ANSYS Advantage Volume 3 Issue 1

ADVAN T AG EE X C E L L E N C E I N E N G I N E E R I N G S I M U L A T I O N

V O L U M E I I I I S S U E 1 2 0 0 9

TM

FRAMEWORKPAGE 6

MULTIPHYSICSPAGE 11

HIGH-PERFORMANCECOMPUTINGPAGE 25

SPOTLIGHT ON ANSYS 12.0

ANSYS Advantage • Volume III, Issue 1, 2009www.ansys.com 1

EDITORIAL

this simulation method has the potential to reduce crackgrowth analysis time by over 90 percent compared with manual methods. The productivity gain will enable engineers to analyze more designs annually, thus keepingup with increased demand for turbochargers around theworld and strengthening the company’s leadership position in this competitive industry sector.

The prediction method is based on improved fracturemechanics capabilities for calculating J integrals, one ofthe many enhancements in ANSYS 12.0. Previewed in theSpotlight section of this issue, the release is a milestonefor the software supplier and a huge step forward for theCAE industry in terms of advancements in individualphysics (structural, fluid, thermal and electromagnetics)and integration of this functionality into a unified multi-physics framework for Simulation Driven ProductDevelopment — an approach leading to top-line revenuegrowth and bottom-line savings for many companies.

Discussion of the business value of simulation is particularly relevant in today’s world as manufacturersface the toughest economic climate of a lifetime. Indeed,with their survival at stake, forward-thinking companiesrecognize the need to invest in engineering simulation nowmore than ever to withstand the current market turbulenceand to strengthen their long-term competitive position,brand value and profitability as conditions improve in thecoming years. ■

John Krouse, Senior Editor and Industry Analyst

Engineering Simulation: Needed Now More Than EverIn a tough economy, forward-thinking companiesare investing in leading-edge simulation technology to drive top-line revenue growth and bottom-line savings.

Time and cost benefits of engineering simulation arewell documented. Predicting product performance anddetermining optimal solutions early in the design phase helpto avoid late-stage problems and to eliminate trial-and-errortesting cycles that drive up costs and bog down schedules.Simulation enables engineers to perform what-if studiesand to compare alternatives, processes that otherwisewould be impractical. Indeed, bottom-line savings are one key benefit that prompts most companies to implementsimulation, and are most readily quantified in return-on-investment calculations.

A second, and potentially greater, benefit is boostingtop-line revenue growth. With simulation, companies candevelop innovative, winning products that stand apart fromothers, make the status quo obsolete or create entirely newmarket opportunities. Brand value can be enhanced by tuning product performance to specific performancecharacteristics. Revenue streams may be expanded byincreasing design throughput of new products or tacklingprojects that otherwise would not be attempted.

How specific companies leverage simulation inachieving these benefits depends on their unique products,engineering challenges and business requirements. Thepossibilities are limitless. Case in point is detailed in thisissue’s article “Predicting 3-D Fatigue Cracks without aCrystal Ball” from Honeywell Turbo Technologies. Engineersused software from ANSYS to predict thermomechanicalfatigue cracks in turbochargers for internal combustionengines. Predicting crack failures early enables engineers tooptimize designs upfront and helps to avoid qualificationtest failures that lead to additional rounds of tests — whichcan be very expensive and take weeks to complete. Further,

© iS

tock

phot

o.co

m/A

trapa

t

15

22

www.ansys.comANSYS Advantage • Volume III, Issue 1, 20092

CONTENTS

Table of ContentsSPOTLIGHT ON ANSYS 12.0

4 ANSYS 12.0Launching a New Era of Smart Engineering SimulationA full generation ahead of other solutions, ANSYS 12.0 takes product design and development to the next level.

6 FRAMEWORK

Introducing ANSYS Workbench 2.0Proven simulation technology is delivered in a truly innovative integration framework.

8 GEOMETRY AND MESHING

Taking Shape in 12.0ANSYS combines depth of simulation with industry experience to providegeometry and meshing tools that realize simulation results faster.

11 MULTIPHYSICSMultiphysics for the Real WorldIn ANSYS 12.0, multiphysics capabilities continue to increase in flexibility,application and ease of use.

14 ELECTROMAGNETICSANSYS Emag 12.0 Generates SolutionsImproved accuracy, speed and platform integration advance the capabilitiesof low-frequency electromagnetic simulation.

15 FLUIDSA Flood of Fluids DevelopmentsA new integrated environment and technology enhancements make fluids simulation faster, more intuitive and more accurate.

18 STRUCTURAL MECHANICSDesigning with StructureAdvancements in structural mechanics allow more efficient and higher-fidelity modeling of complex structural phenomena.

22 EXPLICIT DYNAMICSExplicit Dynamics Goes MainstreamANSYS 12.0 brings native explicit dynamics to ANSYS Workbench and provides the easiest explicit software for nonlinear dynamics.

23 EIGENSOLVERIntroducing the Supernode EigensolverA new eigensolver in ANSYS 12.0 determines large numbers of natural frequency modes more quickly and efficiently than conventional methods.

25 HIGH-PERFORMANCE COMPUTINGThe Need for SpeedFrom desktop to supercomputer, high-performance computing with ANSYS 12.0 continues to race ahead.

28 FUTURE DIRECTIONSFoundations for the FutureThe many advanced features of ANSYS 12.0 were designed to solve today’schallenging engineering problems and to deliver a platform for tomorrow’ssimulation technology.

11

8

4


CONTENTS

3

SIMULATION @ WORK31 AUTOMOTIVE

Predicting 3-D Fatigue Cracks without a Crystal BallANSYS tools quickly predict 3-D thermomechanical fatigue cracking in turbocharger components.

33 HEALTHCARE

Electromagnetics in MedicineElectromagnetic and thermal simulations find use in medicalapplications.

36 ELECTRONICS

Keeping Cool in the FieldA communications systems company gains millions of dollars byusing thermal simulation to bring tactical radios to market faster.

38 BUILT ENVIRONMENT

Designing Against the WindSimulation helps develop screen enclosures that can betterwithstand hurricane-force winds.

40 ENVIRONMENT

Stabilizing Nuclear WasteFluid simulation solidifies its role in the radioactive waste treatment process.

42 OPTIMIZATION

Topology Optimization and Casting: A Perfect CombinationUsing topology optimization and structural simulation helps a casting company develop better products faster.

44 MARINE

Fighting Fire with SimulationThe U.K. Ministry of Defence uses engineering simulation to find alternatives to ozone-depleting substances for fire suppression.

DEPARTMENTS47 TIPS AND TRICKS

Reusing Legacy MeshesANSYS tools enable users to work with finite element modelsin various formats for performing simulations as well as making changes to part geometry.

49 ACADEMIC

Expanding Stent KnowledgeSimulation provides the medical industry with a closer look at stent procedures.

38

49

For ANSYS, Inc. sales information, call 1.866.267.9724, or visit www.ansys.com.For address changes, contact [email protected] subscribe to ANSYS Advantage, go to www.ansys.com/subscribe.

Executive EditorFran Hensler

Managing EditorChris Reeves

Senior Editor andIndustry AnalystJohn Krouse

EditorsErik FergusonShane MoeykensMark Ravenstahl

ContributorsSusan WheelerMarty Mundy

Ad Sales ManagerHelen Renshaw

Editorial AdvisorKelly Wall

DesignerMiller Creative Group

ANSYS Advantage is published for ANSYS, Inc. customers, partners and others interested in the field of design and analysis applications.

Neither ANSYS, Inc. nor the senior editor nor Miller Creative Group guarantees or warrants accuracy or completeness of the material contained in this publication.

ANSYS, ANSYS Workbench, Ansoft Designer, CFX, AUTODYN, FLUENT, GAMBIT, POLYFLOW, Airpak, DesignSpace, FIDAP, Flotran, Iceboard, Icechip, Icemax, Icepak,FloWizard, FLOWLAB, G/Turbo, MixSim, Nexxim, Q3D Extractor, Maxwell, Simplorer, Mechanical, Professional, Structural, DesignModeler, TGrid, AI*Environment, ASAS,AQWA, AutoReaGas, Blademodeler, DesignXplorer, Drop Test, ED, Engineering Knowledge Manager, Emag, Fatigue, Icepro, Icewave, Mesh Morpher, ParaMesh, TAS,TASSTRESS, TASFET, TurboGrid, Vista, VT Accelerator, CADOE, CoolSim, SIwave, Turbo Package Analyzer, RMxprt, PExprt, HFSS, Full-Wave SPICE, Simulation DrivenProduct Development, Smart Engineering Simulation and any and all ANSYS, Inc. brand, product, service, and feature names, logos and slogans are registered trademarksor trademarks of ANSYS, Inc. or its subsidiaries located in the United States or other countries. ICEM CFD is a trademark licensed by ANSYS, Inc. All other brand, product,service and feature names or trademarks are the property of their respective owners.

© 2009 ANSYS, Inc. All rights reserved.

About the CoverANSYS introduces release 12.0,the next-generation technology for Simulation Driven ProductDevelopment. The spotlight begins on page 4.

Circulation ManagerSharon Everts


ANSYS 12.0:Launching a New Era of Smart Engineering SimulationA full generation ahead of other solutions, ANSYS 12.0 takes product design and development to the next level. By Jim Cashman, President and CEO, ANSYS, Inc.

The current economic climate has completely changedthe way most companies view engineering simulation.Leveraging the power of virtual prototyping to compress theproduct development process and drive down costs is nolonger a choice — it’s a requirement for survival in anincreasingly competitive environment.

In nearly every industry, driving product developmentthrough engineering simulation technology has become akey strategy to develop more innovative products, reducedevelopment and manufacturing costs, and accelerate timeto market.

Backed by the unmatched power of ANSYS 12.0 software, progressive companies are taking engineering simulation a step beyond. They have already realized the enormous strategic benefits of virtual prototyping — andare now seeking more from their investments in simulation.ANSYS 12.0 enables these forward-looking companies to maximize the efficiency of their simulation processes, toincrease the accuracy of their virtual prototypes, and to capture and reuse their simulation processes and data. Thisnext level of performance signals a new era of Smart Engineering Simulation, in which product innovations can berealized more rapidly, and more cost effectively, than ever before.

There is no company better qualified to launch this newera. ANSYS has led the engineering simulation industry for nearly 40 years, revolutionizing the field of engineering

simulation in much the same way that the internet and desktoppublishing have revolutionized the broadband distribution of information. As a direct consequence of a long-standing commitment to simulation, ANSYS is the only company offering advanced simulation technologies that span all keyengineering disciplines — and bringing them together in anintegrated and flexible software platform designed specificallyto support Simulation Driven Product Development.

Over the years ANSYS has made significant technologyinvestments, acquisitions and partnership to ensure continuingleadership. We recognize that every technology breakthroughor market accomplishment has only been a stepping stone toour vision. Reflecting these investments — as well as theacquired wisdom of four decades in this industry — ANSYS12.0 represents the fullest expression of our leadership posi-tion. It is the most comprehensive engineering simulationsolution available today.

While the following pages offer a wealth of detail, I’d like tofocus on the high-level benefits that our customers will realizeas they leverage the full depth and breadth of ANSYS 12.0 tomake product development smarter, better, faster and morecollaborative than they ever thought possible.

Smart Technologies = Smart SimulationAt ANSYS, we have applied our long history of tech-

nology leadership to create the world’s smartest solution forengineering simulation — more automated, repeatable,

444 www.ansys.comANSYS Advantage • Volume III, Issue 1, 2009444

Some images courtesy FluidDA nv, Forschungszentrum Jülich GmbH,Heat Transfer Research, Inc., Riello SPA and © iStockphoto.com/iLexx.


persistent and intuitive than existing products. The ground-breaking ANSYS Workbench 2.0 platform is a flexibleenvironment that allows engineers to easily set up, visualizeand manage their simulations. ANSYS 12.0 offersunequalled technical breadth that allows customers toexplore a complete range of dynamic behavior, from frequency response to large overall motion of nonlinear flexible multibody systems. ANSYS has also leveraged its industry-leading capabilities to create an unequalleddepth of simulation physics, including the newly integratedANSYS FLUENT solver, advancements in all key simulationphysics, and enabling technologies for meshing, geometryand design optimization. ANSYS Engineering Knowledge Manager allows engineers to easily archive, search, retrieveand report their simulation data via a local machine or a centralized data repository. Not only does ANSYS 12.0 represent the smartest and best individual technologies, butit brings them together in a customized, scalable solutionthat meets the highly specific needs of every engineeringteam. Powerful and flexible, ANSYS 12.0 can be configuredfor advanced or professional users, deployed to a single useror enterprise, and executed on laptops or massively parallelcomputer clusters. As customer requirements grow andmature, ANSYS 12.0 is engineered to scale up accordingly.

