Figure 5 the addition of manufacturing, supportability and cost to the traditional engineering disciplines constitute the set of CE disciplines, with quality being the CE objective function for optimization. The prerequisite task for that addition is development of realistic, reliable, and easy to use mathematical models for manufacturing, supportability, and cost. In contrast to the traditional engineering disciplines, such models are currently inadequate and this inhibits their incorporation in a formal MDO methodology. Obviously, for military systems cost and operational effectiveness and the tradeoff between them receives high priority [7].

C. Supporting Disciplines

Multidisciplinary design optimization of aerospace vehicles cannot take place without substantial contributions from supporting disciplines. The identified supporting disciplines and methodologies are the Human Interface Aspects of Design, Intelligent and Knowledge-Based Systems, Computing Aspects of Design and Information Integration and Management.

III. HUMAN INTERFACE ASPECTS OF DESIGN

The engineering design process is recognized as a two-sided activity as illustrated in Figure 13. It has a qualitative side dominated by the human inventiveness, creativity, and intuition. The other side is quantitative, concerned with generating numerical answers to the questions that arise on the qualitative side. The process goes forward by a continual question-answer iteration between the two sides. The MDO methodology discards the "push button design" idea in favor of a realistic approach that recognizes the role of human mind as the leading force in the design process and the role of mathematics and computers as indispensable tools. It is clearly recognized that while conceiving different design concepts is a function of human mind, the evaluation and choice among competing, discretely different concepts, e.g., classical configuration vs. a forward swept wing and a canard configuration, requires that each concept be optimized to reveal its full potential. This approach is consistent with the creative characteristics of the human brain and the efficiency, discipline, and infallible memory of the computer.

The middle ground between the two sides of design is occupied by the quasi-intelligent and knowledge-based systems. The area of intelligent and knowledge - based systems deals with a broad variety of ways in which the science and technology of Artificial Intelligence (AI) could contribute to the theory and practice of engineering design. The potential contributions cover much more than what are commonly



inferred to as expert systems. Expert systems as generally implemented with current techniques. have very limited means of knowledge representation and deduction. The problems of design synthesis using multidisciplinary design optimization will usually require more powerful abstractions than provided by the current paradigm of expert systems [8].

A. Design Decision Making

The engineering process can be viewed as a series of decisions which gradually define a new product in more and more detail. As the product evolves from conceptual to preliminary design, to detail design, and then production, the details of the decision making process change radically but its general nature remains the same. Therefore, it can be seen that decision making is at the heart of design. Many different types of decisions must be made in even the simplest case. One must decide where first to look for similar solved problems, how much time should be spent looking at modifications to past or current designs versus new development, which aspects of the design are most important, and how other disciplines are affected. A schematic of how decision makers, using human expertise and expert systems drive the design process is illustrated in Figure 14. These decisions are made in the design process in an environment of uncertainty and risk. Uncertainties come in various forms and the design team faces both upstream and downstream uncertainties. Upstream uncertainties include, for example: uncertainty in the specification of design requirements. This uncertainty relates to the possibility of modification of the original specification that is being designed to. Such changes occur frequently in weapon systems procurements and cause havoc in the design process in terms of schedule slippage and cost increase. Design of space launch vehicles is fraught with uncertainties as to the future mission parameters that may vary in a broad range or vehicle modifications that result in a stretched design. Oftentimes, downstream uncertainties may reflect a lack of knowledge as to the environment in which the product will be used or uncertainties in future availability of spare parts. Uncertainties in manufacturing processes, such as process variability, are also examples of downstream uncertainties from a design standpoint [9].

B. Meta Design

Design viewed as decision-making implies the need to plan the decision-making process. Meta-Design "The design of the design process" addresses the planning activity. As illustrated in Figure 6the aerospace industry has developed a general synthesis and analysis which has proved successful for developing aerospace vehicles from helicopters to spacecraft. However, the existing design process has been geared principally to producing designs optimized for performance considerations without equal regard to cost, schedule, producibility, supportability or quality. As illustrated in Figure 12 design decisions and tradeoffs may have to be reordered among multidisciplines and different decisions may be required. A more flexible design process than illustrated in Figure 6 is required. Plans for integrating CAD/CAE/CAM tools, analysis tools, and design data bases should be directed toward executing a specific concurrent engineering design methodology. The type of design methodology used will depend on the type of design problem being addressed. Implementing a different computer integration scheme for each design methodology would pose a considerable burden in terms of software development. An



alternative approach entails developing a flexible design system capable of supporting the activities of methodology development (meta-design) and methodology execution (design) for multiple design problems. Such a system would be compatible with the evolving idea of a flexible acquisition process and would be analogous to a flexible manufacturing system in that it could be rapidly reconfigured to support products of many different designs. An analytical approach to meta-design that involves providing a framework that allows the design methodologies to be developed and evaluated is addressed in [10].

IV. COMPUTING ASPECTS OF DESIGN

Computer technologies have been changing the environment of engineering design. Therefore, these technologies are a major supporting discipline for MDO. Powerful analysis and simulation programs and CAD workstations are contributing to better solutions. These developments, in turn, are creating new difficulties. In an environment where most of the computer activities still involve stand-alone programs, design engineers often spend 50-80% of their time organizing data and moving it between applications. Integrated processing with database system support should eliminate many of these error-prone manual activities. Data must be shared between disciplines and within disciplines with all the applicable quality, consistency and integrity checks.

It should be emphasized that the MDO methodology calls for extending the type of data available to the designer by the new category of the derivative, or trend data that directly answer the "What If?" questions about the entire vehicle system. Examples of such trend data are the derivative of the aircraft range with respect to the wing aspect ratio, incorporating the aerodynamics-structure interaction, or the derivative of the seat-mile operational cost with respect to the take-off gross weight, accounting for the coupling of the structures, aerodynamics, and propulsion. Since the continual concern about the "what if" questions is what a creative design is all about, having a capability to answer such questions expeditiously and comprehensively will constitute a quantum jump in the design process effectiveness and efficiency.

A. Information Architecture

Several parallel efforts have been and are being undertaken to identify an information framework for integrated design. As a result of a NSF workshop [11], a strong recommendation was made for the establishment of a national research program on engineering information management and suggested that the components include:

Engineering Product and Process Description

Engineering Information Dynamics and Data Models

Very High Level Languages and User Interface Engineering



Decision Support Systems

Conclusions from this NSF workshop were that this research will require the concerned joint efforts of industry, government and academia and that it will require multidisciplinary teams from such areas as engineering, computer science, social science and mathematics.

Another ongoing effort is the work by the Computer-Aided Acquisition and Logistics Support (CALS)/Concurrent Engineering (CE) Mechanical Systems Framework Subtask Group. They have concluded that the information architecture must allow a large multi-disciplinary group to behave as a tightly knit inter disciplinary team, in a concurrent manner in creating product definition information. This architecture includes: concurrent product and process definition, product development team, product life cycle data, and knowledge of customer needs. The architecture may be seen as consisting of an Enterprise Integration framework and an Integrated Information Management System backbone. The Enterprise Integration includes: Product Definition, Process Definition, Configuration Management. Information Exchange, Team Organization, Validation, Metrics, and Enterprise Policy. These elements are peculiar to the enterprise itself. Yet there is an Information Management System that integrates the elements of the enterprise by means of a shared database environment. This includes: Information Modeling, Tool Integration, Information Integrity, Information View, Information Management, Communication, and Resource Definition. The Subtask Group has been assessing the existing environment for Concurrent Engineering from the above stated perspectives. Key topics include:

1) Information architecture,

2) Data exchange standards, such as the Product Data Exchange Specification (PDES),

3) Design - by - Feature,

4) Object - Oriented data management technologies,

5) Storage of (and access to) properties and constraints, material characteristics, and manufacturing methods; and the ability to create (user-specified) multiple views, intelligent libraries, and part, feature, and process information. A first draft of requirements for concurrent engineering information architecture has been completed by the CALS/CE Frameworks Subtask Group [l2].

B. High Performance Computing

The term "supercomputer" is commonly used to denote computing power, but the definition of power in a computer is highly inexact and depends on many factors including processor speed, memory size, and so on. Secondly, there is not a clear lower boundary of supercomputer power. IBM 3090 computers come in a wide range of configurations, some of the largest of which are the basis of supercomputer centers at university, government and industry locations. Finally, technology is changing rapidly and



with it our conceptions of power and capability of various types of machines. Therefore, the general term, "high performance computers (HPC)", is a term that includes a variety of architectures. One class of HPC consists of very large, powerful machines, principally designed for very large numerical applications, such as those encountered in science and engineering. Parallel processing assumes that a problem can be broken into large independent pieces that can be computed in separate processors. Currently, large mainframe HPC's such as those offered by Cray, IBM are only modestly parallel, having as few as two up to as many as eight processors. The trend is toward more parallel processors on these large systems. Some experts anticipate as many as 512 processor machines appearing in the near future. The key problem to date has been to understand how problems can be set up to take advantage of the potential speed advantage of larger scale parallel processing [l3].

A NASA Grand Challenge for high performance computing in aerosciences has been put forth as the integrated multidisciplinary design of aerospace vehicles and their numerical simulation throughout a mission profile [l4]. The goal is to demonstrate the utility of advanced parallel computer systems, including hardware, software and algorithms, capable of delivering teraflop performance for the design of a new generation of aerospace vehicles. Such a demonstration requires separate developments within a number of disciplines as well as the tight integration of those disciplines. Figure 15 and 16 provide some indication of the computational complexity and the present state of the art for two disciplines: aerodynamics and structural analysis. The underlying assumption is that a single simulation must be completed in 15 minutes.

Figure 15 shows a range of configuration complexities from an airfoil through a wing to a full aircraft. Figure 16 also shows a range of computational requirements relative to past and present high performance computers. Again, the configuration complexity moves from a simple laminated material through a component to a full aircraft. The computational requirements implied by these figures are severe in their own right. When one thinks of coupling these and other disciplines that are equally computationally demanding through optimization formulation that requires repeated evaluation of these models the "challenge" is truly "grand" [l4]. To meet that challenge, the MDO technologist recognizes that the usable computing speed is a product of the hardware speed and the algorithm speed. In other words, one cannot get very far by using a multiprocessor computer for executing a method that originated [in a] serial computer environment. It follows that to extract full computational potential from a new type of a computer, one needs to invest a development effort in new solution algorithms comparable to the effort that went into the hardware development itself.

V. OPTIMIZATION ASPECTS OF DESIGN

Optimization methods have been combined with design synthesis and parametric analysis and used in the aerospace industry for the past forty years. The graphically displayed "carpet plot" is a characteristic of this legacy. In the first two decades the most commonly used techniques were graphical methods. Graphical methods were straight forward and easily understood, and had the obvious advantage of showing at a glance the entire interval of interest, calling attention to the function peaks, valleys, and



other instructive features. The important limitation of these methods is that they can paint such a clear picture for only up to three or four variables in one figure, and require large computer resources for generating data points for constructing the plots. For greater number of variables, the combinatorial explosion sets in that would multiply the figures into volumes, and volumes into libraries with the attendant loss of the easy comprehension and interpretation.

During the past two decades much progress has been made in numerical optimization that offers an alternative to the above. Any design can be defined by a vector in multidimensional space where each design variable represents a different dimension. Since we cannot see in more than three dimensions, the general case is beyond our power of visualization. Yet the principle is the same as when we assume only two variables in a base plane and plot above this plane a curved surface representing the objective function which depends on the two variables and which is to be optimized. The objective function may express cost, weight, range, aerodynamic or propulsive efficiency, return on investment, or any combination of parameters. It is subject to functional constraints in accordance with given relationships between variables and parameters and to upper or lower bounds of variables. The side constraints define the permissible part of the curved surface where the optimum value has to be found, e.g. limits due to minimum sheet thickness, maximum stress, stalling speed, etc.

Thus, in a formal notation, the quantitative side of the design problem may be formulated as a problem of Nonlinear Mathematical Programming (NLP):

(1). " find X such that f(X,P) is at minimum constrained by g(X,P)< 0 and h(X,P)= 0"

where X is a vector of the design variables and Xmin and Xmax represent variable bounds, P is a vector of

constant parameters, f is an objective function, g is a vector of inequality constraints, and h is a vector of equality constraints.

Thus, in contrast to the graphical methods, the MDO technology mathematically traces a path in the design space from the initial toward improved designs (with respect to some figure of merit) and does it operating on a large number of variables and functions simultaneously - a feat beyond the power of human mind. However, the visibility of the reasons for the design decisions corresponding to the twists and turns of the search path remain obscured inside a "black box". Making these reasons visible to the designer and presenting graphically the salient features of the design space is a challenge that the MDO technology must recognize and meet, in order to inspire confidence in the optimization results. Post optimality and parameter sensitivity analysis can provide much information that can raise the confidence of the designer.