Better Prototypes, Better Products With its unique multiphysics, high-performance

computing and complete system modeling capabilities,ANSYS 12.0 is a complete solution that takes virtual proto-typing to a new level of accuracy, realism and efficiency.ANSYS 12.0 captures the response of a completelyassembled system and assesses how a range of highly complex, real-world physical phenomena will affect not onlyindividual components but also their interactions with oneanother. Flaws in product functionality can be recognizedbefore investments are made in full-blown physical proto-types — and ideas that are validated in the virtual world canbe fast-tracked to maximize agility and capture emergingmarket opportunities. Powered by fast and accurate solvers,design optimization with ANSYS 12.0 results in prototypeswith a much higher probability of ultimate market success.

Product Design at Warp Speed ANSYS 12.0 automates many manual and tedious tasks

involved in simulation, reducing design and analysis cyclesby days or even weeks. An innovative project managementsystem allows custom simulation workflows to be created,

captured and automated with drag-and-drop ease. ANSYS12.0 amplifies the capabilities and outputs of every memberof the engineering staff, enabling them to work smarter, to intelligently make design trade-offs and to rapidly converge on the best designs. And, because ANSYS 12.0 isbased on the most advanced technology and physics,design and engineering teams can commit to manufacturingoperations with confidence — and without investing timeand money in exhaustive physical testing.

Redefining CollaborationReal-world simulation projects often involve a wide

variety of engineering personnel — and generate large volumes of data that must be shared across the enterprise.With its broad support of simulation disciplines and nativeproject management system, ANSYS 12.0 allowsengineering teams to collaborate more freely, without software barriers or other technology obstacles. Within a single project, several engineers can assess their designswithin individual disciplines, as well as easily coordinatemultiphysics simulations. The single-project environ-ment reduces redundancies and synchronization errorsamong different engineering teams. ANSYS EngineeringKnowledge Manager also provides the tools to manage the workflow of a group of engineers and a myriad of simulation projects.

At ANSYS, we have always believed that engineeringsimulation is a sound investment — and today, it is emergingas one of the smartest investments an organization canmake. We understand the incredible time and cost pressuresunder which our customers operate today, and ANSYS 12.0is specifically designed to help them meet these challenges.

In the new era of Smart Engineering Simulation heraldedby ANSYS 12.0, product development teams can workfaster and more effectively than ever before — with a greaterdegree of confidence in their finished products. Because itprovides a tremendous opportunity for engineers to designhigher-quality, more innovative products that are manu-factured faster, and at a lower cost, ANSYS 12.0 makes themost compelling case yet for engineering simulation as apowerful competitive strategy. But we are far from finished:ANSYS 12.0 is a milestone, not the destination, as we continually work to put our tools in the hands of every engineer who can benefit from them. As the power ofANSYS 12.0 is unleashed by imaginative engineering teamsaround the world, I look forward to the amazing productinnovations that will result. ■

555ANSYS Advantage • Volume III, Issue 1, 2009www.ansys.com 555

www.ansys.comANSYS Advantage • Volume III, Issue 1, 2009666 www.ansys.comwww.ansys.com66

12.0: FRAMEWORK

Introducing ANSYS Workbench 2.0Proven simulation technology is delivered in a truly innovative integration framework.

ANSYS 12.0 delivers innovative,dramatic simulation technologyadvances in every major physics discipline, along with improvements incomputing speed and enhancementsto enabling technologies such asgeometry handling, meshing andpost-processing. These advancementsalone represent a major step ahead on the path forward in Simulation Driven Product Development. ButANSYS has reached even further bydelivering all this technology in an innovative simulation framework,ANSYS Workbench 2.0.

The ANSYS Workbench environ-ment is the glue that binds thesimulation process; this has notchanged with version 2.0. In the originalANSYS Workbench, the user interactedwith the analysis as a whole using theplatform’s project page: launching thevarious applications and tracking theresulting files employed in the processof creating an analysis. Tight integrationbetween the component applicationsyielded unprecedented ease of use forsetup and solution of even complexmultiphysics simulations.

In ANSYS 12.0, while the coreapplications may seem familiar, theyare bound together via the innovativeproject page that introduces the concept of the project schematic.This expands on the project pageconcept. Rather than offer a simplelist of files, the project schematicpresents a comprehensive view ofthe entire analysis project in flow-chart form in which explicit datarelationships are readily apparent.

Building and interacting with theseflowcharts is straightforward. A toolboxcontains a selection of systems thatform the building blocks of the project.To perform a typical simulation, such

as static structural analysis, the userlocates the appropriate analysis system in the toolbox and, using drag-and-drop, introduces it into the projectschematic. That individual system con-sists of multiple cells, each of whichrepresents a particular phase or stepin the analysis. Working through thesystem from the top down, the usercompletes the analysis, starting with aparametric connection to the originalCAD geometry and continuing throughto post-processing of the analysisresult. As each step is completed,progress is shown clearly at the projectlevel. (A green check mark in a cell indi-cates that an analysis step has beencompleted.)

Passing files and data from oneapplication to the next is managedentirely by the framework, and data and state dependencies are directlyrepresented. More-complex analysescan be constructed by joining multiplesystems. The user simply drags a newsystem from the toolbox and drops it onto the existing system in the

schematic. Connections are createdautomatically and data is transferredbehind the scenes, delivering drag-and-drop multiphysics with unprecedentedease of use.

The ANSYS Workbench environ-ment tracks dependencies among thevarious types of data in the project. Ifsomething changes in an upstreamcell, the project schematic shows thatdownstream cells need to be updatedto reflect these changes. A project-level update mechanism allows thesechanges to be propagated through alldependent cells and downstream systems in batch mode, dramaticallyreducing the effort required to repeatvariations on a previously completedanalysis.

Parameters are managed at theproject level, where it is possible tochange CAD and geometry parameters,material properties and boundary condition values. Multiple parametriccases can be defined in advanceand managed as a set of designpoints, summarized in tabular form

The toolbox, at left, contains systems that form a project’s building blocks. In this single-physics example, the user dragsthe system (from left) into the project schematic (at right), then sets up and solves the system, working from the topdown through the cells in the system. As shown, the Fluid Flow system (at right) is complete through mesh generation,as shown by green check marks.


12.0: FRAMEWORK

on the ANSYS Workbench project page.Design Exploration systems can beconnected to these same project-levelparameters to drive automated designinvestigations, such as Design of Experi-ments, goal-driven optimization or Designfor Six Sigma.

In addition to serving as a frameworkfor the integration of existing applications,the ANSYS Workbench 2.0 platform alsoserves as an application developmentframework and will ultimately provide project-wide scripting, reporting, a userinterface (UI) toolkit and standard datainterfaces. These capabilities will emergeover this and subsequent releases. AtANSYS 12.0, Engineering Data andANSYS DesignXplorer are no longerindependent applications: They have beenre-engineered using the UI toolkit and integrated within the ANSYS Workbenchproject window.

Beyond managing individual simu-lation projects, ANSYS Workbenchinterfaces with the ANSYS EngineeringKnowledge Manager (EKM) product for simulation process and data management. At ANSYS 12.0, ANSYSWorkbench includes the single-userconfiguration of ANSYS EKM, calledANSYS EKM Desktop. (See sidebar.)

ANSYS Workbench 2.0 represents asizable step forward in engineering simu-lation. Within this innovative softwareframework, analysts can leverage a complete range of proven simulationtechnology, including common tools forCAD integration, geometry repair andmeshing. A novel project schematic concept guides users through complexanalyses, illustrating explicit data relationships and capturing the processfor automating subsequent analyses.Meanwhile, its parametric and persistentmodeling environment in conjunctionwith integral tools for design optimizationand statistical studies enable engineersto arrive at the best design faster. Looking beyond ANSYS 12.0, theANSYS Workbench platform will befurther refined: The aim is to deliver acomprehensive set of simulation tech-nology in an open, adaptive softwarearchitecture that allows for pervasivecustomization and the integration ofthird-party applications. ■

Judd Kaiser, Shantanu Bhide, Scott Gilmore and ToddMcDevitt of ANSYS, Inc. contributed to this article.

Managing Simulation Data With the ever-increasing use of simulation, keeping track of the

expanding volume of simulation data becomes more and more difficult.The need to be able to quickly locate information for reuse is paramount toincreasing productivity and reducing development costs.

ANSYS EKM Desktop is a new tool, integrated in the ANSYS Workbench environment, that facilitates managing simulation data frommultiple projects. ANSYS EKM Desktop is a single-user configuration of EKM that allows users to add files from any project to a local virtualrepository. Simulation properties and other metadata are automaticallyextracted (or created) from files when added, and users can tag files withunique identifiers at any time. These attributes can all be used to searchand retrieve files based on keywords or complex search criteria. Reportscan be easily generated to allow efficient side-by-side comparison of theattributes of related analyses. Search queries and reports can be saved forlater re-use. Files that are retrieved can be directly launched in their associ-ated simulation application from within the ANSYS EKM Desktop tool.

More-complex analyses involving multiple physics can be built up by connecting systems. Data dependencies are indicated clearly as connections. State icons at the right of each cell indicate whether cells are up to date, require userinput or need to be updated — for example, whether they are just meshed or fully solved.

Two analyses from the schematics shown in the previous figure are shown here in the mechanical simulation application.Launched from the schematic, individual applications may be familiar to existing users.


12.0: GEOMETRY AND MESHING

Taking Shape in 12.0ANSYS combines depth of simulationwith industry experience to providegeometry and meshing tools that realizesimulation results faster.

Engineering simulation softwareusers have been known to spend up to90 percent of their simulation-relatedtime working on pre-processing tasks.By targeting developments in capabilitiesto increase ease of use, simplifying pre-processing tasks, and increasing thecapabilities of pre-processing tools,ANSYS has systematically deliveredexciting advances to increase the efficiency of simulation.

ANSYS has combined rich geometry and meshing techniques with its depth of knowledge andexperience, and the end result isproducts capable of harnessing integrated geometry and meshingsolutions that share core librarieswith other applications. At releases10.0 and 11.0, ANSYS introducedrobust, new meshing capabilitiesfrom ANSYS ICEM CFD and ANSYS

CFX tools into the ANSYS meshing platform — which provides the foundationfor unifying and leveraging meshing tech-nologies, making them interoperable andavailable in multiple applications. Takingadvantage of the enhanced ANSYS Workbench 2.0 framework, the companyprovides further significant improve-ments for ANSYS 12.0 geometry andmeshing applications.

CAD ConnectionsANSYS continues to deliver a leading

CAD-neutral CAE integration environ-ment, providing direct, associative andbi-directional interfaces with all majorCAD systems, including Unigraphics®,Autodesk® Inventor®, Pro/ENGINEER®,CATIA® V5, PTC CoCreate® Modeling,SolidEdge®, SolidWorks®, and Autodesk®

Mechanical Desktop®. Software fromANSYS also supports file-based readers

for IGES, STEP, ACIS®, Parasolid®,CATIA® V4 and CATIA V5. At ANSYS12.0, geometry interfaces have beenenhanced to import more informationfrom CAD systems, including new datatypes such as line bodies for modelingbeams, additional attributes such ascolors and coordinate systems, andimproved support for named selectionscreated within the CAD systems.

For pre-processing larger models,release 12.0 includes support for 64-bitoperating systems, and smart andselective updates of CAD parts. Thenewly introduced ability to selectivelyupdate CAD components allows usersto update individual parts instead of anentire assembly, thus making geometryupdates much faster and more targeted.

Automated cleanup and repair of imported geometry:New tools automatically detect and fix typical problems,such as small edges, sliver faces, holes, seams and faceswith sharp angles. Geometry models can now be preparedfor analysis at a much faster pace. These images show anaircraft model before (top) and after (bottom) cleanup.

“ANSYS 12.0 will set the stage for majorimprovements in our design processes. Two of Cummins’core tools, ANSYS FLUENT and ANSYS Mechanical, arecoming together in the ANSYS Workbench environment. Iam also very pleased to see that geometry import continuesto improve, and we have several more meshing options.”

— Bob Tickel Director of Structural and Dynamic AnalysisCummins, Inc.

Improved surface extension: Users can select and extendmultiple groups of surfaces in a single step, a procedurethat greatly simplifies the process of closing gaps betweenparts after mid-surface extraction. The images show asample model before and after surface extension.

ANSYS Advantage • Volume III, Issue 1, 2009www.ansys.com 9999www.ansys.com 9


Geometry Handling in ANSYS DesignModeler

Geometry modeling in the ANSYSWorkbench environment is greatlyimproved to provide increasedautomation, greater flexibility andimproved ease of use for the task ofpreparing geometry for analysis. Thefeature-based, parametric ANSYSDesignModeler tool, which can be usedto create parametric geometry fromscratch or to prepare an existing CADgeometry for analysis, now includesautomated options for simplification,cleanup, repair and defeaturing.

Merge, Connect and Projectfeatures have been added for improvedsurface modeling in ANSYS 12.0. Faceand Edge merge operations can beused to easily simplify models by eliminating unnecessary features andboundaries, leading to improved meshand solution quality. The Connectoperation can be applied to ensureproper connectivity in models with gapsand overlaps.

Automated cleanup and repaircapabilities have been improved in the12.0 release. New tools automaticallydetect and fix typical problems, such assmall edges, sliver faces, holes, seamsand faces with sharp angles. Geometrymodels can now be prepared for analy-sis at a much faster pace. As always,analysis settings remain persistent afterperforming these operations and areupdated automatically in response tochanges in geometry.

Shell modeling has been enhancedin several ways, including improvedsurface extensions. The ability to selectand extend groups of surfaces greatlysimplifies the process of closing gapsbetween parts after mid-surface extrac-tion. The result is easier modeling ofwelds, for example.

Analysis-specific tools within theANSYS DesignModeler product nowinclude an automated option to extractflow volumes for fluid dynamics analy-ses. In addition, several new features,including user-defined offsets, user-defined cross sections and betterorientation controls, are available forimproved beam modeling for structuralanalyses.

Improved attribute support is available with ANSYS DesignModeler12.0. This includes options to createattributes within ANSYS DesignModeleras well as to import additional attributesfrom external CAD, including namedselections, coordinate systems andwork points.

ANSYS Meshing PlatformA primary focus for ANSYS 12.0 has

been to provide an automated meshingsolution that is best in class for fluiddynamics. With the addition of cap-abilities from GAMBIT and TGridmeshing applications, major improve-ments have been made in the automaticgeneration of CFD-appropriate tetra-hedral meshes with minimal user input.Advanced size functions (similar to thosefound in GAMBIT), prism/tet meshing(from TGrid) and other ANSYS meshingtechnologies combine to provideimproved smoothness, quality, speed,curvature and proximity featurecapturing, and boundary layer capturing.

In the area of hex meshing, the tra-ditional sweep and thin sweep methodshave seen evolutionary improvements.A new method called MultiZone hasbeen integrated into the ANSYS meshing platform. By combining existing ANSYS ICEM CFD Hexatechnology with improvements inautomation, MultiZone allows the userto automatically create hex meshes formany complex geometries withoutrequiring geometry decomposition.

Thin solid sweep method: Using the thin solid sweep meshmethod, complicated sheet metal parts can be easily hexmeshed without the need for midsurfacing or welding. Themesh can be generated to conform to the shared interfaceto increase the accuracy and speed of the solution.

MultiZone mesh method: Using the new MultiZone meshmethod, a user can mesh complicated models with a purehex mesh without the need for geometry decomposition.This brake rotor example can be meshed with a pure hexmesh in a single operation.

Patch conformal tet method with advanced size functions:With minimal input, ANSYS size function–based triangulation and inflation technology can handle advanced CFD meshing challenges, such as this benchmark aircraft model.

In the area of hybrid meshing, theMultiZone method allows for compli-cated regions to be meshed with ahybrid mesh (tet, hex-core, hex-domi-nant), further improving the flexibility andautomation of this meshing approach.For more control in key areas of concern,the Sweep and Patch Conforming methods can be employed with conformal inflation layers throughout.

Though many of these enhance-ments were driven by fluid dynamicsneeds, they also benefit users of othertypes of simulation. For example, usersperforming structural analyses will benefitfrom the improved automation and meshquality. Additional meshing enhance-ments for structural analyses include:

• Physics-based meshing improvements

• Rigid body meshing for contact

• Automated meshing of gaskets

• Improved handling of beams

• Thin solid meshing improvements

• Support for multiple elementsthrough the thickness



Enhancements toTurbomachinery Tools

With release 12.0, a number ofenhancements have been incorpo-rated into ANSYS BladeModeler,the design tool tailored to bladedgeometries for rotating machinery.Within the BladeGen component,the integrated tools for determininginitial blade shape and size (whichwere developed in conjunction withpartner PCA Engineers Limited)have been expanded to cover cen-trifugal compressors and axial fansin addition to radial turbines andcentrifugal pumps. The other com-ponent of ANSYS BladeModeler,BladeEditor, includes new bladegeometry modeling capabilities tocreate and modify one or more bladed components. As an add-in toANSYS DesignModeler, ANSYSBladeModeler provides access toANSYS DesignModeler’s extensivefunctionality to create non-standard geometry componentsand features.

ANSYS TurboGrid softwareincludes a number of evolutionaryimprovements in release 12.0, andintroduces a completely newmeshing technology. This toolfully automates a series of top-ology and smoothing steps tolargely eliminate the need tomanually adjust mesh controls,yet still generates high-qualityfluid dynamics meshes for bladedturbomachinery components.

• Generation of conformal meshesin multi-body parts

• Enhanced and new mesh controls

• Pinch features to help in defeaturing models

• Improved smoothing

• Improved flexibility in size controls and mesh refinement

• Arbitrary mesh matching toimprove node linking and solver accuracy

These improvements, though drivenby structural analysis needs, providebenefits to the entire spectrum ofANSYS users.

ANSYS ICEM CFDFor ANSYS 12.0, ANSYS ICEM CFD

meshing development focused on twoprimary tasks: improved implement-ation of ANSYS ICEM CFD meshing

technology within the ANSYS meshingplatform and continued development toenhance the ANSYS ICEM CFD productfor interactive meshing customers.Because the ANSYS ICEM CFD integra-tion involves the sharing of corelibraries, improvements made for theANSYS meshing platform also enhancethe ANSYS ICEM CFD meshing product(and vice versa).

MultiZone meshing is an example of a crossover technology that hasreceived special attention in bothANSYS meshing and the stand-aloneANSYS ICEM CFD meshing product.This hybrid meshing method combinesthe strengths of various meshers, suchas ANSYS ICEM CFD Hexa and TGrid,in a semi-automatic blocking frame-work. Within the ANSYS Workbenchenvironment, multizone automation provides multi-source, multi-targetand multi-direction sweep capabilities reminiscent of the GAMBIT Cooper tool.In the stand-alone ANSYS ICEM CFDproduct, this is an excellent way tomesh for external aerodynamics in asemi-automated way that providesrapid hybrid meshing with a high degreeof control and quality.

Improvements for ANSYS ICEMCFD 12.0 include process and interfacestreamlining, new hexa features, BFCartmesher enhancements, mesh editingadvancements, output format updatesand more. ■

Ben Klinkhammer, Shyam Kishor, Erling Eklund,Simon Pereira and Scott Gilmore of ANSYS, Inc.contributed to this article.

Hybrid mesh: Using a combination of sweep and tetra-hedral mesh methods, a user can quickly control the meshin regions of interest to improve the accuracy of the solution without the need for a pure hex mesh (and thetime required to generate it).

New developments in the ANSYS TurboGrid softwareare used to create high-quality meshes for bladedcomponents with minimal user input.Geometry courtesy PCA Engineers.

Named selection manager: This new feature allows a userto create and save named selections within CAD systemsand then to use them within ANSYS applications. Thisexample uses the named selection manager withinPro/ENGINEER.

ANSYS ICEM CFD: MultiZone meshing that combinesthe strength of various meshing tools, automaticallygenerated this hybrid grid for a tidal turbine.


Continuing to build on the foundation of prior releases,ANSYS 12.0 expands the company’s industry-leading comprehensive multiphysics solutions. New features andenhancements are available for solving both direct andsequentially coupled multiphysics problems, and theANSYS Workbench framework makes performing multi-physics simulations even faster than before.

ANSYS Workbench IntegrationThe integration of the broad array of ANSYS solver

technologies has taken a considerable step forward withrelease 12.0. The ANSYS Workbench environment has beenredesigned for an efficient multiphysics workflow by inte-grating the solver technology into one unified simulationenvironment. This platform now includes drag-and-dropmultiphysics, which allows the user to easily set up andvisualize multiphysics analysis, significantly reducing thetime necessary to obtain solutions to complex multiphysicsproblems.

Another new enhancement to the ANSYS Workbenchframework is the support for steady-state electric conduc-tion. There is a new analysis system that exposes 3-D solidelectric conduction elements (SOLID231 and SOLID232) inthe ANSYS Workbench platform. All the benefits of thispopular environment — leveraging CAD data, meshingcomplex geometry and design optimization features — arenow available for electric conduction analysis.

Also new in ANSYS Workbench at version 12.0 is sup-port for direct coupled-field analysis. Relevant elements(SOLID226 and SOLID227) are now natively supported in the ANSYS Workbench platform for thermal–electric coupling. There also is a new analysis system for thermal–electric coupling that supports Joule heating problems with

12.0: MULTIPHYSICS

Multiphysicsfor the Real WorldIn ANSYS 12.0, multiphysics capabilities continue toincrease in flexibility, application and ease of use.

temperature-dependent material properties and advancedthermoelectric effects, including Peltier and Seebeck effects.The applications for this new technology include Joule heating of integrated circuits and electronic traces,busbars, and thermoelectric coolers and generators.

Solver PerformanceANSYS 12.0 extends the distributed sparse solver to

support unsymmetric and complex matrices for both sharedand distributed memory parallel environments. This newsolver technology dramatically reduces the time needed toperform certain direct coupled solutions including Peltier andSeebeck effects as well as thermoelasticity. Thermo-elasticity, including thermoelastic damping, is an importantloss mechanism for many MEMS devices, such as block resonators and silicon ring gyroscopes.

ElementsA new family of direct coupled-

field elements is available in ANSYS12.0; these new elements enable themodeling of fluid flow through aporous media. This exciting newcapability, comprising coupledpore–pressure mechanical solids,

The project schematic shows the multiphysics workflow for a coupled electric conduction, heat transfer andsubsequent thermal stress analysis.

The electric potential for the transformer busbar shown here was analyzed within the ANSYS Workbench environment and required the use of temperature-dependentmaterial properties. Courtesy WEG Electrical Equipment.


Coupling Electromagnetics

By joining forces with Ansoft, ANSYS can delivergreater multiphysics capabilities — specifically electro-magnetics — to the ANSYS suite. The plan to integrate thiselectromagnetics technology within the existing ANSYS

simulation environment started almost immediately afterthe acquisition. While the combined development team isworking toward a seamlessly integrated bidirectional solution, several electromagnetic-centric case studies

already have demonstrated the abilityto couple electromagnetic, thermaland structural tools within theadaptive architecture of the ANSYSWorkbench environment.

For example, a high-power elec-tronic connector used in a radarapplication to connect a transmitter toan antenna must be engineered fromelectromagnetic, thermal and structuralperspectives to ensure success. Thesimulation was performed by couplingAnsoft’s HFSS software with theANSYS Workbench environment, usingadvanced thermal and structural capa-bilities. Engineers used HFSS to ensurethat the device was transmitting in the

12.0 MULTIPHYSICS

enables multiphysics modelingof new classes of civil and biomed-ical engineering problems that relyon fluid pore pressures. The elementsallow users to model fluid pore pres-sures in soils (for simulating building foundations)and biometric materials (for modeling bone in order todevelop prosthetic implants).

Fluid Structure InteractionOne of the major enhancements for fluid structure inter-

action (FSI) is a new immersed solid FSI solution. Thistechnique is based on a mesh superposition method inwhich the fluid and the solid are meshed independentlyfrom one another. The solution enables engineers to modelfluid structure interaction of immersed rigid solids withimposed motion. Rotating, translating and explicit motion ofrigid–solid objects can be defined, and the CFD solveraccounts for the imposed motion of the solid object in thefluid. This solution technique provides rapid FSI simulations,since there is no need to morph or remesh the fluid meshbased on the solid motion. The model preparation for thenew immersed solid technique is also very straightforward:The entire setup for the FSI solution can be performedentirely within ANSYS CFX software. This technology isespecially applicable to fluid structure interaction problemswith large imposed rigid-body motions, such as closingvalves, gear pumps and screw compressors. The method isalso useful for rapid first-pass FSI simulations.

Solution scaling of a thermoelectric cooler model with 500,000 degrees of freedom enables a speedup of four times for 12 processors.

Case study procedure of one-way coupling between Ansoft (blue) and ANSYS (yellow) software

Start

End

Create and solve the electromagneticapplication using HFSS

ANSYS Workbench runs HFSS in batchto perform the load interpolation

Export geometry and thermal link filefrom HFSS to ANSYS Mechanical Import surface and/or volumetric

losses using the imported load option(beta) in ANSYS Workbench

Import the geometry into ANSYS Mechanical and create the

corresponding ANSYS thermal model

Solve the ANSYS thermal model andpost-process the thermal results

4

3

2

1

Scal

e of

Sol

utio

n Sp

eed

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

Number of Processors

Sequence of images showing simulation of themotion of a screw pump solved using immersedsolid fluid structure interaction

ANSYS Advantage • Volume III, Issue 1, 2009 13

12.0: MULTIPHYSICS

Another new capability for fluid structure interaction inANSYS 12.0, FLUID136 now solves the nonlinear Reynoldssqueeze film equations for nonlinear transient FSI applica-tions involving thin fluid films. Since the nonlinear fluidic andstructural responses are coupled at the finite element level,the solution is very fast and robust for thin fluid film applica-tions. Any squeeze film application can benefit from thistechnology, including thin film fluid damping often found inRF MEMS switches.