The idea of formulating a design problem in rigorous, mathematical terms, introduced in [15], had spawned a vast body of literature, including comprehensive survey papers, e.g., [l6], [17], [18], and [38], and has become a key component in the MDO methodology. Consistent with its origin, the MDO methodology has thrived to the largest extent in design of light-weight, aerospace structures, but is spreading to other engineering disciplines and non-aerospace applications. The MDO-type methods were particularly successful in space flight for trajectory optimization. Optimization has been applied to trajectory design problems for the past 25 years. Analytic optimization has been applied to solving two



and three burn orbit transfer problems for mission planning (estimating payload transfer capabilities). Boosters (Space Shuttle, Titan, Delta, Atlas) and upper stages (IUS, Centaur, PAM) use some form of trajectory optimization to design flight profiles to maximize payload (or reserve fuel) to some orbital conditions. Reentry problems have also been optimized to obtain maximum cross range or down range trajectories. Additionally the NASP trajectory will have to be optimized to obtain maximum payload to orbit i.e. improvements in the structure or engine efficiency will lead to new trajectories. These individual improvements must be weighed against total system performance to orbit (or some other objective, cost, reliability, or maintainability) to determine if the new system is worth the development cost. It should be noted that optimization does not remove the designer from the loop, but it helps conduct trade studies. The users should be [warned] not to accept solutions without careful examination, because if constraints are omitted from the problem they can often be violated by the optimization which can reduce safety factors and lead to system failure.

Formulation of the design problem for a system life cycle or concurrent engineering concept can be accomplished as a multi-objective optimization problem [l9]:

(2). " find X such that F(fi,(X,P)) is at minimum constrained by g(X,P) < 0 and h(X,P) = 0; where Xmin

< X < Xmax;"

which differs from the single objective formulation in Equation 1 by recognizing a set of individual objective functions fi, i = 1--->NF, which often may be contradictory. The functional relation f( ) may be

as general as admitting all fi's on equal footing and rendering F a vector, or as specific as a weighted sum

of the fi's which reduces F to a scalar. By specifying f( ), the designer defines the desired balance of the

various objectives fi. The multiobjective formulation represents a translation of the customer's ranked

requirements and goals, via the engineering theories and models underlying the design concept, into a mathematical statement of the design problem [9], [l 8].

Numerical optimization capabilities lag in comparative fidelity as characterized by the number of variables describing a design for optimization and for analysis (simulation). Equations are solvable routinely in analysis for tens of thousand, cautiously for hundreds of thousands, and as tour de force for over a million variables. Optimization variables for Nonlinear Mathematical Programming algorithms can not go beyond a few hundred to describe a design, unless there is some special problem structure that can be exploited then the number can be extended to ten thousand. Optimality Criteria (OC) methods do not have any limitation on the number of variables and problems with a million variables have been demonstrated, but they apply only if certain conditions are satisfied, considerably limiting classes of problems for which OC methods may be used. For example they are not applicable in problems whose analysis combines governing equations of very different physical phenomena as is typical for multidisciplinary applications such as the aerodynamics-structures-vehicle performance problem. In contrast, in some applications involving a single physical phenomenon, the OC techniques may be very effective even though they yield only a close approximation to a constrained minimum. The classic example of this is the Fully Stressed Design (FSD) technique that works well for homogeneous material structures but becomes questionable for structures with material mixtures of varying strength to



weight ratios.

Post-optimization analysis of optimal design for sensitivity of the optimal solution to parameters P is often useful for quick assessment of the impact of changes to the original problem formulation [20], [21], [40] .

For instance, if the P values needed to specify the F, g, and h functions in Equation 1 or Equation 2 may vary in an uncertainty range, it may be practical to optimize the design for the most probable P first. Subsequently, a range of new optimum designs may be approximated by extrapolation in the neighborhood of the nominal design using the derivatives of the optimal F and X. For example, consider a launch vehicle trajectory that has been designed to maximize reserve fuel a given mission. If the mission parameters (payload weight, target orbit, or launch vehicle specifications) change significantly then the trajectory for the vehicle must be reoptimized to find the trajectory that maximizes the reserve fuel for the new mission parameters. The optimum sensitivity analysis may also be very useful in multi objective optimization (Equation 2) for evaluation of the effect of the weighting factors subjectively introduced for converting a set fi's to a scalar F. Parameter sensitivity analysis is influenced by numerical

conditioning of the underlying problem and solution accuracy, therefore careful implementation is required to obtain good results [41].

A. System Level Optimization

Why System-Level, Multidisciplinary Optimization?

That question needs to be posed and answered first because a typical disciplinary specialist often tends to strive toward improvement of the objectives and satisfaction of constraints defined in terms of the variables of his discipline. In doing so he generates side effects that other disciplines have to absorb, usually to the detriment of the overall system performance. A classic example is aerodynamic design of a transport aircraft wing for a high lift-to-drag ratio by increasing the wing aspect ratio that may result in a structural weight penalty needed to alleviate flutter. That weight penalty subtracts from the performance benefit of the high lift-to-drag ratio and may actually result in a lower performance comparing to a reduced aspect ratio wing.

To examine the issue in more detail, consider first an approach to airframe structural sizing that is often used for a long-range, subsonic transport aircraft. It may be summarized as follows:

1. Develop aerodynamic shape optimal for the cruise aerodynamic performance (basically, maximizing the L/D).

2. Minimize structural weight under the stress and aeroelastic constraints, including flutter, taking into account that the structural deflections affect the aerodynamic loads and vice versa.

3. From the cruise aerodynamic optimal shape subtract the structural deflections obtained for the optimized structure under that condition to establish a jig shape. This will assure that the ideal



aerodynamic shape will be attained at least at one point during the cruise leg of the mission.

Let us now see what would happen, if we used this approach to a supersonic transport (SST) flying a mission depicted in Figure 17 whose Mach number diagram may look as illustrated in Figure 18. It is a subsonic/supersonic mission and let us suppose that we used the supersonic stage in the above sizing approach. Since there is only one jig shape, if we use it up for the supersonic stage, we will end up having to accept whatever shape the airframe deforms to under the subsonic stage cruise condition. That shape may be aerodynamically suboptimal and cause a drag penalty of deltaD1 relative to the shape aerodynamically optimal for that condition.

If we refer to the subsonic stage in the sizing procedure, we just move the drag penalty to the supersonic stage but do not remove it. To remove or, at least, drastically reduce that drag penalty we have to recognize that there is a three-way mutual dependence of the aerodynamic loads-structural sizing-deflected shape that we, as structures engineers can manipulate to our advantage by changing the structural stiffness, its magnitude and distribution, over the airframe. Without invoking the notion of formal optimization as yet, suppose that by judgment we increase the wing stiffness in the outboard area to reduce the elastic wing twist that contributes to the drag penalty under the subsonic stage cruise condition (the optimal supersonic shape has remained optimal because we compensated by the jig shape). That may cost a structural weight penalty of deltaWI which is, in general, bad for the performance. However, if drag is reduced from deltaDI to deltaD2 < deltaD1, generally a good influence, the performance analysis can be referred to evaluate the deltaWI against (deltaDl<deltaD2) as a trade-off.

The trade-off may come out positive or negative depending on the objective and usually there is a wide choice. A few examples are: the minimum take-off gross weight (TOGW) for given range, payload, and mission profile; the maximum payload for a given range, TOGW, and mission profile.

The above trade-off example is also only one of many. Suppose that the wing is strength-critical in 2.5g pull-up. Then, we may wish to allow the outboard wing more twist flexibility so that it can wash out thus alleviating the wing root bending moment and reducing the structural weight at the price of increased drag of the wing elastically deformed during the subsonic cruise.

Many such trade-offs have to be considered simultaneously, and a complicating factor is that they have to be resolved not only to end up with a positive net impact on the performance objective(s) but they also have to be solved without violating the constraints imposed by each of the participating disciplines, e.g., flutter, allowable stress, vehicle stability, controllability, etc. It is clear that the human judgment needs help from the computer for resolution of such a multitude of trade-offs.

Leaving the above example and returning to the generic discussion, it may be asserted that the user demand that drives the development of multidisciplinary analysis and optimization has been intensifying because:

1. major new aircraft design projects become fewer and farther apart in time, hence the past experience becomes less available as a guide in making the design decisions;



2. advanced aircraft tend to be an enormously complex system of interacting parts and disciplines and its ultimate performance hinges on the myriads of numerical interplay, some of them very subtly and beyond the power of human judgment to evaluate precisely. The ubiquitous challenge of design may be phrased as "How to decide what to change, and to what extent to change it, when everything influences everything else".

The integrated design process that was defined at the end of Section IB is intended to meet the above challenge by creating an environment on the quantitative side of design (Figure 13) in which the designer's decision making will be supported with a comprehensive, and quickly generated, numerical information presented in an easy-to-interpret format. It is not the purpose of this paper to systematically survey the state-of-the-art in the methodology for creation of the above environment or to endorse a particular approach or technique. Rather, its purpose at this point is to illustrate emergence of a new methodology for multidisciplinary design optimization by a few examples of methods whose initial application experience has been encouraging.

It is generally agreed that the challenge posed by the quantitative side of an advanced aircraft design as a complex system needs decomposition that breaks the large, intractable problem into smaller subproblems while maintaining the couplings among the subproblems. In the design office, this approach maps well onto the natural organization of engineers into groups by disciplinary and task specialization. It preserves and nurtures the advantages of the division of labor, including the concurrency of operations - the time-honored principle of industrial management first articulated by Adam Smith in the classic work "The Wealth of Nations" nearly 250 years ago [23].

B. Decomposition and Sensitivity Analysis

The decomposition approach stems from the realization that the analysis and sensitivity analysis that generate data optimization algorithms need may easily account for more than 90% of the total computational optimization cost. Hence the recent emphasis on the efficient sensitivity analysis that exploits modularity in application to complex systems. Numerous decomposition schemes have been proposed in literature and, undoubtedly, more will be developed in the future. For the purposes of this discussion it will suffice to name as two basic examples the methods for a hierarchic decomposition and a non-hierarchic decomposition.

Hierarchic decomposition. The concept of a hierarchic decomposition for engineering design was introduced in [24] using the algorithm from [25] as means for efficient calculation of the optimum sensitivity derivatives. Examples of this type of decomposition applied to structures may be found in [26], and a demonstration of its usefulness in multidisciplinary optimization to aircraft configuration was given in [27]. The hierarchic decomposition method exploits a special way in which the computational and decision making operations may be arranged in the design process of an engineering system. The arrangement is illustrated in Figure 19. Each box represents analysis and optimization of a subset of the entire system problem. The analysis information flow is topdown from the "Parent" black-box to the "Daughter" black-box. For example, a finite element analysis of the entire airframe may be a Parent that



transmits the boundary forces to a Daughter wing substructure and the natural vibration frequencies and modes to another Daughter representing aeroelastic behavior. The topdown flow ends when it reaches the bottom level of the black-box pyramid. Then, each black box solution is available and the optimizations begin progressing from the bottom level up.

Inputs received by a Daughter from a Parent are frozen as constant parameters for the duration of optimization performed inside of the Daughter black-box. Moving up to the Parent, one transmits the results of the Daughter optimization augmented with the derivatives of these results with respect to the parameters that the Parent has sent to the Daughter. These derivatives enable the Parent optimization to account by linear extrapolation on the effect of the Parent design variables on each Daughter constraints.

The procedure continues to the top of the pyramid. The top Parent represents the system level objectives and constraints and is controlled by the system level design variables. The effects of these variables on all the black-boxes in the pyramid below are accounted for by the optimum sensitivity derivatives transmitted from below. Since the procedure is based on first derivatives, it takes a few iterations to converge. Each iteration consists of the analysis sweep top-down and the optimization sweep bottom-up. With careful implementation the optimization on successive iterations becomes more efficient if warm/hot start capabilities are used. Since the Daughters do not communicate at the same level (no information transmission among sisters), the individual black box analyses and optimizations at each level may be performed in parallel.

Non-hierarchic decomposition. The non-hierarchic decomposition method allows for information multidirectional transmission among the black-boxes forming a system as depicted in Figure 20 for an example of a flexible, actively controlled wing. A system like this cannot be arranged into a Parent-Daughter pyramid shown in Figure 19. Its optimization may be executed as a single operation for the entire system and is guided by the system sensitivity measured by the derivatives of the system behavior (response) variables with respect to the system design variables.