Version 12.0 offers another exciting new FSI capability:the ability to perform one-way fluid structure interactionusing ANSYS FLUENT software as the CFD solver. This capability enables one-way load transfer for surface

temperatures or surface forces between ANSYS FLUENTand ANSYS mechanical products based on ANSYS CFX-Post. The most appropriate applications include those thatrequire one-way transfer of fluid pressures or temperaturesfrom CFD to a mechanical analysis, such as automotiveexhaust manifolds, heat sinks for electronics cooling andturbomachinery.

Multi-Field SolverThe multi-field solver (used for performing implicit

sequential coupling) contains a number of new enhance-ments at release 12.0. The first is a new solution option thatcontrols writing a multiframe restart file. This capabilityallows a user to restart an analysis from any multi-field timestep, which allows for better control over the availability of arestart file with less hard drive usage. Another enhancementis more-flexible results file controls. This capability reducesthe results file sizes for the multi-field solver, and it allows forsynchronizing the fluid and mechanical results in an FSIsolution. The final improvement is new convergence con-trols for the multi-field solution to provide more flexiblesolution controls for nonlinear convergence of the multi-fieldsolver. The applications for these enhancements are anymultiphysics application using sequential coupling includingfluid structure interaction. ■

Stephen Scampoli of ANSYS, Inc. and Ansoft LLC technical specialists contributed to this article.

The results of an RF MEMS switch solved by coupling the electrostatic, fluid andmechanical behavior of the switch in one analysis using FLUID136 to represent squeezefilm effects. Image courtesy EPCOS NL and Philips Applied Technologies.

Deformation of the high-power electronic connector can be predicted by combiningAnsoft HFSS and ANSYS Mechanical software.

solenoid. The power loss was used as an input for a thermalsimulation performed with ANSYS Mechanical software todetermine the temperature profile of the device. Subse-quently, the application predicted how the device deformeddue to the rise in temperature. Such coupling delivers a powerful analysis framework needed to solve these complex,interrelated physics problems. Thus, engineers canaddress electro-thermal-stress problems associated withoptimizing state-of-the-art radio frequency (RF) and electro-mechanical components including antennas, actuators,power converters and printed circuit boards (PCBs).

proper path, by calculating the high-frequency electro-magnetic fields, power loss density distribution and S-parameters. In such high-power applications, it is criticalto determine the temperature distribution to ensure thedevice stays below temperatures that cause material failure,such as melting. The power loss density results from the

HFSS simulation were used as the source for the thermal simulation performed withinANSYS Mechanical software,which simulated the tempera-ture distribution of the device.

In another case, a valve-actuating solenoid applicationused a coupled ANSYS andAnsoft simulation to analyzetemperature distribution.Maxwell software was used tocalculate the power loss fromthe low-frequency electro-magnetic fields within the

Eddy current and conduction loss calculated by Ansoft’s Maxwell software

www.ansys.comANSYS Advantage • Volume III, Issue 1, 20091414 www.ansys.com14

12.0: ELECTROMAGNETICS

ANSYS Emag 12.0Generates SolutionsImproved accuracy, speed and platform integration advancethe capabilities of low-frequency electromagnetic simulation.

As the combined development teams from Ansoft andANSYS set out to integrate the world-class Ansoft electronicdesign products into the ANSYS portfolio, ANSYScustomers can benefit immediately from improved andextended electromagnetics capabilities in release 12.0.

ElementsA new family of 3-D solid elements for low-frequency

electromagnetic simulation is included in the 12.0 release ofANSYS Emag software. Solid elements (SOLID236 andSOLID237) are available for modeling magnetostatic, quasi-static time harmonic, and quasi-static time-transientmagnetic fields. These two elements are formulated usingan edge-based magnetic vector potential formulation,which allows for improved accuracy for low-frequency electromagnetic simulation. The elements also provide atrue volt degree of freedom — as opposed to a time-integrated electric potential — enabling circuit couplingwith discrete circuit elements and simplifying pre- and post-processing for electromagnetic simulation.SOLID236 and SOLID237 also include much fastergauging than prior releases, which significantly reducesoverall solution times. Users can apply this new elementtechnology to most low-frequency electromagneticapplications, such as electric motors, solenoids, electromagnets and generators.

SolversAt release 12.0, the distributed sparse solver includes

support for low-frequency electromagnetics. SOLID236

and SOLID237 elements support both distributed andshared-memory parallel processing for low-frequency electromagnetic solutions. As a result of faster simulationspeeds, users can solve much larger and more complexlow-frequency electromagnetic models.

ANSYS Workbench IntegrationRelease 12.0 offers several ANSYS Workbench

enhancements for electromagnetic simulation. A newcapability facilitates multiple load step analysis for magneto-statics. This allows users to compute the magnetostaticresponse to time-dependent loading, specifying voltage andcurrent loads with time-dependenttabular data. The results aremore flexibility for magneto-static problems withtime-dependent loadsalong with transientsimulation for elec-tromagnetics, withthe addition of asimple commandsnippet, within theANSYS Workbenchenvironment.

The integrated plat-form also includes anoption for a meshed representation of astranded conductor.The current density for the new stranded conductor supports tabular loading for the new multi-step mag-netostatic analysis. This capability allows for a moreaccurate representation of current, improves overall simulation accuracy and leverages existing CAD data forcoil geometry. This new ANSYS Workbench technologycan be applied to any electromagnetic application subject to time-dependent loading, including electricmachines, solenoids and generators. ■

Stephen Scampoli of ANSYS, Inc. contributied to this article.Solution scaling of a SOLID237 model with 550,000 degrees of freedom

5

4

3

2

1

DANSYS for Low-Frequency Electromagnetics

Solu

tions

Spe

edup

Number of Processors1 2 3 4 5 6 7 8

SOLID2363-D 20-node brick

SOLID2373-D 10-node tetrahedron

Nonlinear transient rotational test rig solved in the ANSYS Workbench environment using SOLID236, SOLID237and the new stranded conductor option(TEAM24 benchmark)


12.0: FLUIDS

With release 12.0, ANSYS con-tinues to deliver on its commitment todevelop the world’s most advancedfluid dynamics technology and makeit easier and more efficient to use.Through its use, engineers can develop the most competitive prod-ucts and manufacturing processes possible. In addition to deliveringnumerous new advancements inphysics, numerics and performance,ANSYS has combined the function-ality of both ANSYS CFX and ANSYSFLUENT into the ANSYS Workbenchplatform. Customers can use thisintegrated environment to leveragesimulation technology, includingsuperior CAD connectivity, geometrycreation and repair, and advancedmeshing, all engineered to improvesimulation efficiency and compressthe overall design and analysis cycle.

Integration into ANSYS Workbench ANSYS 12.0 introduces the full

integration of its fluids products intoANSYS Workbench together with thecapability to manage simulation workflows within the environment. Thisallows users — whether they employANSYS CFX or ANSYS FLUENT soft-ware (or both) — to create, connectand re-use systems; perform auto-mated parametric analyses; andseamlessly manage simulationsusing multiple physics all within one environment.

The integration of the core CFDproducts into the ANSYS Workbenchenvironment also provides users with

access to bidirectional CADconnections, powerful geometrymodeling and advanced mesh genera-tion. (See the article Taking Shape in12.0.) Users can examine analysisresults in full detail using CFD-Post,also available within the ANSYS Workbench environment.

MultiphysicsIn some cases, fluid simulations

must consider physics beyond basicfluid flow. Both ANSYS CFX andANSYS FLUENT technologies providemany multiphysics simulation optionsand approaches, including coupling to ANSYS Mechanical software to analyze fluid structure interaction (FSI) within the ANSYS Workbenchenvironment.

Another new capability is theimmersed solid technique in ANSYSCFX 12.0 that allows users to includethe effects of large solid motion in their analyses. (See the article Multiphysics for the Real World.)

General Solver ImprovementsANSYS continues to make

progress on basic core solver speed, abenefit to all users for all types of appli-cations, steady or transient. A suite ofcases that span the range of industrialapplications has consistently shownincreases in solver speed of 10 to 20percent, or even more, for both ANSYSCFX and ANSYS FLUENT software.Beyond core solver efficiency, improve-ments to various aspects of parallelefficiency address the continued

A Flood of FluidsDevelopmentsA new integrated environment and technology enhancements make fluidssimulation faster, more intuitive andmore accurate.

growth and needs of high-performancecomputing. (See the article The Needfor Speed.)

The perennial goal of improvingaccuracy without sacrificing robustnessmotivated numerous developments,including new discretization optionssuch as the bounded second-orderoption in ANSYS FLUENT and theiteratively-bounded high-resolutiondiscretization scheme in ANSYS CFX.Being able to consistently use higher-order discretization schemes meansthat users will see further increases inthe accuracy of flow simulations withoutpenalties in robustness.

User Interface Ease of use has been enhanced in

various ways. Most noticeably, theANSYS FLUENT user interface hastaken a significant step forward byadopting a single-window interfaceparadigm, consistent with otherapplications integrated in ANSYSWorkbench. A new navigation paneand icon bar and new task pages andtools for graphics window manage-ment all reflect a more modern andintuitive interface while providingaccess to the previous version’s menubar and text user interface.

Fuel injector model with close-up of vapor volume fraction contours at the injector surface


12.0: FLUIDS

16

Multiphase Multiphase flow modeling con-

tinues to receive a great deal ofdevelopment attention, in terms ofnumerics and robustness improve-ments as well as extended modelingcapabilities. ANSYS FLUENT softwareextends the single-phase couplingtechnology, introduced previously forthe pressure-based solver, to includeEulerian multiphase simulations. Thisenhancement provides more robustconvergence, especially for steady-state flows. ANSYS CFX users will findthat improvements to the option toinclude solution of the volume fractionequations as part of the coupled set ofequations make it more broadly usablein applications with separate velocityfields for each phase. Other modelingenhancements include the implemen-tation of a wall boiling model andadditional non-drag forces in ANSYSCFX as well as more robust cavitationand immiscible fluid models in ANSYSFLUENT.

Turbomachinery The significant proportion of cus-

tomers using products from ANSYS forthe design and optimization of rotatingmachinery ensured that this fieldreceived a substantial developmentfocus. This latest release contains avariety of enhancements to core solvertechnology that couple rotating andstationary components more robustly,more accurately and more efficiently.ANSYS BladeModeler and ANSYSTurboGrid, specialized products for

bladed geometry design and mesh gen-eration, continue to evolve and improve.(See the Geometry and Meshing articlefor more details.)

An exciting new development for turbo-machinery analysts is the introduction of the through-flow code ANSYS Vista™ TF.Developed together with partner PCA Engineers Limited, Vista TF complementsfull 3-D fluid dynamics analysis to providebasic performance predictions on one ormore bladed components in a matter ofseconds, allowing users to quickly and easily screen initial designs.

And More … These enhancements represent just

the tip of the iceberg in new andimproved models and capabilities withincore fluids products from ANSYS. Someother new developments include:

• Turbulence modeling extensionsand improvements

■ Reynolds-averaged Navier–Stokes (RANS) models

■ Laminar–turbulent transition

■ Large eddy simulation (LES)

■ Detached eddy simulation (DES)

■ Scale-adaptive simulation (SAS)

• Ability to use real gas propertieswith the pressure-based solver inANSYS FLUENT and, therefore,include these in reaction modeling

• Faster, more accurate chemistryacross the board

• Dramatic speedups in view factorcalculations in ANSYS FLUENT

For ANSYS CFX software, a hostof improvements have been added tothe graphical user interface (GUI).There is a completely new capabilitythat allows users to customize GUIappearance, including the option tocreate additional input panels. Thesecustom panels provide the ability to encapsulate best practices and common processes by givingthe user control over GUI layout andrequired input.

Specific Focus Areas Internal Combustion Engines

Internal combustion (IC) enginesare a primary target application for the development of numerous features. While this development isdriven by the specific needs of ICengine simulations, it benefits manyother applications and users:

• New options and flexibility forhandling variations in physicscomplexity required at differentphases of analyses

• Further-integrated options and controls for remeshing,including an IC-specific optionfor setting up an entire enginesimulation

• Extensions and improvementsto discrete particle-trackingcapabilities

• Numerous enhancements tocombustion models and theirusability

Internal combustion engine simulation is one of the focus applications for ANSYS 12.0. This snapshot from atransient simulation of the complete engine cycle shows the flow just after the intake valves open and the direct injection of fuel. New flow feature extraction options inCFD-Post are used to highlight vortex structures withvelocity vectors. Image courtesy BMW Group.