The derivatives may be computed without finite differencing on the entire system analysis by a technique that:

1. solves the system at a baseline design point,

2. computes the partial sensitivity derivatives of the output from each black-box with respect to its input from other black-boxes and with respect to the design variables,

3. uses the above partial derivatives as coefficients to form a set of simultaneous, linear, algebraic equations whose solution yields the system sensitivity derivatives.

A review of various types of decomposition, including the hierarchic and non-hierarchic approaches, was provided in [28]. The mathematical concept underlying the non-hierarchic approach was introduced in [29] and [30]. Its applications in aerospace design were compared to that of the hierarchic decomposition in [31], and an example of its industrial use was described in [32]. Common to



optimization by both hierarchic and non-hierarchic decomposition is its reliance on the sensitivity analysis as a generic numerical method in engineering analysis [33] as well as a disciplinary method of the type described for structures in [34] and for aerodynamics in [35].

How to decompose a system. When the system at hand is new and there is no past experience in guiding its decomposition, one may benefit from the use of a formal technique that converts a set of randomly sequenced black-boxes into a set ordered into a hierarchic, nonhierarchic, or a mixed, hierarchic/non-hierarchic arrangement. The technique formalism requires that each black-box be defined as a source and a recipient of information. As a source, the blackbox sends information through its vertical sides, horizontally, to the left and to the right. This definition is illustrated in Figure 21.

Initial random sequencing is presented by a diagonal chain of modules shown in Figure 22. The execution sequence is initially assumed to proceed from the upper left corner to the lower right corner and the modules are positioned randomly along the diagonal. Each off-diagonal dot marks a data interface indicating that the output moving along the horizontal line is directed along the intersecting vertical to the recipient module. The dots in the upper right triangle mean feeding the data forward (downstream), by the same token the lower triangle dots mean feedback (upstream). Each instance of a feedback calls for an iteration because module A upstream depends on the output from a successor module B downstream.

By a systematic row and column permutation executed by a computer program, the random picture of Figure 22 may be transformed into an ordered sequencing shown in Figure 23. The transformation goal was to eliminate as many feedback instances as possible. It was not possible to eliminate them all in this particular case. However, their number was reduced and the remaining feedback instances have been clustered. That clustering suggests decomposition shown in Figure 24. It is a hybrid decomposition, hierarchic with respect to the clusters, each represented by a box in the pyramid, and non-hierarchic inside each cluster. Software tools became recently available for generating this type of decomposition from the initial, unorganized set of computational modules as described in [36].

C. Concluding Remarks on Optimization

The above examples of methods now under development and testing should not be regarded as the last word but only the beginning in evolution of a new methodology for quantitative support of the design process. One common thread of the examples discussed in the foregoing is the concern about creating an environment in which the engineer's mind and computer interact drawing on the best resources of each. This concern is expected to alleviate misgiving some practicing engineers may have about the formal design methodology that was offered, on occasion in the past as an "automated design". That was a misrepresentation that might have been an underlying cause of the lag of applications behind the theoretical developments noted in the survey in [16].

The other common thread is the concern about modularity of implementation necessary to ensure flexibility, open-endedness, and ability to accommodate a variety of the information sources, including



judgmental estimates, statistics, references, and experiments, in addition to computer programs. Modularity is also seen as a prerequisite for exploiting the computer technology progress in multiprocessor machines and distributed computing. Finally, there is a pervading concern for making the information exchanged among disciplinary specialists quantified and precise to provide a basis for the qualitative discourse these specialists are engaged in.

With these concerns in mind, one may foresee further developments as encompassing new algorithms for decomposition, disciplinary and system sensitivity analysis, effective search and optimization of the design space, and AI-based tools making all this user-friendly.

The central role of the disciplinary and system sensitivity analyses was apparent in the above method examples. Disciplinary sensitivity analysis by quasi-analytical approach is now routine only in structures and immediate emphasis is needed on developing a similar capability in CFD - the other major consumer of computer resources in aircraft design. The system optimization will become well-rounded when all contributing disciplines are liberated as much as possible and practical from the tedium of finite differencing by augmenting their analyses with sensitivity algorithms. Progress in the techniques for search and optimization in the design space is also important for the overall effectiveness and efficiency of the methodology as are procedures for tying together that search with analysis, sensitivity analysis, and approximate analysis, including the approach of statistically-fitted response surface methods. Improvements in the search techniques are needed for effective identification of multiple local minima - a vexing problem that thus far lacks a rigorous mathematical solution for cases with more than a few variables. One should also keep in the field of view the optimality criteria as an alternative to the search of the design space. Finally, the development should be kept open to accommodate innovations such as the self-learning neural nets, and genetic algorithms, to mention but a few examples of the cutting-edge approaches.

As always in methodology development, the ultimate test of usefulness is in applications. Therefore, a systematic cooperation of the theoreticians, implementers, and users who apply the tools and influence the theory and implementation with their observations and wishes must be an intrinsic part of that development. The benefits from introduction of the new methodology will be amplified if that methodology is applied early in the design process where most of the leverage is available.

VI. TRANSITIONING TO THE MDO ENVIRONMENT

The previous sections of this white paper have reviewed different aspects of MDO. This section will provide some thoughts on how to evolve to a concurrent engineering (CE) environment and the role MDO for aerospace systems will play in this transition. The goal is to achieve the compression of the tasks in the Design Phase illustrated in Figure 3 and redistribution of the effort among the engineering disciplines as indicated by the horizontal bars in Figure 12. The expected end result is more design freedom retained longer into the design process and more information about the object of design gained earlier in the process as portrayed by the curves in Figure 12.



To accomplish the above one needs to develop an environment for the integrated design process as defined at the end of Section IB The following specific tasks should constitute that development:

1. Identify information exchange requirements - each discipline describes its input and output information.

2. Establish unified numerical modeling parameterized in terms of the design variables - a consistent vehicle geometry must be the basis for all mathematical models, and changes to the geometry must be centrally coordinated.

3. Establish a data management system for a quick and easy location and transfer of the information needed by the engineers and by the computational tasks, and for generation of good initialization data for the optimization tasks.

4. Develop mathematical models for manufacturing, reliability, supportability, and life cycle cost, to augment the classical discipline models for a complete implementation of the CE idea.

5. Assemble an efficient design-oriented analysis capability. A design-oriented analysis is tailored to support applications in design characterized by: repetitive use with only a subset of the input changed in each repetition, need for sensitivity data, use of the mathematical models of varied degree of refinement to trade accuracy for computational cost.

6. Efficiently generate discipline design sensitivities.

7. Assemble a system sensitivity analysis for vehicle optimization - system design variables will be identified and used to quantify the effects of design changes on the system behavior.

8. Improve optimization algorithms for effective handling of very large number of design variables, disjoint and nonconvex design spaces, multiple minima, and multiobjectives.

9. Improve post-optimum sensitivity analysis for greater computational efficiency, and for effectiveness in the extrapolations across the points where the set of active constraints changes its membership (see [20] for the description of a problem caused by changes in the active constraint set).

10. Develop a method for systematic developments and evaluation of design changes toward meeting the objectives and constraints in form of an iterative, multidisciplinary optimization process.

The above development will result in a new, higher level of the state-of-the-art in engineering design. It is anticipated that industries, government laboratories, and universities will all contribute building blocks. There will be an accumulation of generic and proprietary, product-tailored tools, and of partial implementations of the entire process. Pilot projects will accumulate experience, demonstrate benefits,



and build confidence. Gradually, a complete, new, integrated design process will evolve and be used for creating aerospace vehicles.

That process will be a logical expansion to the Design Phase of the CE concept defined in [37]. The most important ten CE characteristics from the above reference (slightly rephrased for the context of this discussion) and their relationship to MDO are listed in Table 1 to emphasize once again the view of MDO as a key new component in CE. In that development, the AIAA TC-MDO has a role described in Section VIII.

TABLE I

TEN CHARACTERISTICS REQUIRED FOR THE

CURRENT ENGINEERING PROCESS

CHARACTERISTICS WHAT IS REQUIREDMDO

RELATIONSHIP

1Compreh. Sys. Eng. Proc. Using Top-Down Design Approach

Authoritative, but Particip. Top Mgt ; System Eng. Mgt. Plan (SEMP) ; Automated Config. Mgt/Control

Decomposition

2Strong Interface with Customer

Methods for Translation of Voice of Customer Into Prod/Process Characts.

Optim. Methods

3Multi-Function Sys. Eng. and Design Teams

Management and Peer Acceptance; Equal or Near Equal Analysis - Cap.

Decomposition and Sensitivity Analysis

4 Continuity of the TeamsTraining org. accept and Incentive Program

5Practical Eng. Optim. of Product & Process Characts.

Methods for Incorp. Qual. & Quant. Optim. Methods

Compat. of Num. Optim. Methods with Other Methods

6Design Benchmarking Through Creation of a Dig. Prod. Model

Design by Feature Methods Plus Data Exchange Stands

Sensitivity Analysis and Optim. Methods



7Simul. of Product Perf. and Manuf. Process

Destrib. Simul. Cap. with Varying Levels of Fidelity

Sensitivity Analysis and Optim. Methods

8Experiments to Confirm/Change High Risk Predictions

Design of Experiments Methods for Variability Reduction of High Risk Characs.

9Early Involvement of Subcontractors/Vendors

Accept. by Top Mgt. and Peers Plus Organ. Decomposition

Decomposition

10Corporate Focus on Contin. Improve. & Lessons Learned

Design Tracking and Library Access through an Autom. Config. Mgt./Control System

Decomposition, Sensitivity Analysis and Optim. Methods.

VII. CONCLUSIONS

Multidisciplinary Design Optimization (MDO) has been rapidly gaining recognition as a new, engineering discipline that assumes a key role in development of advanced aerospace vehicles whose common characteristic is that they are complex engineering systems. In its role of a catalyst and conciliator of the disciplinary requirements and interactions, MDO becomes as important for success of design as any traditional engineering disciplines. MDO has been reviewed in the historical context of the aerospace design process evolution and in the context of the present day and future challenges posed by advanced aircraft and spacecraft. If this White Paper were written a decade ago, in all likelihood it would have emphasized design optimization for improved performance. The recently evolved understanding that performance is only a subset of the overall product quality that must include the cost of development, manufacturing, and maintenance has replaced that emphasis in this paper with one that includes the entire life cycle of the aircraft or spacecraft, with the cost of that life cycle as one of the key objectives. This meshes very well with idea of Concurrent Engineering whose main goal is to move the manufacturing and supportability considerations upstream into the design process in order to compress the entire development and to assure that these considerations get in the design process an attention equal to that traditionally afforded the vehicle performance. This basic idea of Concurrent Engineering - the compression of the major life cycle phases of Design, Manufacturing, and Maintenance that were sequentially arrayed heretofore - applies also to the phase of design. That phase also may be "compressed" in the sense of staggering the conventional sequence of operations and decisions.

MDO is seen as a means by which to achieve the above compression by bringing more information about the entire life cycle and the vehicle performance and cost aspects earlier into the design process. This will enable engineers to make design decisions on a rational basis that gives equal consideration to all the influences disciplines exert on the system, directly, or indirectly through their complex interactions. Doing this early in the process exploits the leverage of the uncommitted design variables.



On the other hand, it is equally important to extend the MDO-based approach to the later phases of the design process in order to take advantage of the new information that becomes available during that process through creative thinking, analysis, experimentation, and exploration of alternatives. In order to do that, the design variables that in the conventional design process are decided and set early, need to be retained as free variables much longer into the process. Using the MDO technology one may achieve this because the overall methodology of system analysis, and optimization based on sensitivity data remains the same throughout the process. The variable element is analysis that deepens as the process moves on.

The MDO methodology is well-suited to blend in the above analysis the traditional, performance-oriented design considerations with those posed by the remainder of the life cycle because it is generic and capable of including anything represented by a mathematical model, whether that model is derived rationally or established heuristically. However, it is necessary to develop such models first and this is one of the several specific developments identified in the White Paper. Another development direction of a high pay-off potential pointed out is toward the probabilistic methods, multiobjective capability, and facilities to accommodate the "soft" (negotiable) constraints as distinct from the hard constraints in optimization - as required by the applications of MDO extended to manufacturing, maintenance, and economics.

The key premise expounded for the MDO approach in the White Paper is that it is not a "push button" design. Instead, MDO is an environment in which the human ingenuity combines with the power of mathematics and computers in making design decisions. The boundary between the formal mathematical methods and the human judgment is, of course, fluid. Nothing should prevent an engineer either from delegating a repetitive tedious routine to a formal method or from substituting judgment for a formal method or from overriding the method results.

Based on that premise, the MDO-enhanced design process has the clear potential for radically improved product quality achieved by systematic exploration of the alternatives created by human ingenuity and bringing each of these alternatives to the optimal state among which a fair choice can be made by engineer's judgment.

VIII. THE ROLE OF THE AIAA MDO TC

The TC-MDO should be a focus for MDO activity, providing a forum through which the efforts of researchers can be disseminated to users and potential users in industry and government establishments. At the same time, feedback from users will establish future requirements and goals.

In order to maintain such a forum, the TC should seek membership among all engineers and computer specialists, involved in design, and design support, of aerospace vehicles of all major categories such as aircraft, launch vehicles, spacecraft, missiles, transatmospheric vehicles, etc.

To achieve the goals called for by its charter, the TC should undertake the following tasks:



(1) DEFINE the technological sphere of interest in multidisciplinary design optimization regarded as a new engineering discipline and one of the key elements in concurrent engineering and total quality management.

(2) GATHER information on MDO

- university research

- industry practices and applications

- government research and requirements

(3) EDUCATE

- upper and middle management in industry and government

- R&D engineers in industry and government

- university graduate and post graduate students

(4) GUIDE research efforts by suggesting areas for study, and future goals.

To accomplish these tasks the following TC-MDO subcommittees have been formed:

(1) White Paper - Act as a focal point for a periodic generation of a white paper expressing the collected views of the TC and describing state-of-the-art in integrated MDO.

(2) Computer Technology and Optimization - Act as a focal point for information concerning optimization algorithms and their application and advances in computer technology.

(3) Education - Act as a focal point on all issues relating to education in MDO.

(4) Liaison - Act as a focal point to coordinate activities and provide a channel of communication with other active AIAA TC's.

(5) Conference Support - Act as a control focus of activity and resources of the TC-MDO in support of AIAA sponsored and co-sponsored conferences, symposiums, and shows.

(6) Publications - Act as a focal point for generation and distribution of all publications of the TC- MDO.



(7) Benchmark - Act as a focal point for devising effective and practical test cases for MDO methods.

(8) Emerging Methods - Act as a focal point for identifying emerging methods applicable to MDO.

(9) Material Optimization - Act as a focal point for coordinating research efforts in the area of optimum design of materials, and their inclusion into the design of complex systems together with the other relevant disciplines.

(10) Awards - Act as a focal point for identifying and recognizing significant contributors to MDO.

REFERENCES

1. U. Haupt, "Decision-Making and Optimization," NPS-67 Hp 77021A, February 1977.

2. R. I. Winner, et al., "The Role of Concurrent Engineering in Weapons System Acquisition," Report R-338, IDA, Washington, DC, December 1988.

3. D. P. Schrage, "The Impact of TQM and Concurrent Engineering on the Aircraft Design Process," Keynote Address Vertical Lift Aircraft Design Conference, San Francisco, CA, Jan 17, 1990.

4. Sobieszczanski-Sobieski, J.; Barthelemy, J.-F. M.; and Giles, G. L.: "Aerospace Engineering Design by Systematic Decomposition and Multilevel Optimization," 14-th Congress of the International Council of the Aeronautical Sciences (ICAS), Proceedings of; Toulouse, France, Sept. 1984; also published as NASA TM 85823 NASA; Langley Research Center, Hampton, VA, June 1984.

5. "Aeronautics Technology, Possibilities for 2000: Report of a Workshop," National Academy Press. Washington, D.C. 1984.

6. J. L. Hunt and J. G. Martin, "Hypersonic Airbreathing Vehicle Conceptual Design (Focus on Aero-Space Plane)," Recent Experiences in Multidiscipiinary Analysis and Optimization, Hampton, VA, September 28-30, 1988.

7. "The Future of Military R&D: Towards a Flexible Acquisition Strategy," Institute for Defense Analysis (IDA), July 1990.

8. "Goals and Priorities For Research In Engineering Design: A Report to the Design Research Community," The American Society of Mechanical Engineers (ASME), New York, NY, July 1986.

9. E. Tse and W.E. Cralley, "Management of Risk and Uncertainty in Product Development Processes," IDA Paper P-2153, June 1989.



10. J. E. Rogan and W. E. Cralley, "Meta Design," IDA, Paper P-2152, Jan 1990.

11. R. E. Fulton and J. I. Craig, "Information Framework Technology For Integrated Design/Engineering Systems," Results of NSF and Georgia Lnstitute of Technology Workshop, Callaway Gardens, GA, March 13-15, 1989.

12. First Draft, "Requirements For Concurrent Engineering Infonnation Architecture," CALS/ISG/CE Framework Subtask Group, June 11, 1990.

13. "High Performance Computing Networking For Science - Background Paper," Congress of the United States, Office of Technology Assessment, September 1989.

14. R. G. Voigt: "Requirements For Multidisciplinary Design of Aerospace Vehicles on High Performance Computers," NASA Contractor Report 181915, September 1989.

15. Schmit, L. A.: "Structural design by Systematic Synthesis," Proc. of the Second National Congress on Electronic Computation, Structures Division ASCE, Pittsburgh. PA, Sept. 1960. pp. 105-132.

16. Ashley, H.: "On Making Things the Best - Aeronautical Uses of Optimization," AIAA J. of Aircraft, Vol 19, No 1. Jan. 1982, pp 5-28.

17. Siddall, J. N.: "Frontiers of Optimal Design," ASME J. of Mechanical Design, Oct. 1983.

18. Betts, J. T.: "Frontiers in Engineering Optimization," ASME J. of Mechanisms, Transmissions. and Automation in Design; June 1983.

19. Eschenauer, H., Koski, J.. and Osyczka, A. (Editors): "Multicriteria Design Optimization," Springer Verlag, 1990.

20. Sobieszczanski-Sobieski, J.; Barthelemy, J.-F. M.; and Riley, K. M.: "Sensitivity of Optimum Solutions to Problem Parameters," AIAA Paper 81-0548R, and AIAA J, Vol 20, No 9, September 1982, pp. 1291-1299.

21. Barthelemy, J.-F. M.: Sobieszczanski-Sobieski, J.: "Extrapolations of Optimum Designs Based on Sensitivity Derivatives," AIAA J., Vol 21, No 5, May 1983, pp. 797-799.

22. Aviation Week and Space Technology, June 18, 1990, pp. 98.

23. Adam Smith: "The Wealth of Nations," 1791.

24. Sobiesznanski-Sobieski, J.: "A Linear Decomposition Method for Large Optimization Problems -



Blueprint for Development," NASA TM 83248, February 1982.

25. Barthelemy, J. F.; and Sobieszczanski-Sobieski. J.: "Optimum Sensitivity Derivatives of Objective Functions in Nonlinear Programing," AIAA J, Vol 22, No 6, June 1983, pp. 913-915.

26. Sobieszczanski-Sobieski. J.; James, B. B.; and Dovi, A. R.: "Structural Optimization by Multilevel Decomposition," AIAA J., Vol 23, No 11, November 1985, pp. 1775-1782.

27. Wrenn, G. A.; and Dovi. A. R.: "Multilevel Decomposition Approach to the Preliminary Sizing of a Transport Aircraft Wing," AIAA Journal of Aircraft, Vol 25, No 7, July 1988, pp. 632-638.

28. Barthelemy, J. F.: "Engineering Design Applications of Heuristic Multilevel Optimization Methods," Second NASA/Air Force Symposium on Recent Advances in Multidisciplinary Analysis and Optimization; Hampton, VA, September 28-30, 1988, Proceedings to be published as NASA CP - No 3031.

29. Sobieszczanski-Sobieski, J.: "On the Sensitivity of Complex, Internally Coupled Systems," AIAA/ASME/ASCE/AHS 29th Structures, Structural Dynamics and Materials Conference, Williamsburg, VA.,April 1988; AIAA Paper No CP-88-2378. and AIAA J.. Vol 28, No 1, Jan. 1990, also published as NASA TM 100537, January 1988.

30. Sobieszczanski-Sobieski, J.: "Sensitivity Analysis of Complex Coupled Systems Extended to Second and Higher-Order Derivatives," AIAA J.. Vol 28, No 4, Apr. 1990, also published as NASA TM 101587, April 1989.

31. Sobieszczanski-Sobieski , J.: "Sensitivity Analysis and Multidisciplinary Optimization for Aircraft Design: Recent Advances and Results," Int'l Council for Aeronautical Sc., Proceedings of 16th Congress, Jerusalem. Aug.- Sept. 1988; Vol 2, pp. 953-964.

32. Abi, F.F.; Ide. H.; Shankar, V. J.; and Sobieszczanski-Sobieski, J.: "Optimization for Nonlinear Aeroelastic Tailoring Criteria," Int'l Council for Aeronautical Sc., Proceedings of 16th Congress. Jerusalem, Aug.-Sept.. 1988; Vol 2. pp 1083-1091.

33. Proceedings of the Symposium on Sensitivity Analysis in Engineering, NASA Langley Research Center, Hampton, VA, Sept. 1986; Adelman, H. M.; and Haftka, R.T. - editors. NASA CP-2457, 1987.

34. Adelman. H. A; and Haftka, R. T.: "Sensitivity Analysis of Discrete Structural Systems," AIAA J., Vol 24, No 5, May 1986, pp. 823-832.

35. Yates, E. C.: "Aerodynamic Sensitivities from Subsonic, Sonic, and Supersonic Unsteady, Nonplanar Lifting-Surface Theory," NASA TM 100502, September 1987.



36. Rogers, J. L.: "A Knowledge-Based Tool for Multilevel Decomposition of a Complex Design Problem," NASA TP 2903, 1989.

37. CALS Technical Rep 002: "Application of Concurrent Engineering to Mechanical Systems Design," Final Report of RM Mechanical Design Study, June 16, 1989.

38. Schmit, L. A.: "Structural Synthesis - Its Genesis and Development," AIAA J., Vol. 19, No 10, 1981, pp. 1249-1263.

39. Betts, J. T.; and Huffman, W. P.: "The Application of Sparse Nonlinear Programming to Trajectory Optimization," AIAA Paper 90-3448, Aug. 1990, Proceedings AIAA Guidance Navigation and Control Conference.

40. Fiacco, A.: "An Introduction to Sensitivity and Stability Analysis in Nonlinear Programming," 1983, Academic Press.

41. Hallman, W. P.: "Sensitivity Analysis for Trajectory Optimization Problems," AIAA Paper 90-0471, Jan. 1990.

APPENDIX I

SURVEY OF THE INDUSTRY MDO PRACTICES

In the summer of 1990, the AIAA Technical Committee for Multidisciplinary Design Optimization conducted an industry survey on the use of the MDO technology. The survey was taken to their companies in the U.S.A. and in Europe by the TC members who used their company contacts to answer the survey questions. Thus the answers received were representative of the company rather than individual opinions.

The first part of this appendix defines the survey purpose and background. A Summary of the results is given in the second part.

Survey Definition

Purpose The survey purpose is to determine the ways and means the aerospace industry uses to resolve trade-offs that arise in design process of aerospace vehicles, with emphasis on the trade-offs that involve two or more engineering disciplines.

Background The following examples illustrate the notion of a trade-off. By increasing the aspect ratio of a transport wing, the drag-due-to-lift is reduced thus improving range for a given payload. However, a higher aspect ratio wing, in general, will weight more tending to decrease range. The net effect of



change in aspect ratio on range may then be positive or negative, depending on the strength of the drag and weight influences.

The kill probability of an air-to-air missile may be increased by making the missile more agile, or making the fighter that launches the missile more agile, or both. There is a cost associated with adding agility to the missile and another cost of adding agility to the fighter. In what proportions should one allocate a fixed total budget to the missile development and to the fighter development to get a missile/fighter system of the maximal kill probability?

The pointing accuracy of a large antenna dish attached to a spacecraft constructed as a large, actively-controlled structure, may be improved by making the structure more rigid, or by adding more capability to the control system. There are weight penalties, and cost penalties for both alternatives. What is the "best" mix of added structural rigidity and added capability of active-control system to achieve the required pointing accuracy?

As the examples illustrate, the trade-off arise at high-level (system level) as well as more detailed level, in all classes of vehicles. For proper resolution they involve numerical information and judgment.

Regarding numerical information, there is a body of mathematical methods such as: disciplinary and system analyses, sensitivity analysis (to compute derivatives of the dependent variables with respect to independent variables by analytical, quasi-analytical, or finite difference techniques), parametric studies, and formal optimization. On the judgment side, the approaches range from unstructured decision making to highly organized and disciplined procedures for generation, evaluation, and recording of the judgmental decisions.

It is not clear, however, where the center of gravity lies between the extremes of the all mathematical and all judgmental ways of resolving the trade-offs, and what are the most often used techniques in both categories. It is also not clear whether things are as they should be with regard to the above, or whether they should be changed. It is important to know the industry opinion on this issue for effective planning and development of the pertinent methodology and engineering education. This survey should shed some light on the issue.