Evolution of the free surface of oil in a reciprocating compressor. The blue area is the gas/oil rotating domain insidethe shaft, and the gray surface at the bottom shows the oil level of the reservoir. As the shaft rotates, oil is pumpedup due to body forces. Image courtesy Embraco.


12.0: FLUIDS

1717

“ANSYS CFX 12.0 showed a 30 percent solver speedupin comparison with the previous release. This significant improvement allows

us to examine more design variations in the same time, enabling further design

optimization and considerably reducing the total development time. This helps

Embraco bring our products to the market more quickly.”

— Celso Kenzo TakemoriProduct and Process Technology ManagementEmbraco

CFD-Post can be used to compare multiple designs directly, both by examiningthem side by side and by looking at the calculated difference between results.Geometry courtesy CADFEM GmbH.

In work sponsored by BMT Seatech, partially-filled tanks on marine vessels are beingsimulated by researchers at the University of Southampton to predict structural loadsand changes in vessel behavior due to the sloshing of the fluid.

• Inclusion of convective terms insolids to model conjugate heattransfer in moving solids inANSYS CFX

• Ability to model thin surfaces inANSYS CFX

• Much more in areas such as particle tracking, fuel cells,acoustics, material propertiesand population balance methods

CFD-Post An exciting introduction is the

common post-processing applicationCFD-Post. The result of combiningtechnologies from both ANSYS FLUENT and ANSYS CFX tools and building upon the well-established

CFX-Post application, CFD-Post pro-vides a complete range of graphicalpost-processing options to allow usersto visualize and assess the flow predic-tions they have made and to createinsightful 2-D and 3-D images and animations. The application includespowerful tools for quantitative analysis,such as a complete range of options forcalculating weighted averages andautomatic report-generation capabil-ities. All steps can be scripted, allowingfor fully automated post-processing.Among the specific enhancements inrelease 12.0 are the ability to open andcompare multiple cases in the sameCFD-Post session and the addition oftools to locate vortex cores in the predicted flow field.

ConclusionThis is only a sampling of what the

fluid dynamics development teamshave produced for ANSYS 12.0. Thecombined depth and breadth of CFD knowledge and experience isdelivering benefits to all users as technologies are combined and devel-opment teams drive simulationtechnology to new levels of achieve-ment. With release 12.0, ANSYScontinues its commitment to provideleading-edge CFD technology. ■

This article was written through contributionsfrom Chris Wolfe and John Stokes of ANSYS, Inc.


12.0: STRUCTURAL MECHANICS

Designing with StructureAdvancements in structural mechanics allow more efficient and higher-fidelity modeling of complex structural phenomena.

The ability to drive the engineering design process instructural applications has taken a significant step forwardwith the improvements in release 12.0. New features andtools, many integrated into the ANSYS Workbench platform,help reduce overall solution time. Specific improvementsfocus on elements, materials and contact and solverperformance, along with linear, rigid and flexible dynamics.

ElementsThe most notable new element in release 12.0 is the

four-noded tetrahedron for modeling complex geometries inhyperelastic or forming applications. The element providesa convenient way to automate the meshing of complexstructures, avoiding the need for pure hexahedral meshes.This reduces the time it takes to develop a case from geom-etry through solution, while maintaining the accuracy of thesolution. See the table below for a summary of new andenhanced elements.

When simulating a nonlinear process, large deformationcan introduce too much distortion of the elements. Resolving

this requires local remeshing during the simulationprocess. The 2-D rezoning introduced with release 11.0extends further in ANSYS 12.0, increasing the flexibility ofthe remeshing process: The user can now define transitionregions within the refined zones and use meshes createdin external meshing tools.

MaterialsAccounting for proper cyclic softening or hardening or

damage of materials is a key factor for elastomer applica-tions and, more generally speaking, any structure whosematerial variation depends on the strain rate. Release 12.0introduces several additions to the wide choice of materi-als already available. Other feature improvements include:

• Rate-dependent Chaboche plasticity, which canbenefit turbine and engine design

• Bergström–Boyce model to enhance elastomermodeling capabilities

• New damage model based on the Ogden–Roxburgh formulation

Element New Improved Capability Applications

Four-noded tetrahedron X Provides a convenient way to automatemeshing of complex structures, avoidingneed for pure hexahedral meshes

Modeling complex geometries for forming or hyperelastic applications

General axisymmetric element X Supports contact Compatible with 3-D non-axisymmetric loading and canuse arbitrary axis of rotation

Various pipe model elements X Increased accuracy To provide refined behavior of structures in case of ovalization, warping or similar deformations of cross section for thin or moderately thick pipes and nonlinear material behavior support

Shell: linear, quadratic, axisymmetric X Improved shell thickness updating schemeand improved convergence

Provides greater accuracy in the behavior of shell modelsas well as a faster solution for nonlinear problems

Beam X Supports cubic shape function Provides additional accuracy to coarse meshes and greater support of complex load patterns

Reinforcement elements X Allows modeling of discrete fibers with a variety of nonlinear material behavior

Stresses in reinforcements can be analyzed separately from host elements

Summary of new and enhanced element features in ANSYS 12.0 structural analysis products

Warping and ovalization of pipe structureswith the new pipe elements



• Anand’s viscoplasticity model, useful for metalforming applications such as solder joints

• Improvements in the calculation of J-integrals toaccount for mixed-mode stress intensity factors,which benefit improvements in fracture mechanics

• Initial strain and initial plastic stress import capabilities that allow for state transfer from a 2-D model to a 3-D model

ContactAs assemblies have become a de facto standard in

simulation, the need for advanced contact features hasgrown accordingly. ANSYS 12.0 developments include anumber of additional contact modeling features as well assignificant improvements in solving contact problems.

While Coulomb’s law for friction is widely used, there arecircumstances in which more elaborate modeling isrequired, such as wear modeling or pipelines resting on seabeds. Release 12.0 supports a friction coefficient definitionthat depends upon the contact state itself and accounts forcomplex frictional behavior. Specifically, the user is able todefine the dependency of the friction on contact parame-ters, such as sliding distance or contact pressure.

A typical contact application involves seals that are sub-ject to fluid pressure. Release 12.0 provides support of fluidpressure penetration, to model scenarios in which pressurerises higher than the contact pressure around the seal.Pressures in such cases can be applied only on the freefaces of the structure and evolve with the contact state.

Contact simulation is usually a time-consumingprocess. The latest release introduces contact modelingimprovements that significantly reduce computation timeand results file size. These enhancements include new

Performance of new modal solver

80,000

60,000

40,000

20,000

0

Number of Modes

contact search algorithms, contact trimming logic andsmart over-constraint elimination for multipoint constraint(MPC) contact.

Solver PerformanceSolver performance has improved in many different

areas. ANSYS 12.0 introduces a new modal solver, calledSNODE, that increases the speed of computation for prob-lems with a large number of modes — in the realm ofseveral hundred — on large structures that typically haveover a million degrees of freedom. This solver is well suitedfor automotive or aerospace applications and for largebeams and shell assemblies. Beyond its ability to computea larger number of modes in a reduced amount of time,SNODE also significantly reduces the amount of I/Orequired to compute the solution. (See the SupernodeEigensolver article.)

Many enhancements have been made to the distributedsolver to improve the scalability of the solution. (See the article on High Performance Computing.) More solvertechniques are supported, including:

• Partial solve capability that computes only a portionof the solution

• Prestressed analysis

• Models that employ the use of unsymmetric matrices, which are useful for scenarios that involvehigh-friction coefficients, for example

These new features can be combined for applicationssuch as brake squeal, which might combine the partialsolve and unsymmetric matrix capabilities.

Crack tip analysis of turbine bladeCourtesy PADT

Linear DynamicsSome of these element, material, contact and solver

improvements benefit the field of linear dynamics as well.They are complemented by enhancements specific to thissimulation area, especially for mode superposition analysis.For harmonic or transient loadings, the mode superpositionmethods exhibit better performance, especially during the

CPU

Tim

e (s

econ

ds)

100 1,000 4,000 8,000

Block Lanczos

Supernode

so-called expansion pass that computes results at each frequency or time step on the full model. For very largestructures, the total computation effort can be reduced by up to 75 percent. The mode combination for spectralanalysis benefits from similar advancements. Instability predictions, such as the case of brake squeal, can becomputed faster due to several enhancements to thedamped eigensolver.

The introduction of ANSYS Variational Technology provides faster mode computation for cyclic symmetricstructures, such as those found in many turbine applications. Using this technique can typically improve

Instability analysis for brake squeal Modal analysis of a cyclic–symmetric geometry

Courtesy PADT, Inc.

solution speed by a factor of three or four — the greater thenumber of sectors, the better the performance.

Rotating machinery applications profit from an extendedset of capabilities for rotordynamics analysis. These includethe extension of the gyroscopic effect to shell and 2-D elements and inclusion of rotating damping that takeshysteretic behavior into account.

Random vibration and spectral analysis users gain newtools as well as a greater flexibility in modeling structures,including support of spectrum analysis in the ANSYS Workbench platform. New tools include the United StatesNuclear Regulatory Commission–compliant computation ofmissing masses and support of rigid modes, along with theability to use residual vectors to account for higher energy modes. The global number of spectra appliedsimultaneously to the structure has been increased up to50 as has the number of modes used in a combination —now up to 10,000.

When analyzing design variations, comparing datafrom different simulation cases, or correlating simulationand test data, comparison between modal content of themodels is required. The modal assurance criterion (MAC)in release 12.0 provides a convenient tool to compare theresults of two modal analyses. Typical use cases for thecriteria include tuning of misaligned turbine blades orvalidation of new component designs, each with respectto their vibration behavior.

New Element Reduces Meshing TimeZF Boge Elastmetall GmbH develops, manufactures and

supplies vibration control components and parts for theautomotive industry. These components include plasticparts, energy-absorbing elements for vehicle safety, and rubber–metal components such as chassis suspensionmounts, control arm bushes (also known as bushings) andengine mounts.

The German company uses simulation to reducedevelopment time and costs. When developing models forcomponents with hyperelastic material properties, companyengineers require an element type that can be freelymeshed; can accommodate extreme deformation, stablecontact and short computing time; and can provide reliable results.

By using the new SOLID285 four-noded tetra-hedron element available in ANSYS 12.0, ZF BogeElastmetall engineers considerably reduced meshingtime. Close correlation between the simulation and physicalmeasurement allowed them to determine the spring rate ofstrongly deformed structures without the complex and

Deformation ofan automotivesuspensionmount



time-consuming meshing that was previously requiredwhen using hexahedral elements. Boge’s work proved thatby employing this new element, users can determinethe stresses and strains for a durability calculation in a reasonable time.


ANSYS Workbench IntegrationThe integration of the structural applications within

the ANSYS Workbench platform provides additionalproductivity to users, including:

• New meshing techniques to improve mesh quality

• Support of additional elements, such as gasket elements as well as quadratic shells and beams that include offset definitions

• Boundary condition definitions that provide a spatial dependency for loads

• Coupling conditions

• Remote points

Multibody DynamicsAt release 12.0, a number of improvements in the

general area of multibody dynamics enable the rapid designand analysis of complete mechanical systems undergoinglarge overall motion. ANSYS Rigid Dynamics software has anew Runge–Kutta 5 integrator, the preferred solution for longtransient simulations. A new bushing joint, a “stops andlocks” option for most other joint types, and the ability to specify preload for springs give new flexibility when simulating complex multiple-part assemblies andcomponent interactions.

For complex assemblies, conducting an initial simulationwith the ANSYS Rigid Dynamics product is the key toachieving robust flexible dynamics results. Creating over-constrained assemblies is an inconvenient reality; release12.0 adds a redundancy analysis and repair tool to identifyoverconstrained assemblies, points out which joints ordegrees of freedom are redundant, and allows selectiveunconstraining to create a properly constrained mechanism.

A number of improvements to data and process handling increase ease of use for multibody simulations:

• Enhanced load data fitting (no longer requires curve fitting)

• Ability to read in complex load input,such as simulated or measured multi-channel road surface or seismicdata, and apply as load data to partsor joints

• Ability to use remote solution manager(RSM) to offload the solving effort to aserver or other capable CPU (benefits long-duration and multi-channel input transientsimulations)


Multibody dynamics capabilities were used to simulate this leaf spring suspension.

• Ability to export forces and moments at any timewithin a transient simulation

For durability studies, exported loads can be used in astatic structural analysis as an efficient first-pass failureanalysis. Although it won’t provide the complete pictureobtained from comprehensive flexible dynamics simulation,a static structural simulation is typically much less compu-tationally expensive. Flexible dynamics simulations benefitat release 12.0 from robust component modal synthesis, orCMS. This method uses an internal substructuringapproach and requires that the CMS parts of an assemblyare constructed with linear materials. The procedure simpli-fies a problem by accounting only for a few degrees offreedom, which results in solution times that are often afraction of those found using the standard full computationmethod. Time-to-solution reductions of several hundredpercent are not uncommon.

• Ability to associate contact to the top or bottom of shell face

Post-processing capabilities have drastically improvedwith release 12.0. The user can now plot any structural sim-ulation data stored in the results files. Mathematicaloperations involving elementary results can be introducedto create additional user-defined criteria. Complex modeshapes, plotting on linear paths, stress linearization (whichdepends upon path plotting), and the ability to displayunaveraged results at element nodes complement the list ofthe features that increase productivity at ANSYS 12.0. ■

Pierre Thieffry and Siddharth Shah of ANSYS, Inc. contributed to this article.


12.0: EXPLICIT DYNAMICS

Explicit Dynamics Goes MainstreamANSYS 12.0 brings native explicit dynamics to ANSYS Workbench and provides the easiest explicit software for nonlinear dynamics.

ANSYS has expended significanteffort in the area of explicit dynamics forrelease 12.0 — including the addition ofa new product that will make this tech-nology accessible to users independentof their simulation experience. In addi-tion, enhancements to both the ANSYSAUTODYN and ANSYS LS-DYNAproducts provide considerable benefitsto their users.

Newly introduced in ANSYS 12.0,ANSYS Explicit STR software is the firstexplicit dynamics product with a nativeANSYS Workbench interface. It is basedon the Lagrangian portion of the ANSYSAUTODYN product. The technology willappeal to those who want to model transient dynamic events such as droptests, as well as quasi-static eventsinvolving rapidly changing contact conditions, sophisticated material failure/damage and/or severe displace-ments and rotations of structures. Inaddition, it will appeal to users who canbenefit from the productivity provided byother applications integrated within theANSYS Workbench environment.Those who have previous experienceusing ANSYS Workbench will find that

they already know most of what is needed to use ANSYS Explicit STR.

The ANSYS Explicit STR tool is wellsuited to solving:

• Drop tests (electronics and consumer goods)

• Low- to high-speed solid-to-solidimpacts (a wide range of applica-tions from sporting goods toaerospace)

• Highly nonlinear plastic buckling events (for ultimate limit state design)

• Complete material failure applications (defense and homeland security)

• Breakable contact, such as adhesives or spot welds (electronics and automotive)

The real benefit of ANSYS ExplicitSTR software is the work flow afforded by operating in the ANSYS Workbench environment. While many differentsimulation processes are possible, here is an example of the typical steps a usermight take:

• Associatively link to a parametricCAD model or import a geometry

• Create a smooth explicit mesh using the new explicit preferenceoption and/or patch-independentmesh method within the ANSYSmeshing platform; automatically create part-to-part contact by usingthe new body interactions tool

• Fine-tune contact specifications ifdesired by utilizing breakable or eroding contact options

• Load and/or support an assemblyand/or parts as usual

• Assign material properties from thecomprehensive material library

• Solve interactively either in thebackground or via remote solutionmanager (RSM)

• View progress of solution in real time using concurrent post-processing capability, new toANSYS Workbench at 12.0

• Explore alternative design ideas via parametric changes to the CAD model and easily perform re-solves,just like other ANSYS Workbenchbased applications

• Use the ANSYS Design Explorationcapability to automate the para-metric model space exploration

In addition, users of the full version ofANSYS AUTODYN (structural- plus fluids-capable) have access to the ANSYSExplicit STR interface; consequently, theywill be able to transfer implicit solutionsfrom the ANSYS Workbench environmentfor doing implicit–explicit solutions, suchas bird strike analysis of a pre-stressed fanblade. ANSYS LS-DYNA software userswill be able to use the pre-processing portion of ANSYS Explicit STR and outputa .K file for solving and post-processingoutside of ANSYS Workbench. ■

Wim J. Slagter of ANSYS, Inc. is available toanswer your questions about explicit dynamics.ANSYS Explicit STR is the first explicit dynamics product with a native ANSYS Workbench interface.


12.0: EIGENSOLVER

Introducing the Supernode EigensolverA new eigensolver in ANSYS 12.0 determines large numbers of natural frequency modes more quickly and efficiently than conventional methods.By Jeff Beisheim, Senior Development Engineer, ANSYS, Inc.

In a wide range of applications, parts are subject tocyclic mechanical loading, and engineers must use aneigensolver to determine the structure’s natural frequencies— also known as eigen modes. With some modes, largevibration amplitudes can interfere with product performanceand cause damage, such as fatigue cracking. In mostcases, only the first few modes with the largest deforma-tions are of particular interest, though determining evendozens of modes can be common.

In the CAE industry, the block Lanczos eigensolver istypically used more than any other for these types of calcu-lations. This proven algorithm has been used in many finiteelement software packages, including ANSYS Mechanicaltechnology. It brings together the efficiency and accuracy ofthe Lanczos algorithm and the robustness of a sparse directequation solver. The software works in a sequential fashionby computing one mode (or a block of modes) at a time untilall desired modes have been computed.

Although the method is considered efficient in solvingfor each of these eigen modes, the amount of time andcomputer resources (both memory and I/O) required addsup when many dozens of eigen modes must be found.Elapsed solution times of several hours — or days — aretypical in applications that involve thousands of modes.Generally, determining large numbers of modes is requiredin capturing system response for studies such as transientor harmonic analyses using the mode superpositionmethod.

For such cases, the ANSYS release 12.0 includes a newsupernode eigensolver. Instead of computing each modeindividually and working with mode shapes in the globalmodel space, the supernode algorithm uses a mathematicalapproach based on substructuring to simultaneously deter-mine all modes within a given frequency range and tomanage data in a reduced model space.

By utilizing fewer resources than block Lanczos, thissupernode eigensolver becomes an ideal choice when solving on a desktop computer, which can have limitedmemory and relatively slow I/O performance. When com-bined with current eigensolver technology already availablein mechanical software from ANSYS, virtually all modalanalyses can be efficiently solved.

The ANSYS supernode eigensolver is well suited for applications such as seismic analysis of power plant cooling towers, skyscrapers and other structures in which hundreds of modes must be extracted to determine the response of the structures to multiple short-duration transient shock/impact loadings.

Comparing EigensolversA sample comparison shows that the super-

node eigensolver offers no significant performanceadvantage over block Lanczos for a low number ofmodes. In fact, supernode is slower when 50 or fewer modes are requested. However, whenmore than 200 modes are requested, the supernode eigensolver is significantly fasterthan block Lanczos — with efficiency increasingconsiderably as the number rises.

Performance of block Lanczos and supernode eigensolvers at 1 million DOF

25,000

20,000

15,000

10,000

5,000

0

Solv

er E

laps

ed T

ime

(sec

onds

)

10 50 100 200 500 1,000 2,000

Number of Requested Modes

Block Lanczos

Supernode

Linux 64-bit server: 3.4 GHzsingle-core Xeon®, Red Hat® 4,64 GB RAM, 200 GB disk

© is

tock

phot

o.co

m/M

atej

ay


12.0: EIGENSOLVER

Using Supernode EigensolverThe supernode eigensolver can be selected in the

ANSYS Mechanical traditional interface using the SNODElabel with the MODOPT command or via the Analysis Optionsdialog box. ANSYS Workbench users can choose thiseigensolver by adding a command snippet that includes theMODOPT,SNODE command.

The MODOPT command allows users to specify the num-ber of natural frequencies and what range those frequencieslie within. With other eigensolvers, the number of requestedmodes primarily affects solver performance, while the fre-quency range is, essentially, optional. Asking for moremodes increases solution time, while the frequency rangegenerally decides which computed modes are computed.

The supernode eigensolver behaves completely oppo-site: It computes all modes within the specified frequencyrange regardless of how many modes are requested.Therefore, for maximum efficiency, users should input a range that covers only the spectrum of frequenciesbetween the first and last mode of interest. The number of modes requested on the MODOPT command thendecides how many of the computed frequencies are provided by the software.

Today, with the prevalence of multi-core processors, thefirst release of this new eigensolver will support shared-memory parallelism. For users who want full control of thesolver, a new SNOPTION command allows control over several important parameters that affect accuracy and efficiency.

Controlling ParametersThe supernode eigensolver does not compute exact

eigenvalues. Typically, this is not an issue, since the lowestmodes in the system (often used to compute the dominantresonant frequencies) are computed very accurately — gen-erally within less than 1 percent compared to using blockLanczos. Accuracy drifts somewhat with higher modes,however, in which computed values may be off by as muchas a few percent compared with Lanczos. In these cases,the accuracy of the solver may be tightened using the range

Examining Real-World PerformanceA heavy-equipment cab model with over 7 million equations was

used to demonstrate the power of the supernode eigensolver. Thismodel was solved using a single core on a machine with the Windows® 64-bit operating system with 32 gigabytes of RAM. Timespent computing 300 modes with block Lanczos was about 31.8hours. The solution time dropped to 15.7 hours (a two-timesspeedup) using the supernode eigensolver. The model illustratesreal-world performance for a bulkier model with only 300 modesrequested. For modal analyses in which hundreds or thousands ofmodes are requested, users often see a speedup of 10 times or morewith the supernode eigensolver compared with block Lanczos. Inone recent project, a major industrial equipment manufacturerreduced analysis run time from 1.5 hours to just 10 minutes byswitching from block Lanczos to supernode eigensolver.

Total displacement for the tenth-lowest natural frequency is plotted for a heavy-equipment cab modelrepresented by more than 7 million equations.Model courtesy PTC.

factor (RangeFact) field on the SNOPTION command.Higher values of RangeFact lead to more accurate solu-tions at the cost of extra computations that somewhat slowdown eigensolver performance.

When computing the final mode shapes, the super-node eigensolver often does the bulk of I/O transfer to andfrom disk, and the amount of I/O transfer is oftensignificantly less than a similar run using block Lanczos. Tomaximize supernode solver efficiency, I/O can be furtherminimized using the block size (BlockSize) field on theSNOPTION command. Larger values of block size willreduce the amount of I/O transfer by holding more data inmemory during the eigenvalue/eigenvector output phase,which generally speeds up the overall solution time. However, this is recommended only if there is enoughphysical memory to do so.

Application GuidelinesThe following general guidelines can be used in deter-

mining when to use the supernode eigensolver, which istypically most efficient when the following three conditionsare met:

• The model would be a good candidate for using thesparse solver in a similar static or full transient analysis (that is, dominated with beam/shell elementsor having thin structure).

• The number of requested modes is greater than 200.

• The beginning frequency input on the MODOPTcommand is zero (or near zero).

For models that have dominantly solid elements orbulky geometry, the supernode eigensolver can be moreefficient than other eigensolvers, but it may require highernumbers of modes to consider it the best choice. Also,other factors such as computing hardware can affect thedecision. For example, on machines with slow I/O perform-ance, the supernode eigensolver may be the better choice,even when solving for less than 200 modes. ■


12.0: HIGH-PERFORMANCE COMPUTING

25

The Need for SpeedFrom desktop to supercomputer, high-performance computing withANSYS 12.0 continues to race ahead.

Tuning software from ANSYS on the latest high-performance computing technologies for optimal performancehas been — and will continue to be — a major focus area within the software development organization at ANSYS. Thiseffort has yielded significant performance gains and new functionality in ANSYS 12.0, with important implications formore productive use of simulation by customers.

High-performance computing, or HPC, refers to the use ofhigh-speed processors (CPUs) and related technologies tosolve computationally intensive problems. In recent years,HPC has become much more widely available and afford-able, primarily due to the use of multiple low-costprocessors that work in parallel on the computationaltask. Today, clusters of affordable compute servers makelarge-scale parallel processing a very viable strategy forANSYS customers. In fact, the new multi-core proces-sors have turned even desktop workstations intohigh-performance platforms for single-job execution.

This wider availability of HPC systems is enablingimportant trends in engineering simulation. Simu-lation models are getting larger — using morecomputer memory and requiring morecomputational time — as engineersinclude greater geometric detail andmore-realistic treatment of physical phenom-ena (Figure 1). These higher-fidelity models are critical forsimulation to reduce the need for expensive physical testing.HPC systems make higher-fidelity simulations practical byyielding results within the engineering project’s required timeframe. A second important trend is toward more simulations— enabling engineers to consider multiple design ideas, conduct parametric studies and even perform automateddesign optimization. HPC systems provide the throughputrequired for completing multiple simulations simultaneously,thus allowing design decisions to be made early in the project.