Format The survey subject is really too complex to boil down to a simple, check-a-box, questionnaire. Therefore, a free format essay is preferred (please, include identification of your company, your position, and give an example of a product to which the issues raised in this survey would, typically, apply). The minimum length for a meaningful answer is probably less than one single-spaced page. To facilitate the evaluation, the maximum length should not exceed 3 pages. However, a questionnaire format is also available, if time for a free-format answer cannot be found.

Summary of the Survey Results: Questions and Answers.

Most of the survey returns came in the Questionnaire Format but several were in an all free-format



narrative. The survey Questionnaire Format questions are reproduced in full. Most questions called for a numerical answer. The numerals following each question represent averages of the survey return. The answers were also illustrated by placing the averages on the numerical axis. Since there is no uniform definition of design stages, the answers were classified as pertaining to early and late phases of design and marked by E and L, respectively. The averages include also the information extracted judgmentally from the free narrative results. Questions 4 and 6 in the Questionnaire called for free-format answers and are followed by paraphrased extracts from these answers and from those returns that came in an all-free-format narrative.

1. Assuming a scale from -5 (all mathematical) to +5 (all judgmental), place on the scale the center of gravity of the ways by which the design trade-offs are being resolved, for each design stage. Notes: 1) results are reported for early/late design stages, 2) "system" means a complete vehicle.

-1.1/ -2.2 (early/late)

Mathematical.......|.........Judgmental

-5...-4...-2...-1...0...1...2...3...4...5

...........L....E....|........................

2. In the judgmental decision making, where is the center of gravity between the extremes of very formal organizational procedures (-5) and unstructured process (+5). Use a format as in answer 1.

-1.1/ -2.2

Mathematical.......|.........Judgmental

-5...-4...-2...-1...0...1...2...3...4...5

...........L....E....|........................

3. For the numerically generated information, please, evaluate how much does your organization rely on the following mathematical tools, using a scale from 0 (not used) to +5 (used very often, regarded as essential).

Analysis

Disciplinary analysis 4.2/4.4

0.....1.....2.....3.....4.....5



...........................EL

System sensitivity by parametric study: 3.0/3.5

0.....1.....2.....3.....4.....5

...................E..L.........

System sensitivity by finite differences: 2.8/1.5

0.....1.....2.....3.....4.....5

..........L......E..............

System sensitivity by analytical/semi-analytical method: 3.0/2.3

0.....1.....2.....3.....4.....5

...............L...E............

Optimization

Parametric study/disciplines: 4.0/3.5

0.....1.....2.....3.....4.....5

.......................L.E......

Parametric study/system: 4.2/4.2

0.....1.....2.....3.....4.....5

..........................L/E...

Formal numerical optimization/disciplines: 3.0/2.8

0.....1.....2.....3.....4.....5

..................LE............



Formal numerical optimization/system: 3.0/2.0

0.....1.....2.....3.....4.....5

.............L.....E............

4. If formal, numerical optimization is used, name a few techniques, e.g., nonlinear programming (NLP), linear programming (LP), optimality criteria, and names of a few optimization programs (Early/H for in-house developed, A for acquired from outside).

NLP, LP, Fully Stressed Design, Optimality Criteria (FASTOP), Design of experiments (DOE), Mix of in-house and acquired, Most of NLP at early stages, little in Aerodynamics, OC and FSD at later states in Structures. Formal optimization of the configuration in early stages, after that structural optimization with the configuration frozen.

5. For each design stage indicate whether the present system adequately identifies the best design options and configurations, accounting for complex interactions among the system parts and governing disciplines. Use scale from 0 (very inadequate) +5 (completely adequate).

2.9/3.2

0.....1.....2.....3.....4....5

...................E.L.........

6. Finally, indicate whether you are satisfied with status quo or would like to see a change.

Formal optimization applied to configuration (system) very early, then configuration frozen, optimization limited to structures and control.

The above confirms the paradox: In the design process, "the knowledge increases with time, the freedom to act on that knowledge decreases with time".

Present ways adequate to design good vehicles, not adequate "to prevent problems from occurring late in the design cycle which require costly and sometimes futile efforts to correct".

After the configuration is frozen, problems arising in a particular discipline are expected to be solved by a fix limited to that discipline (e.g., flutter fixed by stiffening of the wing structure or by balance masses).

Organizational structure and culture must change to bring about an effective MDO into the design process.



The best place for MDO is in the middle of the design process when enough hard information is available but before too many variables get frozen and before the problem size mushrooms.

Better infrastructure is essential: faster, bigger computers, visualization, data bases.

Lack of the system sensitivity information hampers the design process.

"Higher order" disciplines (e.g., aeroelasticity) are particularly limited by the above.

High priority should go to a complete automation of the routine engineering tasks, including AI methods.

MDO should be used at ALL stages of design

Mathematical models of different degree of refinement should be used in a coordinated manner throughout the design process.

Doing work faster = the MDO advantage.

Need a better handle on the multiple minima problems and more visibility into the optimization process to gain confidence in the results.

MDO has a potential as a crucial component in the Concurrent Engineering.

The best way to introduce MDO is by incremental changes.

Trajectory optimization is a good example of an application where optimization is used because no other means would do.

APPENDIX II


Multidisciplinary Design Optimization (MDO)

Membership Roster

NAME ORGANIZATION



Dr. Jaroslaw Sobieski, Chairman

NASA Langley Research Center, MS 246

Hampton, VA 23681-0001

Mr. Jan Aase

Engineering Computing Systems Technology

MD 24043

General Electric

1000 Western Ave.

Lynn, MA 01910

Dr. Frank Abdi

Rockwell International

P.O. Box 92098

201 N. Douglas St. #GB15

El Segundo, CA 90009

Dr. Ramesh K. Agarwal

McDonnell Douglas Research Laboratories

Dept. 222/B.110

P.O. Box 516; MC 1111041

St. Louis, MO 63017

Dr. Todd J. Beltracchi

The Aerospace Corporation

P.O. Box 92957

Los Angeles, CA 90009-2957

Dr. Laszlo Berke

NASA Lewis Research Center

21000 Brookpark Rd.

Cleveland, OH 44135



Mr. Christopher Borland

Boeing Commercial Airplane Group

P.O. Box 3707; MS 7H-94

Seattle, WA 98124

Dr. Kyung K. Choi

College of Engineering

The University of Iowa

Iowa City, IA 52242

Mr. Robert D. Consoli

General Dynamics Fort Worth Div.

Dept. 0635

P.O. Box 748 MZ 2872

Ft. Worth, TX 76101

Dr. Evin Cramer

Boeing Comp. Services

P.O. Box 24346, M/S 7L-21

Seattle, WA 98124-0346

Mr. Alan J. Dodd

Douglas Aircraft Co.

McDonnell Douglas Co.

3855 Lakewood B., M/S 18-86

Long Beach, CA 90846

Prof. George S. Dulikravich

Aerospace Engineering Dept.

233 Hammond Bldg.

The Pennsylvania State University

University Park, PA 16802



Mr. George C. Greene

Fluid Mechanics Div., MS 163

NASA Langley Research Center

Hampton, VA 23665

Dr. Zafer Gurdal

Engineering Sc. and Mechanics Dept.

Virginia Polytechnic Institute

Blacksburg, VA 24061

Dr. Prabhat Hajela

Dept. of Mechanical Engineering

Aeronautical Eng. and Mechanics

5020 Jonsson Eng. Ctr.

Rensselaer Polytechnic Institute

Troy, NY 12180

Dr. Wayne Hallman

The Aerospace Corporation

P.O. Box 92457

Los Angeles, CA 90009

Dr. K. Scott Hunziker

Boeing Aerospace

P. O. Box 3999, M/S 82-97

Seattle, WA 98124-2499

Dr. Erwin H. Johnson

MacNeal Schwendler Co.

815 Colorado Blvd.

Los Angeles, CA 90041



Dr. Ilan Kroo

Dept. of Aeronautics and Astronautics

Stanford University

Stanford, CA 94305

Mr. Michael Love

General Dynamics Fort Worth Div.

P. O. Box 748, MZ 2824

Ft. Worth, TX 76101

Dr. John K. Lytle

NASA Lewis Research Center, MS AAC-1

21000 Brookpark Rd.

Cleveland, OH 44135

Mr. Philip Mason

Grumman Aircraft Systems Div.

MS B43/35

Bethpage, NY 11714

Dr. Hirokazu Miura

System Analysis Br.

NASA Ames Research Center, MS 237-11

Moffett Field, CA 94035

Mr. Douglas Neill

Northrop Aircraft Div.

Dept. 3854/82, 1 Northrop Ave.

Hawthorne, CA 90250

Mr. Larry G. Niedling

McDonnell Aircraft Co.

P. O. Box 516, M/C 03412 80

St. Louis, MO 63166



Ms. Beth Paul

General Dynamics Ft. Worth Div.

P. O. Box 748, MZ 2208

Ft. Worth, TX 76101

Dr. Nick Radovcich

Lockheed Aeronautical Systems Co.

Dept. 76-12, Bldg. 63GE, Plant A-1

P. O. Box 551

Burbank, CA 91520

Mr. Bruce A. Rommel

Douglas Aircraft Co.

McDonnell-Douglas Corp.

M/S 18-86

Long Beach, CA 90846

Dr. Vijaya Shankar

Rockwell Int'l. Science Center\

P. O. Box 1085

Camino del Rios

Thousand Oaks, CA 91360

Dr. Daniel P. Schrage

School of Aerospace Engineering

Georgia Institute of Technology

Atlanta, TA 30332

Mr. Otto Sensburg

MBB Ottobrunn

P. O. Box 80 11 60

8000 Munich 80 Germany



Mr. J. Tulinius

Rockwell International Corp.

North American Aerospace Oper. 011 GC02

P. O. Bhox 92098

201 N. Douglas St.

El Segundo, CA 90009

Dr. Gary Vanderplaats

VMA Engineering

5960 Mandarin Ave., Suite F

Goleta, CA 93117

Dr. Vipperla Venkayya

Air Force Wright Research & Dev. Center

FIBR

Wright-Patterson AFB, OH 45433-6553

Dr. B. P. Wang

University of Texas at Arlington

P. O. Box 19023

Arlington, TX 76019

Mr. John W. Hayn

McDonnell-Douglas Missile Co.

P. O. Box 516, M/C 270 0120

St. Louis, MO 63166


http://endo.sandia.gov/AIAA_MDOTC/sponsored/Fig.1.gif

http://endo.sandia.gov/AIAA_MDOTC/sponsored/Fig.1.gif12/29/2006 12:29:36 PM











































What is MDO?

What is MDO?

Some popular definitions for Multidisciplinary Design Optimization (MDO):

● A methodology for the design of complex engineering systems and subsystems that coherently exploits the synergism of mutually interacting phenomena.

● Optimal design of complex engineering systems which requires analysis that accounts for interactions amongst the disciplines (or parts of the system) and which seeks to synergistically exploit these interactions.

● "How to decide what to change, and to what extent to change it, when everything influences everything else."

For a more detailed description, refer to the Current State of the Art On Multidisciplinary Design Optimization (MDO) white paper.




http://endo.sandia.gov/AIAA_MDOTC/faq/MDOTC_whatisMDO.html12/29/2006 12:29:56 PM


Functions performed by the MDO TC

What functions are performed by the MDO TC?

The AIAA MDO TC sponsors a variety of activities to further the development, application, and teaching of MDO technology:

● Service ❍ Conferences supported by the MDO TC ❍ Short Courses supported by the MDO TC ❍ White Papers and Publications prepared by the MDO TC ❍ Awards to recognize outstanding contributions in the field of MDO. ❍ Assessment and recommendation of improvements in the teaching of MDO technology.

● Technical ❍ Provision of a MDO test problem suite for the benchmarking of optimization methods in

the aerospace community. ❍ Exchange and dissemination of MDO application information within the aerospace

community.