Software from ANSYS takes advantage of multi-processor and/or multi-core systems by employingdomain decomposition, which divides the simulation modelinto multiple pieces or sub-domains. Each sub-domain is thencomputed on a separate processor (or core), and the multipleprocessors work in parallel to speed up the computation. In theideal case, speedup is linear, meaning that the simulation turn-around time can be reduced in proportion to the number ofprocessors used. Parallel processing also allows larger problems to be tackled, since the processing power and memory requirements can be distributed across the cluster ofprocessors. Whether performed on a multi-core desktop work-station, desk-side cluster or scaled-out HPC system, parallel

HPC on Workstations?While purists might argue whether workstations can

be considered high-performance computing platforms,the performance possibilities for ANSYS 12.0 running onworkstations are noteworthy. With the latest quad-coreprocessor technology, an eight-core workstation runningWindows® can deliver a speedup of five to six times forusers of mechanical products from ANSYS (Figure 2)and over seven times for users of its fluid dynamicsproducts (Figure 4). This means that parallel processingnow provides tremendous ROI for both large engineeringgroups and individual workstation users, enabling fasterturnaround, higher-fidelity models and parametric modeling. With release 12.0 and 2009 computing platforms, parallel processing improves productivity for all simulation types, from workstation to cluster, formechanical or fluids simulations.

Figure 1. Simulations as large as 1 billion cells are now supported at release 12.0.This 1 billion-scale racing yacht simulation was conducted on a cluster of 208 HPProLiant™ server blades. (For more information, visit www.ansys.com/one-billion.) Image courtesy Ignazio Maria Viola.



Figure 3. Scaling of a 10M DOF simulation using the ANSYS Mechanical 12.0 iterativePCG solver on a cluster of Intel Xeon 5500 Processor series. All cores on these quad-core processors are fully utilized for the benchmark.

processing provides excellent return on investment by improvingthe productivity of the engineers who perform simulation.

ANSYS 12.0 provides many important advances in areasrelated to parallel processing and HPC, delivering scalabilityfrom desktop systems to supercomputers. For users of theANSYS Mechanical product line, release 12.0 introducesexpanded functionality in the Distributed ANSYS (DANSYS)solvers, including support for all multi-field simulations, pre-stress effects and analyses involving cyclic symmetry. Inaddition, DANSYS now supports both symmetric and non-symmetric matrices as well as all electromagnetic analyses.Mechanical simulations benefit from significantly improvedscaling on the latest multi-core processors. Simulations in thesize range of 2 million to 3 million degrees of freedom (DOF)now show good scaling on eight cores (Figure 2). Based onbenchmark problem performance, customers can expect toget answers back five to six times faster on eight cores. Evenmore impressive is the scale-out behavior shown in Figure 3,with a 10 million DOF simulation showing solver speedup of68 times on 128 cores.

With turnaround times measured in tens of seconds,parametric studies and automated design optimization arenow well within the grasp of ANSYS customers who performmechanical simulations. These benchmarks are noteworthy,in part, as they show execution with all cores on the clusterfully utilized, indicating that the latest quad-core processorshave sufficient memory bandwidth to support parallel processing for memory-hungry mechanical simulations. Software tuning has contributed to improved scaling as well,including improved domain decomposition, load balancingand distributed matrix generation. To help customers maxi-mize their ANSYS solver performance, the online help systemnow includes a performance guide that provides a compre-hensive summary of factors that impact the performance ofmechanical simulations on current hardware systems.

Explicit simulations using ANSYS AUTODYN technologytake great advantage of HPC systems at release 12.0. Full 64-bit support is now available, allowing much larger simu-lations to be considered from pre-processing to solution and post-processing.

For users of fluid dynamics software from ANSYS, release12.0 builds on the strong foundation of excellent scaling inboth the ANSYS FLUENT and ANSYS CFX solvers. These fluids simulation codes run massively parallel, with sustainedscaling at hundreds or even thousands of cores. The releaseincorporates tuning for the latest multi-core processors,including enhanced cache re-utilization, optimal mapping andbinding of processes to cores (for better memory locality andsystem utilization), and leveraging the latest compiler opti-mizations. The resulting ANSYS FLUENT and ANSYS CFXperformance on the newly released Intel® Xeon® 5500 Processor series is shown in Figure 4, with outstandingspeedup of over seven times for many benchmark cases. Inaddition, the new release delivers significant performanceimprovements at large core counts, the result of generalsolver enhancements and optimized communications overthe latest high-speed interconnects. Figure 5 demonstrates

80706050403020100

1 2 4 8 32 64 128

Core

Sol

ver S

peed

up

Number of Cores

9.00

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

0.00

Ideal

BMD-2

BMD-3

BMD-4

BMD-5

BMD-6

BMD-7

12

48

8.00

5.745.46

8.83

5.53 5.83

4.55

Numbe

r of C

ores

Core

Sol

ver S

peed

up

8.00

7.00

6.00

5.00

4.00

3.00

2.00

1.00

0.00

Ideal

FL5L2

FL5L3

Sedan_4M

Truck_14M

Indy Car

LeMans

Turbine1

24

8

8.00

7.16 7.18

6.23

7.58 7.55

6.796.47

Numbe

r of C

ores

Core

Sol

ver S

peed

up

ANSYS FLUENTANSYS CFX

Figure 4. Scalability of ANSYS FLUENT and ANSYS CFX benchmark problems on theIntel Xeon 5500 Processor series quad-core platform. Simulations running eight-wayparallel show typical speedup of over seven times.

Figure 2. Speedup of Distributed ANSYS Mechanical 12.0 software using the 11.0 SP1 benchmark problems. Simulations running eight-way parallel show typicalspeedup of between five and six times. Data was collected on a Cray CX-1 PersonalSupercomputer using two quad-core Intel Xeon Processor E5472 running Microsoft®

Windows HPC Server 2008.



27

Figure 5. Scalability of ANSYS CFX 12.0 on a 10M node transonic airfoil benchmarkexample. Data was collected on a cluster of AMD Opteron™ 2218 processors, showingthe benefit of a high-speed interconnect.

Figure 7. Parallel I/O in ANSYS FLUENT 12.0 using the Panasas© file system, comparedto serial I/O in the previous release using NFS. Parallel treatment of I/O provides important speedup for time-varying simulations on large clusters.

273

345334

PanFS – FLUENT 12

NFS – FLUENT6.3

Figure 6. Scaling of ANSYS FLUENT 12.0 software is nearly ideal up to 1,024 processorsand 78 percent of ideal at 2,048 processors. Data courtesy SGI, based on the SGI Altix®

ICE 8200EX using quad-core Intel Xeon Processor E5472 with Infiniband®.

scaling achieved by ANSYS CFX software on a cluster ofquad-core AMD processors. Nearly ideal linear scaling to1,024 cores — and very good efficiency up to 2,048 cores —has been demonstrated with ANSYS FLUENT (Figure 6). Bothfluids codes provide improvements to mesh partitioning thatenhance scalability. ANSYS FLUENT software now providesdynamic load balancing based on mesh- and solution-derived criteria. This enables optimal scalability forsimulations involving multiphysics, such as particle-ladenflows. The ANSYS CFX code delivers improved partitioningfor moving and/or rotating meshes, yielding important reductions in memory use and improved performance for turbomachinery and related applications. Finally, ANSYSFLUENT users will benefit from several usability improve-ments, including built-in tools for checking system networkbandwidth, latency and resource utilization — all helping toidentify potential scaling bottlenecks on the cluster.

Beyond solver speedup, the ANSYS 12.0 focus on HPCaddresses issues related to file input and output (I/O). BothANSYS FLUENT and ANSYS CFX software have updated I/Oalgorithms to speed up writing of results files on clusters,enhancing the practicality of periodic solution snapshotswhen checkpointing or running time-dependent simulations.ANSYS FLUENT includes improvements in the standard file I/Oas well as new support for fully parallel I/O based on parallel filesystems. Order of magnitude improvements in I/O throughputhave been demonstrated on large test cases (Figure 7), virtuallyeliminating I/O as a potential bottleneck for large-scale simula-tions. ANSYS CFX improves I/O performance via datacompression during the process of gathering from the clusternodes, therefore reducing file write times. Proper I/O configura-tion is also an important aspect of cluster performance for theANSYS Mechanical product line.

Recognizing that cluster deployment and managementare key concerns, ANSYS 12.0 includes a focus on compati-bility with the overall HPC ecosystem. ANSYS products areregistered and tested as part of the Intel Cluster Ready pro-gram, confirming that these products conform to standards ofcompatibility that contribute to successful deployment(www.ansys.com/intelclusterready). In addition to supportingenterprise Linux® distributions from Red Hat® and Novell,ANSYS 12.0 products are supported on clusters based onMicrosoft Windows HPC Server 2008. ANSYS has alsoworked with hardware OEMs, including HP®, SGI®, IBM®, Dell®,Cray® and others, to define reference configurations that areoptimally designed to run simulation software from ANSYS(www.ansys.com/reference-configs).

As computing technology continues to evolve, ANSYS isworking with HPC leaders to ensure support for the break-through capability that will make simulation more productive.Looking forward, important emerging technologies includemany-core processors, general purpose graphical processingunits (GP-GPUs) and fault tolerance at large scale. ■

Contributions to this article were made by Barbara Hutchings,Ray Browell and Prasad Alavilli of ANSYS, Inc.

Spee

dup

Number of Cores

Ideal

Infiniband

Gigabit

120.00

100.00

80.00

60.00

40.00

20.00

0.00

2,048

1,536

1,024

512

00 512 1,024 1,536 2,048

Number of Cores

I/O R

ate

for D

ata

Writ

e (M

Byte

s/se

c) 400

300

200

100

0

273

39X

345334

64 128 256

Solv

er S

peed

up

Number of Cores

1,581

991

521

265128

0 16 32 48 64 80 96

PanFS — FLUENT 12.0

NFS — FLUENT6.3

Speedup

Ideal

Transonic Airfoil BenchmarkModel — 10M Nodes

Truck Benchmark Model — 111M Cells

31X 20X

12.0: FUTURE DIRECTIONS

Foundationsfor the FutureThe many advanced features of ANSYS 12.0 were designed to solvetoday’s challenging engineering problems and to deliver a platform fortomorrow’s simulation technology.

As this special spotlight in ANSYS Advantage attests,release 12.0 delivers a compelling advancement in what theCAE industry has, until now, only envisioned — a full rangeof best-in-class simulation capabilities assembled into aflexible multiphysics simulation environment specificallydesigned to increase engineering insight, productivity andinnovation. Whether the need is structural analysis, fluidflow, thermal, electromagnetics, geometry preparation ormeshing, ANSYS customers can rely on release 12.0 for thedepth and breadth of simulation capabilities to overcometheir engineering challenges.

Staying true to our commitment to develop the mostadvanced simulation technologies, release 12.0 has furtherexpanded the depth of individual physics and more intimately coupled them to form an engineering simulationcapability second to none. A multitude of new materialmodels, physics and algorithms enable simulating real-world operating conditions and coupled physical phenomena, while new solver technology and parallel processing improvements have dramatically reduced runtimes and made complete system simulations more computationally affordable.

Shouldering the array of technology in release 12.0 is our next-generation simulation platform, ANSYSWorkbench 2.0. Seamlessly spanning all stages of engineering simulation, ANSYS Workbench 2.0 has beenengineered to manage the complexities of today’s simu-lations and to accelerate innovation.

Release 12.0 is a notable milestone in the company’snearly 40-year history of innovating engineering simulation,and it sets the stage for a new era of Smart EngineeringSimulation — an era in which ANSYS customers will gainmore from their investment in simulation by increasing theefficiency of their processes, increasing the accuracy oftheir virtual prototypes, and capturing and reusing their simulation processes and data. However, the advance-ments of ANSYS 12.0 notwithstanding, the journey is farfrom complete. To address the simulation challenges on thehorizon, ANSYS will continue to reinvest in research anddevelopment and to explore new technologies. In particular,there are a few areas that we consider vital in the pursuit ofSimulation Driven Product Development — areas in whichANSYS has laid strong foundations and remains committedto build upon as we look beyond release 12.0.

Physics FirstANSYS customers rely heavily on simulation before

making commitments to product designs or manufacturingprocesses. High-fidelity engineering simulation is absolutelyparamount when upstream engineering decisions candetermine the overall success of a product and, in somecases, the company’s financial success. At ANSYS, webelieve our customers should never have to compromise bymaking broad-based engineering assumptions due to limitations in their analysis software. That is why we havetaken a comprehensive multiphysics approach to simula-tion, and it starts with a foundation of individual physics.Looking beyond release 12.0, ANSYS will continue to investand demonstrate leadership in all the key physics. And as



we develop tomorrow’s advanced capabilities, we will con-tinue to allow them to be combined in ways that freeengineers from making the assumptions associated withsingle-physics simulations. Within the ANSYS Workbenchsimulation paradigm, we will enable engineers to routinelyconsider the effects of fully coupled physical phenomena.

High-Performance ComputingAs one might expect, high-performance computing

(HPC) is a strategic enabling technology for ANSYS. Theappearance of quad-core machines on the desktop and theincreased availability of compute clusters have ushered in a new era of parallel and distributed computing for our customers. ANSYS has kept pace with the exponentialincrease in computational horsepower with prolific develop-ment in the areas of parallel and distributed computing andnumerical methods. The result is improved scalability anddramatically reduced run times for large-scale fluid flow,structural and electromagnetic simulations.