For more information, see the MDO TC Operations Manual and AIAA Technical Activities.




http://endo.sandia.gov/AIAA_MDOTC/faq/MDOTC_what.html12/29/2006 12:29:56 PM

http://www.aiaa.org/information/technical/index.html


1.0 INTRODUCTION

AIAA MULTIDISCIPLINARY DESIGN OPTIMIZATION

TECHNICAL COMMITTEEOPERATING PLAN

June 1998

TABLE OF CONTENTS

1. INTRODUCTION 2. PURPOSE AND ORGANIZATION 3. SUBCOMMITTEE FUNCTIONS

❍ 3.1 Communications Subcommittees ■ Education ■ Internet ■ Liaison ■ Publications ■ White Paper

❍ 3.2 Technical Subcommittees ■ Applications ■ Benchmarking ■ Conference support ■ MA&O Symposium Support

❍ 3.3 Planning ■ Awards ■ Charter ■ Membership

4. ACTIVITIES AND SCHEDULES ❍ 4.1 Committee operation and membership ❍ 4.2 TC Chair Selection Process ❍ 4.3 Officer Responsibilities ❍ 4.4 Subcommittee activities and schedules

■ Education ■ Internet ■ Liaison

http://endo.sandia.gov/AIAA_MDOTC/operations/MDOTC_man.html (1 of 18)12/29/2006 12:29:58 PM

1.0 INTRODUCTION

■ Publications ■ White Paper ■ Applications ■ Benchmarking ■ Conference support ■ MA&O Symposium Support ■ Awards ■ Charter ■ Membership

5. APPENDICES ❍ 5.1 Latest 3 year plan ❍ 5.2 Sample Liason Report ❍ 5.3 AIAA Short Course Review Procedure ❍ 5.4 TC Operations/Calander Even Years ❍ 5.5 TC Operations/Calander Odd Years ❍ 5.6 Plan for Electing New General Chair for Multidisciplinary Analysis and Optimization

Symposium

1.0 INTRODUCTION

This manual describes the activities of the AIAA Multidisciplinary Design Optimization Technical Committee in support of the broad AIAA objective of advancing the state of technology for a wide range of aerospace systems. The purpose of this manual is to describe the overall schedule and activities of the committee in order to maintain continuity of committee functions as members change from year to year. This is therefore a working document which should be reviewed yearly and updated as required to accurately reflect committee activities.

2.0 PURPOSE AND ORGANIZATION

CHARTER: "To provide an AIAA forum for those active in development, application, and teaching of a formal design methodology based on the integration of disciplinary analyses and sensitivity analyses, optimization, and artificial intelligence, applicable at all stages of the multidisciplinary design of aerospace systems".

The key mission of the technical committee as outlined in the charter is "to provide an AIAA forum..." for members of the technical community active in the study of formal design methodology based on an efficient integration of analysis and optimization methods. Such a forum currently exists in the form of technical meetings and publications of the AIAA, and can be further enhanced by input from the technical committee. In particular, the committee must ensure that the subject of these meetings and


1.0 INTRODUCTION

publications include all scientific disciplines involved in creating an effective multidisciplinary design optimization environment.

To achieve this broad mission, the technical committee is organized around a number of subcommittees with narrower, more specific charters. Some of the subcommittees are standing committees to handle the ongoing business of the technical committee and others have finite lifetimes consistent with the nature of their activities. It is the expressed intent of the technical committee that the subcommittees be active, solicit committee member involvement, and conduct the primary business of the technical committee.

3. SUBCOMMITTEES/FUNCTIONS

3.1 Communications Subcommittees

Education: To assess and make recommendations for improvement on all issues related to the teaching of multidisciplinary design optimization methods in both the university and the industry R&D environment and to develop a program for fostering a greater overall awareness of multidisciplinary design methods.

Internet: To facilitate timely communications among TC members using the Internet for electronic messages and postings. This entails managing the TC World Wide Web site, administering TC electronic mailing lists, and archiving TC operational reports and data.

Liaison: To coordinate activities and provide a channel of communication with other technical committees and outside organizations which contribute to advancing the state of technology in the design of aerospace systems.

Publications: To compile and edit information for the yearly "highlights" article, and to distribute an updated committee operating manual and roster each year.

White Paper:The white paper subcommittee is charged with initiating all MDO TC white paper efforts. The subcommittee reviews current white paper activities at the first of each year to determine if any new initiatives are needed during the coming year.

3.2 Technical Subcommitees

Applications: To provide for an exchange of information related to the application of MDO methods at a level that can be explicitly measured in vehicle mission performance, weight, or cost. Also serves as a forum for industry to relate real world needs to government and academia.

Benchmarking: To develop a set of test cases (formulations and solutions) for optimization in several MDO/Aerospace disciplines and in MDO itself, to be offered to the aerospace community as a means to measure performance of optimization methods.


1.0 INTRODUCTION

Conference support: To provide support to AIAA sponsored and co-sponsored conferences which include sessions or papers related to MDO.

MA&O Symposium Support: To provide planning and support to MA&O symposiums which are co-sponsored by AIAA and run by the MDO TC.

3.3 Planning Subcommitees

The planning subcommittee has the task of insuring the long-range continuity of the MDO TC, including developing plans for the membership transition to continue a balanced mixture of disciplines and organizations, and to stimulate and coordinate planning in the subcommittees. The planning subcommittees are responsible for choosing new TC members from among the applicants. The planning subcommittee also has responsibility for keeping the Operations Manual up to date. Finally, this subcommittee is in charge of determining when the White Paper subcommittee should be reformed in order to generate a state of the art version.

Awards: To act as a focal point in the recognition of outstanding contributions in the area of multidisciplinary design optimization.

Charter: This subcommittee reviews current charter activities at the first of each year to determine if any new initiatives are needed during the coming year. It is anticipated that the charter will be updated on an annual basis.

Membership: This subcommittee is charged with maintaining the current membership roster. The committee will issue a call for MDO TC member applications in late August of each year. This committee along with the current chair will review member applications and make recommendations to the full TC at the January TC meeting.

4.0 ACTIVITIES AND SCHEDULES

4.1 Committee membership and operation:

The TC has a nominal membership limit of 35 members. Membership is for one year (May 1 to April 30) annually renewable for three years contingent on active participation in the TC and continued membership in the AIAA. These requirements may be relaxed under unusual circumstances; for example, to insure that membership terms are staggered for a new TC or to insure continuity of leadership or other specific duties. Nominations for committee members are requested in the early fall and may be submitted by any AIAA member (including self nomination). Non-AIAA members may be nominated but must join immediately after accepting membership on the TC. The TC chairman with the help of the Planning Subcommittee makes the membership selection and notifies new members in the early spring. The TC seeks to maintain a membership which reflects: a broad background in the


1.0 INTRODUCTION

technologies which contribute to multidisciplinary design optimization, a cross section of industry, government, and academic interests, a resonable geographic distribution, and a balance of technical and management skills.

The TC chairman plays a vital role in the life of the committee. The TC chairman is responsible for running the biennial election process to ensure a smooth transition of leadership on the technical committee. Note that, while the TC chair is required to have been a member for one full year by the time the elections are conducted, the vice-chairs do not have to have prior TC experience; they must be full members, however. There is no limit as to how many times one may hold a particular position and, should your term on the TC expire during you tenure, it should normally be extended to enable you to fulfill your responsibility.

Candidates should contact the current TC Chair to clearly identify the position they are running for. In addition, the candidates need to provide a write-up (~1-2 pages) describing their vision for the TC and what they intend to accomplish in their new position. These write-ups will be posted on the web for the membership to peruse, prior to voting. While any TC member may nominate another candidate, the TC would request that the nominee be contacted prior to nomination in order to assure their availability and willingness to run for office.

The voting will be conducted electronically by the TC Chair, and the TC will need a quorum of the voting members (associate members are non-voting members) before validating the elections. TC members should give serious consideration to running for these positions.

4.2 TC Chair Selection Process

The terms for the new Chair and Vice Chair/Technical, Vice Chair/Communications run for two years begining in May and ending in April. The curent term for new officers will run from May 1998 to April 2000. The election process should be completed prior to the Fall MDO TC meeting in the year preceeding the transition. This will enable the new office to overlap with the current one for 3 consecutive meetings. Note that the position of secretary is appointed rather than elected.

Here is the timetable for election of new officers (1997):

● Call for nominations 1 July 1997 ● Nominations/background material due 30 July 1997 ● Background material posted on the Web by 3 August 1997 ● Electronic votes due 2 September 1997

Please consult the web page (http://endo.sandia.gov/AIAA_MDOTC/main.html) for responsibilities and conditions of eligibility (under Operations Information/TC Operations Slides).

4.3 Officer Responsibilities


1.0 INTRODUCTION

TC Chair

Elected for 2 years from members who have at least one year of experience with the TC, in odd years, between spring and fall meeting, comes into office on May 1, even years.

● Attend all TC meetings ● Run planning committee ● Interface with TAC and AIAA ● Run elections for chair and vice-chairs ● Select subcommittee chairs from recommendations of vice-chairs, odd years ● Select subcommittee membership on recommendation of vice-chairs, review every year between

Fall and Reno meeting, or when appropriate ● Present subcommittee reports for planning committee if chair absent

TC Vice-Chairs

Elected for 2 years, in odd years, between spring and fall meeting, comes into office on May 1, even years.

● Attend all TC meetings ● Coordinate Technical and Communications committees ● Lead subcommittees in reviewing their charters ● Stand-in for chair if absent ● Recommend subcommittee memberships ● Recommend subcommittee chairs ● Present subcommittee reports at meeting if subcommittee chair absent

TC Subcommittee Chairs

Appointed for 1 years by TC chair, every year by SDM meeting prior to entering service at Fall meeting, upon recommendation from vice-chair.

● Attend two out of three of all TC meetings, must have substitute fo the other meetings. ● Define, update subcommittee charter. ● Provide subcommittee information to webmaster. ● Run subcommittee business between meetings. ● Draft and present subcommittee written reports, submit to secretary. ● Recommend subcommittee membership to vice-chair. ● Recommend subcommittee chair upon leaving position.

Secretary


1.0 INTRODUCTION

Appointed by the chair for two years, in even years, by SDM meeting prior to entering service at Fall meeting.

● Collect subcommittee reports, submit to webmaster within 30 days of meeting ● Deliver minutes of meeting to webmaster within 30 days of meeting

Membership

Selected yearly at Reno meeting by membership subcommittee

● Membership begins on May 1 of year selected ● Membership is renewable annually, typically for up to 3 years ● Membership may be renewed annually, beyond 3 years, at chair's discretion, if member has taken

on a responsibility that extends beyond his/her tenure. ● Each member must participate actively in the activities of at least one subcommittee for

continued membership

4.4 Subcommittee activities and schedules

Subcommittee activities and schedules as they are currently available are included in this version of the manual. Others will be added as they are generated in the normal review and update process of this manual.

Education: The activities of this subcommittee are of an ongoing nature. The subcommittee must develop a plan for both increasing the general awareness of the potential advantages of a formal design methodology and for coordinating the delivery of educational programs in the general area of multidisciplinary design optimization. In order to achieve these goals:

The subcommittee plans to continue assisting in the development of requirements for various AIAA sponsored design competitions to encourage use of formal optimization tools at the senior undergraduate level. It has initiated contact with the AIAA Education committee to provide input for ABET requirements for design in the undergraduate aerospace curriculum and is currently soliciting specialized course topics in multidisciplinary design from MDO TC members.

A sequence of courses related to multidisciplinary analysis and optimization are being offered. These can be attached to AIAA or ASME meetings, arranged at central sites, or delivered at specific industry/R&D facilities.

Informational articles can be written or solicited. These articles should be targeted at an audience without formal education in optimization and be published in Aerospace America. The same material can also be organized in a general informational lecture, suited for AIAA sectional or regional meetings.


1.0 INTRODUCTION

Develop a general informational database that includes the preparation of audio visual material in the subject area, and possible publication by AIAA.

The subcommittee must play a key role in promoting the teaching of formal optimization techniques in engineering curricula. This must go beyond instruction of a traditional optimization course, and must focus to some degree on the integration of analysis and design. Considering the already stretched requirements of most undergraduate programs, this may be no easy task. The most likely place for incorporating such material is in capstone design courses at the senior level. Input and suggestions from the AIAA to the ABET accreditation board must be effectively used for this purpose.

The education subcommittee can serve as a catalyst to encourage the inclusion of formal design optimization in undergraduate curricula by seeding and providing AIAA/Industry sponsorship to projects/papers/contests directly related to the application of optimization methods. A possible approach to this could be along the lines of the current AIAA design competitions.

The subcommittee can also adopt an active role in reviewing the available literature and educational material in the subject area. Publications of reviews on new material as well as suggestions to major publishing houses on needs in specific areas can provide a very useful service to the technical community.

Internet: This subcommittee provides three primary services to the MDO TC. First, the internet subcommittee maintains the TC World Wide Web site. This involves the following operations:

● maintaining accurate committee and subcommittee membership information, white papers, conference and short course information, an operations manual, a FAQ list, and related site information.

● posting of TC meeting location and agenda, meeting minutes, subcommittee reports, and current action items.

● updating the site to reflect advances in web technology and software.

Second, the internet subcommittee administers electronic mailing lists for the TC in order to centralize management of the TC email distribution. This involves the following operations:

● maintaining software (e.g., Majordomo) for list serving. ● maintaining accurate list subscriptions by updating member addresses and trouble-shooting email

problems. ● informing TC members of list purpose and proper usage. ● creating additional email lists as required to accomplish further automation of TC operations.

Third, the internet subcommittee archives both Web site postings and mailing list traffic in order to provide a browsable reference on past and present TC issues. This involves the following operations:


1.0 INTRODUCTION

● maintaining software (e.g., LWGate) for mailing list archival. ● maintaining menus of past and present postings on the TC Web pages.

Liaison: This subcommittee works closely with the TC chair to coordinate the activities of the MDO TC with those of the other AIAA TCs or other organizations. The subcommittee chair is appointed by the TC chair and is primarily responsible for assigning liasons to other TC meetings and collecting and reporting significant activities. These liasons communicate with the other TC's via their TC chair or liason chair or by attending their meetings when possible. The liason the communicates MDO related activities back to the MDO TC via the Liaison Chair. This communication is in the format of the MDO TC Liaison Report; found in the Appendix of this manual. Liaison assignments can be found in the subcommittee membership list. Currently representatives are assigned to the Aircraft Design TC, the Applied Aero TC, the Structures TC, the Structural Dynamics TC, the CAD/CAM TC, the Guidance and Control TC, Public Policy TC, the Thermophysics TC, and to ASME and PMEC.

Publications: This subcommittee distributes the roster and operating manual to new members at the first meeting of the year or as soon after as practical. It compiles and edits information for the yearly "highlights" article at the end of each year and reviews the operating manual for possible update.

White Paper: The committee will periodically generate and review a white paper expressing the collected views of the technical committee and describing the state-of-the-art in integrated multidisciplinary design analysis and optimiztion.

Applications : This subcommittee provides for the exchange and dissemination of information related to the application of MDO methods at a level that can be explicitly measured in vehicle mission performance, sized weight, and/or cost. Emphasis will be placed on current and planned activities to integrate MDO methods in the standard design pro-active of aerospace systems. The methods themselves, as well as design experiences, overall benefits, and impediments to formal MDO methods, will be subjects to be considered by the subcommittee. The subcommittee will also serve as a forum for industry participants to relate the "real world" needs associated with MDO to the academic and government members of the MDO TC. Specific activities include:

● Promote presentation and publication of experiences with and plans for the application and utilization of MDO methods into the vehicle synthesis/design environment

● arrange invited speakers at conferences and MDO Technical Committee meetings ● arrange special sessions at conferences ● encourage publication of journal articles ● Identify and disseminate information pertaining to the benefits of formal MDO methods to

overall vehicle performance, size, range, and/or cost ● Assess and report the MDO needs of advanced (conceptual/preliminary) vehicle design or

synthesis organizations ● Define multidisciplinary metrics and figures of merit ● Examine issues related to overall vehicle synthesis that MDO can influence. E.g. shorten design


1.0 INTRODUCTION

cycle, improve product, reduce development cost, reduce manpower required for design, capture human experience, etc.

Benchmarking: This subcommittee is to develop a set of test cases (formulations and solutions) for optimization in several MDO/Aerospace disciplines and in MDO itself, to be offered to the aerospace community as a means to measure performance of optimization methods. Also, available and future capabilities in multidisciplinary design optimization are to be compared benchmarked according to a standard set of problems and performance measures.

Conference support: The MDO TC has had an evolving presence at conferences over the past several years. The current focus is on the SDM (various locations, April), Aircraft Engineering, Technology, and Operations (various locations, August), Aerospace Sciences (Reno, January), and MA&O (various locations, September bi-annually) conferences but this is expanding. Note that support of the MA&O Symposium is handled by the MA&O Conference Support subcommittee.

The activities include participating in conference planning, organizing and chairing sessions, writing calls for papers, and paper review to ensure that meetings reflect the appropriate involvement of the scientific disciplines which contribute to multidisciplinary design optimization.

The schedule for session organization and paper review for these conferences is keyed to the conference date.

MA&O Symposium Support: The MDO TC has primary control and responsibility of the MA&O Symposium which is held in even years during September at various locations. The subcommittee is responsible for planning, organizing, writing the call for papers, and choosing chairing sessions, and paper review. Papers are chosen to ensure that symposium reflects involvement multidisciplinary analysis, design, and optimization.

The schedule for the call for papers, paper review, and session organization is keyed to the conference date.

Awards: At the beginning of each year, the Awards subcommittee will survey the new members of the MDO TC and advocate the promotion of deserving members to Senior Member, Associate Fellow, and Fellow status. Members who are upgraded shall be recognized during the appropriate TC meeting. The requirements for upgrades are found below in brief and in detail at the AIAA Membership Upgrades:

● Senior Members - Persons who have demonstrated a successful professional practice in the arts, sciences, or technology of aeronautics for the equivalent of at least eight years, or the applicant shall have at least eight years of continuous professional membership. Applications are reviewed by the Membership Committee on a monthly basis. Senior Members receive a certificate and lapel pin.


http://www.aiaa.org/Membership

1.0 INTRODUCTION

● Associate Fellows - Persons who have accomplished or been in charge of important engineering or scientific work, or who have done original work of outstanding merit, or who have otherwise made outstanding contributions to the arts, sciences or technology of aeronautics or astronautics. Nominees must be Senior Members with at least 12 years of professional experience (four years of post-graduate studies may be included, if applicable). Three AIAA member references with the standing of the Associate Fellow, Fellow or Honorary Fellow are required. Nomination forms are due by April 15; references are due by May 15. Newly elected Associate Fellows receive a certificate and lapel pin. A list of newly elected Associate Fellows is published every January in Aerospace America.

● Fellows - Persons of distinction in aeronautics or astronautics, and shall have made notable valuable contributions to the arts, sciences, or technology thereof. Nominees must be of Associate Fellow status. Five references are required. Nomination forms are due by June 15; references are due by July 15. Newly elected Fellows receive a certificate and lapel pin and are honored at the Honors Night Banquet, in conjunction with the Global Air & Space Conference.

During the Reno (January) and SDM (April) names are collected for who the TC should nominate for the rank of fellow. During the SDM (April) meeting the TC votes on who, if anyone, the awards committee should nominate for the rank of fellow.

During the year the subcommittee will contribute to the selection of award recipients for existing AIAA awards based on established schedules and AIAA calls for participation. This includes the SDM award and the new AIAA award in the area of multidisciplinary optimization which is given bi-annually at the MA&O symposium. All of the awards procedures can be found in the AIAA Honors and Awards Manual.

The Multidisciplinary Design Optimization Award was established by this TC. The chairman of the awards subcommittee should remind the TC members during the fall meeting to submit nominations to AIAA for this award. The selection of the recipient of this award is made by this subcommittee as follows:

1. In October of the year proceeding the MA&O conference the call for nominations is published in Aerospace America. A copy of the call for nominations, as well as the nomination form, are mailed to each TC chairman.

2. The nominations must be submitted to AIAA headquarters by mid January (January 12 in 1996).

3. The nominations are passed on to this subcommittee in late January.

4. Members of a selection committee are chosen by this subcommittee using the following AIAA guidelines:

❍ The selection committee should be a representative sampling of the professional peer


http://www.aiaa.org/Membership/H&A_Manual_00.doc


1.0 INTRODUCTION

group(s) in the technical specialty(ies) of concern. They are not required to be current TC members, but must be members of AIAA.

❍ At least two members should be members of the TC. ❍ Former award winners are often useful and willing to be selection committee members.

They may not be able to serve as regular TC members but are often quite willing to undertake this less-time-consuming task.

❍ Whenever possible (without violating other important guidelines), the total selection committee membership should be an odd number to eliminate tie votes.

5. The selection committee must determine the award recipient and award citation by the end of February. The information is passed along to the Awards Subcommittee, which in turn, passes it along to AIAA headquarters.

6. The award recipient is notified in mid March.

7. The award is presented at the MA&O conference in mid September.

The nomination material for candidates that do not receive the award are kept for two years. These candidates are automatically considered for the next award. More details on the technical awards procedure can be found in the AIAA Honors and Awards Manual.

The MDO TC chooses the recipient of the best paper award for the MA&O symposium. AIAA has no set procedure for determining the best paper of a conference. The choice is completely up to the TC sponsoring the conference. Requirements for best paper:

1. Full length papers only.

2. The paper must have been presented at the conference.

3. The paper must be in the printed proceedings. This is so people can go back and read the best paper after it has been announced.

The selection committee wants to determine the best paper without having to read through all of the conference papers. The potential best papers were determined to be:

● The best paper in each session as determined by the session co-chairs

● The top 15 most highly rated abstracts from the conference paper selection reviews

Following is the procedure in use to determine the recipient of the best paper award. This procedure may be modified in the future. There will be about 50 papers in consideration for the best paper and 5



1.0 INTRODUCTION

reviewers.

● STEP 1: Break the papers into 5 groups of about 10 and have each reviewer review 2 groups. This way each paper is reviewed by two reviewers. Each reviewer will choose the 5 best papers out of the about 20 reviewed.

● STEP 2: There will be up to 25 papers after STEP 1. These will be broken into 5 groups of 5 and each reviewer will again review two groups. No reviewer will review a paper that he reviewed in STEP 1. This way each paper is reviewed by four reviewers. Each reviewer will choose the 2 best papers out of the up to 10 reviewed.

● STEP 3: There will be up to 10 papers after STEP 2. Each reviewer will review all of the papers and rank them 1-10. The papers will be assigned points based on their rank, i.e. 1 for the best, 2 for the second best, etc. The points for each paper for each review will be added up. The three papers with the lowest point total will go to STEP 4.

● STEP 4: The reviewers will decide which of the 3 papers from STEP 3 should receive the best paper award.

As soon as the best paper is chosen, the authors should be notified by mail that they will be receiving an award at the next MA&O conference. The selection should be made by the SDM conference following the MA&O symposium. Also, the authors of the 10 papers selected in Step 2 will be notified by mail that their papers were finalists in the best paper award selection procedure.

Draft policy for MDO TC Support of Fellow Nomination

1. The MDO TC will endorse a maximum of one nomination for AIAA Fellow Status each year.

2. The endorsement will take the form of a letter signed by the TC chair and logistical support provided by the Honors and Awards subcommittee.

3. Each year at the Reno meeting the TC will initiate the process of at most one candidate to receive its nomination. Names and bios will be solicited after the Reno meeting, and an electronic vote to select a nominee will be completed within two months.

4. Each suggested candidate will have a nominator who is responsible for writing the nomination letter and obtaining commitments from five references.

5. Candidates for the TC's endorsement must have made significant contributions to the field of MDO.

Charter: This subcommittee reviews current charter activities at the first of each year to determine if any


1.0 INTRODUCTION

new initiatives are needed during the coming year. It is anticipated that the charter will be updated on an annual basis.

Membership: This subcommittee is charged with maintaining the current membership roster. The committee will issue a call for MDO TC member applications in late August of each year. This committee along with the current chair will review member applications and make recommendations to the full TC at the January TC meeting.

5.0 APPENDICES

5.1 Latest 3 year plan (to be added when available)

5.2 Sample Liaison Report

MDO TC Liaison Report

Submitted by:

TC:

TC Chairperson:

Contact for TC meeting minutes:

Their Liaison to MDO TC:

Last TC meeting date:

Last TC meeting location:

Last TC meeting attendance:

Items of interest to MDO TC:

5.3 AIAA Short Course Review Procedure

Proposals for AIAA Short Courses which seek support or sponsorship by the AIAA MDO TC should be sent to the AIAA MDO TC Education Subcommittee Chair. Proposals must include a syllabus with identification of the courses MDO content, a list of speakers and their resumes, with emphasis on prior work and teaching experience. This syllabus should include an hourly breakdown of topics where applicable. The education subcommittee chair will arrange for up to three AIAA MDO TC members


1.0 INTRODUCTION

(not necessarily education subcommittee members) to review the Short Course proposal. Reviewers will provide a brief assessment of the short course proposal and will recommend that the AIAA MDO TC accept or decline sponsorship of the short course. Reviewers will be asked to pay particular attention to the MDO content of any short course proposal. In order that a short course be sponsored by the AIAA MDO TC there must be some significant content which addresses issues pertinent to MDO. For example; numerical optimization, system decomposition and synthesis, multidisciplinary design methodologies and strategies, etc.. Once these reviews are completed they will be distributed to the Education Subcommittee Members via electronic mail. The Education Subcommittee will vote as a committee (via email) to accept or decline sponsorship of the short course. The Education Subcommittee's recommendation will be passed onto MDO-TC chair and executive committee. It is expected that proposals submitted via this procedure will be acted on and that the person submitting the proposal will be notified of acceptance or declination within two months from the date of submission.

The Education Subcommittee Chair may elect to return the initial reviews to the person submitting the proposal for revision of the short course proposal prior to forwarding the reviews to the Education Subcommittee. Any revisions would need to be reviewed prior to moving the proposal forward to the chair and executive committee. After the short course is held, AIAA will return reviews to the MDO-TC and Education Subcommitee for review

5.4 TC Operations/Calander Even Years

Januarywinter meeting, ASM/Renoselect new membersbegin Fellow programannual report due group directornew roster due AIAA

Februaryselect subcommittee chairsselect SDM representative (follow. year)

March

Aprilspring meeting, SDMbegin selection of next MA&O team

Maymembership year beginsfellow nomination(s) due


1.0 INTRODUCTION

Junefellow recommendation(s) due

July

August

Septemberfall meeting/MA&O conference (all-day)new office first meetingMDO award ceremonyselect subcommittee membershipbegin best MDO paper selection

October

November

Decemberselect ASM representative (follow. year)

5.5 TC Operations/Calander Odd Years

Januarywinter meeting, ASM/Renoselect new membersbegin Fellow programannual report due group directornew roster due AIAA

Februaryselect subcommittee chairsselect SDM representative (follow. year)

March

Aprilspring meeting, SDM

Maymembership year beginsfellow nomination(s) due


1.0 INTRODUCTION

Junefellow recommendation(s) due

Julybegin office election process

August

Septemberfall meeting/location tbdannounce new office new office first meetingselect subcommittee membership

October

November

Decemberselect ASM representative (follow. year)

5.6 Plan for Electing New General Chair for Multidisciplinary Analysis and Optimization Symposium

Process for Identifying New General Chair:

In August or September of the year preceding the biennial Multidisciplinary Analysis and Optimization Symposium, the TC Chair will make an announcement that the TC will be accepting nominations from individuals who wish to serve as General Chair for the next (three years out) MAO Symposium (this announcement should be made to all present and past members of the TC, as well as to ISSMO members). Past and present TC Members can nominate themselves or be nominated by others. However, if the nomination is not a self-nomination, there must be an assurance that the individual being nominated wants to serve in this capacity.

Nominations must be written nominations which outline the resources that the individual can commit to the task and any experience/background which might make him/her appropriate for handling the outlined responsibilities below. The nominee can also comment on what procedure he/she would use to identify a Technical Chair.

Prior to the fall TC meeting, the TC membership will review nominations. At the fall TC meeting a vote of the TC will elect the next General Chair for the MA&O Conference (three years out). The next


1.0 INTRODUCTION

General Chair is encouraged to select a Technical Chair before the ASM meeting in Reno. The election three years prior to the MA&O conference will provide the Technical Chair with greater flexibility in coordinating the site selection process with AIAA and the TC. The new chairs can be introduced during the banquet at the MA&O Symposium the following September. A briefing meeting of the new chairs and present chairs will then take place at the MA&O Symposium.

Responsibilities of General Chair:

One of the first responsibilities of the General Chair will be to choose an appropriate Technical Chair for the conference. The General Chair works with AIAA to identify the conference site (this typically entails traveling to at least one and perhaps more potential sites, funding for which must come from the General Chair or his company), plans all 'extra-curricular' activities (e.g. banquets, dinners, etc.) and makes arrangements for plenary and banquet speakers. Other responsibilities would be worked out with the Technical Chair and often includes such things as preparing or assisting in preparing the Call for Papers and any other literature associated with the conference, and assisting with the organization of the sessions (which would again involve travel if the Technical Chair is geographically located elsewhere).

Responsibilities of Technical Chair:

The Technical Chair interfaces with AIAA and the General Chair on all technical matters pertaining to the conference. This includes preparing literature mailings (i.e. Call for Papers, etc.), receiving all abstracts, arranging for review of abstracts, determining acceptance or rejection of abstracts (often with assistance from the General Chair and/or Organizing Committee), organizing special panel or technical sessions, etc. The last two conferences have used Superchairs who are responsible for arranging abstract reviews within topic areas. The Technical Chair is responsible for identifying and interfacing with Superchairs, if this organizational structure is used. If used, volunteers for Superchair duty will be accepted at the MDO TC meetings in the year prior to abstract submission.


Last Updated: 13 December 2001




MDO TC Subcommittee Membership


The AIAA MDO TC has two types of subcommittees: those which support AIAA services and activities (such as conferences and awards), and those which address particular technical topics of interest to TC members and the MDO community at large.

TC members are invited (new members especially) to review the enclosed subcommittee descriptions and contact the subcommittee Chairman to volunteer for service on any particular subcommittee. Subcommittees meet at the discretion of their Chairs, who report any activity at the regular TC meetings. Each Chairman should also review the subcommittee membership for completeness and correctness, and notify the TC Chair with changes of subcommittee membership.

Also, if you feel that any area of important potential TC activity is not adequately covered by the existing subcommittee structure, please feel free to propose new subcommittees to the TC Chairman (of course, you will probably be asked to head the new subcommittee!).

Subcommittees

● Applications ● Awards ● Education ● Publications

Applications

Purpose: To provide for an exchange and dissemination of information related to the application of MDO methods at a level that can be explicitly measured in vehicle mission performance, weight, or cost. Also serves as a forum for industry to relate real world needs to government and academia.

Chair: Tim Purcell [email protected]

Members:

● Stephen Batill [email protected]

http://endo.sandia.gov/AIAA_MDOTC/membership/MDOTC_sub.html (1 of 4)12/29/2006 12:29:59 PM




● Elias Bounajem [email protected] ● Evin Cramer [email protected] ● Guru Guruswamy [email protected] ● Ram Krishnamachari [email protected] ● Achille Messac [email protected] ● Vassili Toropov [email protected] ● Brett Wujek [email protected]

Awards

Purpose: To act as a focal point for recognition of outstanding contributions in the field of multidisciplinary optimization. The Awards Subcommittee was instrumental in establishing the biennial MDO Award, and on an ongoing basis is principal in the selection of candidates for this award, nominations and selection for other AIAA awards such as the SDM, and in nomination of AIAA Associate Fellows for upgrade to Fellow membership status.

Chair: Lt. Col. Robert Canfield [email protected]

Members:

● Max Blair [email protected] ● Srinivas Kodiyalam [email protected] ● Achille Messac [email protected] ● Somanath Nagendra [email protected] ● Hans Schweiger [email protected] ● Afzal Suleman [email protected] ● Karen Wilcox [email protected] ● Scott Zink [email protected]

AIAA Information: AIAA Honors and Awards Manual.

Education

Purpose: To coordinate activities related to education in multidisciplinary design optimization, both within AIAA and related activities outside of AIAA. These activities include coordinating and


















http://www.aiaa.org/Membership/index.hfm?mem=4


sponsoring Short Courses, soliciting or writing informational articles about MDO activities for the education community, promoting teaching of formal optimization and MDO techniques in engineering curricula, supporting AIAA sponsorship of student projects and competitions, reviewing the available literature and educational material, etc.

Chair: Kemper Lewis [email protected]

Members:

● Kurt Anderson [email protected] ● Oktay Baysal [email protected] ● Wei Chen [email protected] ● Bernie Grossman [email protected] ● Zafer Gurdal [email protected] ● Doug Smith [email protected] ● Afzal Suleman [email protected]

Publications

Purpose: Oversee publication activities of the MDO TC. These have included a White Paper (published as an AIAA Report), Newsletter (now maintained at this Web site), and the annual Aerospace America Highlights article.

Chair: Fred Striz [email protected]

Members:

● Vladimir Balabanov [email protected] ● Dan DeLaurentis [email protected] ● Tony Giunta [email protected] ● Narendra S. Khot [email protected] ● James Reuther [email protected] ● Ravindra V. Tappeta [email protected]

Last Updated: 16 January 2002





















How to join the AIAA MDO TC

How to join the AIAA Multidisciplinary Design Optimization Technical Committee

You may nominate yourself or someone else for membership on a technical committee. For each nominee you must submit a nomination form along with a resume or biographical data. If you nominate someone for more than one committee, list each technical committee (TC) on one form. If you are invited to be on more than on TC, you must select one committee for membership, (as you can only be on one TC at a time). If you are appointed to a TC and you are not a member of AIAA, you must join AIAA within one year of your appointment. If you are not appointed, your nomination will automatically be considered for the following year. TC membership is for one year, with two additional years possible. Committee members are automatically considered for a second and third year of membership and do not have to submit new forms. Deadline for receipt of nominations is November 1. Download the nomination form from the AIAA Web Site and then mail it to AIAA Technical Committee Nominations, 1801 Alexander Bell Drive, Reston, VA 20191, or FAX it to 703/264-7551.

For more information, see Nomination Requirements and Nomination Form Instructions.


Last Updated: 17 October 2002


http://endo.sandia.gov/AIAA_MDOTC/faq/MDOTC_how.html12/29/2006 12:30:00 PM

http://www.aiaa.org/forms/tcnom.pdf

http://www.aiaa.org/about/index.hfm?abo=530

http://www.aiaa.org/about/index.hfm?abo=540


Operations Slides

Operations Slides

http://endo.sandia.gov/AIAA_MDOTC/operations/operations.html (1 of 7)12/29/2006 12:30:02 PM

Operations Slides


Operations Slides


Operations Slides


Operations Slides


Operations Slides


Operations Slides







As approved at 18 Jan 1996 meeting, AIAA/ASM, Reno, NV

TC Operations/Proposed TC Structure

OFFICECHAIR, VICE-CHAIRS

PAST-CHAIR, SECRETARY

APPLICATIONS BENCHMARKING CONFERENCE SUPPORT MA&O SYMPOSIUM

PLANNINGTC CHAIR

TECHNICALVICE-CHAIR/TEC.

COMMUNICATIONSVICE-CHAIR/COM.

AWARDS CHARTER[MEMBERSHIP][POLICY]

EDUCATION INTERNETLIAISON NEWSLETTER PUBLICATIONS[WHITE PAPER]

[] indicate subcommittees that are not in-place at this time


TC Operations /TC Chair• Elected for 2 years from members who have at least one year of

experience with the TC, in even years by SDM meeting prior to entering service at Fall meeting.

• Attend all TC meetings• Run planning committee• Interface with TAC and AIAA• Run elections for chair, even years (next will be before SDM1998)• Run elections for vice-chairs, even years, at the same time as the above

(next will be before SDM 1996)• Select subcommittee chairs from recommendations of vice-chairs, odd

years• Select subcommittee membership on recommendation of vice-chairs,

review every year between Fall and Reno meeting, or when appropriate• Present subcommittee reports for planning committee if chair absent


TC Operations/TC Vice-Chairs

• Elected for 2 years, in even years by SDM meeting, prior to entering service at Fall meeting

• Attend all TC meetings• Coordinate Technical and Communications committees• Lead subcommittees in reviewing their charters

• Stand-in for chair if absent• Recommend subcommittee memberships• Recommend subcommittee chairs• Present subcommittee reports at meeting if subcommittee chair absent


TC Operations/Subcommittee Chair

• Appointed for 1 years by TC chair, every year by SDM meeting prior to entering service at Fall meeting, upon recommendation from vice-chair.

• Attend two out of three of all TC meetings, must have substitute fo the other meetings.

• Define, update subcommittee charter.

• Provide subcommittee information to webmaster.

• Run subcommittee business between meetings.• Draft and present subcommittee written reports, submit to

secretary.

• Recommend subcommittee membership to vice-chair.

• Recommend subcommittee chair upon leaving position.


TC Operations/Secretary

• Appointed by the chair for two years, in even years, by SDM meeting prior to entering service at Fall meeting

• Collect subcommittee reports, submit to webmaster within 30 days of meeting

• Deliver minutes of meeting to webmaster within 30 days of meeting


TC Operations/Membership

• Selected yearly at Reno meeting by membership subcommittee• Membership is renewable, typically for up to 3 years

• Each member must participate actively in the activities of at least one subcommittee for continued membership


TC Operations/Calendar

JanFeb

MarApr

MayJun

JulAug

SepOct

NovDec

Even Year

Reno meetin

g

membership selection

SDM meetin

g

MA&O conference

Chair, vice-chair,

secretary,

subcommitte

e chair new

member f

irst m

eeting

Subcommittee m

embership selection

Chair/vice-chair e

lection

secretary, subcommitte

e

chair selectio

n

JanFeb

MarApr

MayJun

JulAug

SepOct

NovDec

Odd Year

Reno meetin

g

membership selection

SDM meetin

g

AeTOC conference

subcommittee chair,

new m

ember firs

t meetin

g

Subcommittee m

embership selection

Subcommittee chair

selection

Documents

Aiaa Mdo Tc 1998 White Paper on Mdo