Solving large-scale problems with meshes exceeding 1 billion cells has been the latest stretch goal for fluid flow simulation. Recently, HPC and software from ANSYSwere combined to investigate the aerodynamics of a racing yacht using 1 billion computational cells. Breakingthis barrier demonstrates our conviction for high-performance scientific computing. As computationalresources increase and engineering simulations becomelarger and more complex, we will continue to ensure thatour solvers scale appropriately. Moreover, our forwarddeployment of HPC technology is not limited to solvers. The complexity of today’s models and massive amounts of results data require more-scalable solutions forpreparing models and interpreting results as well.

ANSYS Workbench Framework The ANSYS Workbench 2.0 platform is a powerful multi-

domain simulation environment that harnesses the corephysics from ANSYS; enables their interoperability; andprovides common tools for interfacing with CAD, repairinggeometry, creating meshes and post-processing results.Instrumental to the successful integration of this unparal-leled breadth of technology is a “well-architected,” openand extendable software framework.

The ANSYS Workbench framework is designed to provide common services for engineering simulation

applications — data management, parameterization, scriptingand graphics, among others. Release 12.0 relies heavily onthe framework’s data management and parameterizationservices to integrate existing applications into the ANSYSWorkbench environment, where they have become highlyinteroperable. Over subsequent releases, these applicationswill leverage the framework’s graphical toolkit to establish aconsistent user interface and further blend the variousapplications integrated into the platform. At the onset ofdeveloping ANSYS Workbench 2.0, we identified scriptingand journaling as fundamental requirements of the newarchitecture. As such, a top-level scripting engine has beenthoughtfully designed and lays the groundwork for futureANSYS Workbench customization and batch processing.Looking beyond release 12.0, all these services will be further refined and will fuel rapid add-in development and a further expansion of capabilities. Over time, ANSYS customers and partners will leverage the framework’s open architecture, enlisting its services to create tailoredapplications, and will elevate ANSYS Workbench as an application development platform for the engineering simulation community.

Simulation Process and Data Management ANSYS Workbench 2.0 is an environment in which a

single analyst creates and executes one or more steps of anengineering simulation workflow. ANSYS EngineeringKnowledge Manager (EKM) extends ANSYS Workbench byproviding the tools to manage the work of a group of analysts and myriad simulation workflows. This includes system-level services to manage and foster collaborationon thousands of models, terabytes of results, hundreds ofdefined processes and huge investments in simulation.

Looking forward, ANSYS believes that managing dataand processes will become integral with engineering simu-lation. Ten years ago, simulation comprised three discreteand sequential phases: pre-processing, solving and post-processing. With the evolution of ANSYS Workbench, wenow look at engineering simulation as a continuous workflow intertwining these steps. In the same way,process and data management will become intertwined



with simulation, expanding its role and aligning it withbusiness processes such as product lifecycle and supplychain management.

Electromechanical System SimulationThe ANSYS acquisition of Ansoft anticipates a trend in

the realm of engineering and design: The mechanical, elec-trical and software engineering worlds will rapidly converge.Several years ago, the synchronization of these worlds was coined “mechatronics,” and, today, the combineddisciplines are responsible for engineering the electro-mechanical systems found in everything from washingmachines to airplanes. A simple examination of the auto-motive industry reveals that the more recent and excitingadvancements have relied on mechatronics. So, at a timewhen greeting cards and tennis shoes contain micro-processors and sensors, mechatronics is not just forhigh-end cars and appliances; rather it is the key tounleashing innovation in every industry.

For many years, electrical and mechanical engineeringteams have increasingly relied on simulation to accelerateinnovation, but each camp has adopted simulation toolsthat were not fully capable of addressing the needs of theother — until now. As the separation between the electronicand mechanical worlds becomes increasingly blurred,

ANSYS has extended its range of simulation technology by incorporating Ansoft’s world-class product portfolio.Standardizing on ANSYS Workbench for Simulation DrivenProduct Development means establishing a common platform on which to further develop both mechanical andelectronic components and analyze the behavior of thecombined systems. Driving innovation with mechatronicswill require a comprehensive electromechanical simulationenvironment developed by a leader in both mechanical andelectronic simulation software.

The Future Begins NowWith its advancements in individual physics, high-

performance computing, multidomain simulation, meshing,and key enabling technologies such as simulation workflowand data management, release 12.0 clearly delivers on theANSYS vision for Simulation Driven Product Development.But even though we have come a long way with the adventof ANSYS 12.0, there is still an exciting journey ahead.Standing on the strong foundation of all that ANSYS haslearned and developed in almost 40 years of leadership inengineering simulation, we see many new opportunities onthe horizon that will extend the reach of how customers useour technology. The ANSYS vision and strategy continue toset our bearings, and we continue to invest in pioneeringnew frontiers of the industry. And most important is that weremain committed to enabling customers to use simulationto develop innovative products that perform better, costless and are brought to market faster. ■

This article was written through contributions from Todd McDevitt of ANSYS, Inc.

As mechanical and electrical engineering worlds converge, the combination of ANSYS and Ansoft technologies will allow engineers to analyze the behavior of combined systems. © iS

tock

phot

o.co

m/w

eich

eltfi

lm,©

iSto

ckph

oto.

com

/med

uedi

k


ANSYS Advantage • Volume III, Issue 1, 2009

AUTOMOTIVE

www.ansys.com 31

Turbochargers increase the powerand boost the fuel efficiency of internalcombustion engines, but engineeringteams find they pose unique designchallenges. For example, because theturbine is driven by the engine’s ownhot exhaust gases, components mustwithstand widely varying thermalstresses as temperatures cyclebetween 120 and 1,050 degrees Celsiusfor engine speed variations relating toidle, acceleration and braking.

In particular, components such asthe cast-iron housing that directs hotgases into the turbine are subject tothermomechanical fatigue cracking —a problem that often is not discovereduntil parts fail in qualification tests. Toreplicate four to five years of severethermal shock loading — far greaterthan parts would experience in normaloperation — engineers perform roundsof tests that each can be very expen-sive and take weeks to complete.Several of these rounds generally mustbe performed before arriving at a work-able design that passes scrutiny. Manystress intensity factor formulas areavailable in handbooks for predictingfatigue crack growth with simplified 2-Dgeometries; typically, though, these formulas are not applicable forcomplex part geometries under elastic–plastic conditions in high-temperature

environmentswith multi-axialloading. As a result,many part designsare based on modifyingprevious geometries, trial-and-error testing cycles and, inmany cases, “crystal ball” best-guesspredictions based partly on conjectureand simplified assumptions.

Honeywell Turbo Technologiesovercomes these limitations by usingANSYS Mechanical software togetherwith the ANSYS Parametric DesignLanguage (APDL) scripting tool to calculate the probability of a crack initiating as well as its most likelygrowth rate, length and 3-D path. Predicting crack fractures in this man-ner at the early stages of componentdevelopment enables engineers tooptimize designs upfront and helpavoid qualification test failures. Conversely, the analysis gives engi-neers information on the presence ofsmall benign cracks that do not lead toloss of component functionality (forexample, gas leakage or turbine wheelrub) and can, therefore, be ignored.

For this application, J-integralanalysis capabilities in ANSYS 12.0provide a robust solution to predictcrack behavior at high temperatures.The J-integral is a path-independent

fracture mechanics parameter that calculates energy release rate and intensity of deformation at the crackfront for linear and nonlinear materialbehaviors. The J-integral approach generally works best with hexahedralmeshes for the highest possible accuracy. But representing the entirestructure with a hex mesh is atremendous drain on computationalresources. So in this case, HoneywellTurbo engineers used two separatemeshing techniques: hexahedralelements for representing the instanta-neous crack front (a cylindrical volumearound the crack front called the cracktube) and tetrahedral elements for theremaining part volume.

Connectivity between the two different mesh patterns is assured withANSYS transition elements. The size of

Predicting 3-D FatigueCracks without a Crystal BallANSYS tools quickly predict 3-D thermomechanical fatigue cracking in turbocharger components.By Shailendra Bist, Senior Engineer, and Ragupathy Kannusamy, Principal Engineer, Structures and Fatigue Group, Honeywell Turbo Technologies, California, U.S.A.

Honeywell Turbo Technologies produces nearly 9 million turbochargers annually for the automotive industry. Because turbochargersundergo wide thermal swings, they are subjectto thermomechanical fatigue cracking.


AUTOMOTIVE

the 3-D crack tube depends on the volume of the crack path’s plastic zoneand is based on the number of rings of elements and the number of contours to be used in calculating J-integral values using the ANSYSCINT command. The number of element rings and contours shouldbe high enough to maintain pathindependence and accuracy of energyrelease rate.

In this way, ANSYS software calculates J-integral values at eachincrement of crack propagation alongseveral user-defined virtual crackextension directions. The crack featureis updated in a third-party CAD code ateach increment, then imported intoANSYS Mechanical software where it is

meshed, solved and post-processed.The cycle continues until a target criterion is reached. All processes are integrated and controlled using in-house APDL scripts. By leveragingimproved fracture mechanics capabil-ities in ANSYS 12.0 for calculatingJ-integrals, the method provides anew approach to model and simulatearbitrary 3-D crack growth and tocompute mixed mode stress intensityfactors along the crack front within thesimulation software.

This method requires calculationsto be performed iteratively for thou-sands of crack-growth cycles — aprohibitively labor-intensive and time-consuming task if performed manuallybut one well-suited to the automation

capabilities of the APDL scripting tool.Along with techniques such as submodeling and load blocks for moreefficient solution processing, suchautomation radically increases thespeed of performing these iterativecalculations.

Honeywell Turbo analyzed a testcase using this method to predictgrowth behavior of paths in a cruciformspecimen under uniaxial and biaxialloading. The uniaxial load case showsprominent crack turning while the biaxial case shows near planar growth.The results obtained validate theapproach. The team completed furtherruns to validate crack growth rates that show promising results.

Using this automated ANSYSfatigue crack prediction process hasthe potential to increase engineeringproductivity significantly, with crackgrowth analysis time reduced by morethan 90 percent compared to manualmethods. This speedup has significantvalue, since Honeywell Turbo engi-neers must analyze as many as 400designs annually, and demands willlikely increase in the coming years asturbochargers are implemented on agrowing number of vehicle modelsaround the world. In this way, tech-nology from ANSYS is playing a criticalrole in enabling the turbocharger company to strengthen its leadershipposition in this competitive industrysector. ■

Hexahedral elements represent the expected path of 3-D crackpropagation (called the crack tube), and less-complex tetrahedralelements are used for the remaining volume of the part.

Crack front withvirtual crackextension directions

X5

X9

X4

X8

X3

X7

X2

X6

X1

The 3-D crack growth direction determining the propagationpath is based on a virtual extension direction angle in whichmaximum energy is released.

Crack pathfrom test

Crack pathfrom testresult

Crack path directions in cruciform specimens under uniaxial loading (top) and biaxial loading (bottom)

Electromagnetic fields are used more and more inadvanced medical applications such as magnetic reso-nance imaging (MRI), implants and hyperthermia treatment.As the state of the art advances, devices are becomingmore complex and simulation more indispensable in theproduct design phase. With simulation, a designer canstudy device functionality and address safety concernswithout exposing a patient to harm or otherwise.

In the design of an open MRI system, for example, thedetails of the radio-frequency (RF) coils, a human bodymodel, and the large volume of the entire examination roommust all be included in an electromagnetic simulation modelto determine the resulting field accurately. The finite elementmethod found in HFSS (High-Frequency Structure Simulator)software, an electromagnetic field simulation tool new to theANSYS portfolio, is well suited for this purpose as it usessmall mesh elements where refinement is needed and largermesh elements elsewhere. The human body model availablethrough ANSYS comprises 300 objects that, detailed downto the millimeter, represent organs, bones and muscles.

Frequency-dependent electromagnetic material parametersare also included in the model.

The RF coil design requires optimization for appropriateimage quality: The coils need to resonate at 42.6 MHz for a1 tesla system and produce a rotating magnetic field that isstrong and smooth in the region of interest but minimizesundesired field components. If the field varies strongly,some parts of the image will appear to be overexposed,while other areas will remain too dark, both of which aredetrimental for contrast. Once the specifications related toimage quality are satisfied, the designer needs to make surethat specific absorption rate (SAR) safety regulations aremet. SAR is a measure of how much RF power is absorbedby, and thus creates heat in, the body. When limits are exceeded in any part of the body, the patient can experience discomfort and tissue damage.

Model of the open MRI system, which combines anMRI model generated by Philips Healthcare with theANSYS human body model

Electromagneticsin MedicineElectromagnetic and thermal simulations find use in medical applications.By Martin Vogel, Senior Member of the Technical Staff, Ansoft LLC


HEALTHCARE

© iS

tock

phot

o.co

m/ k

wan

isik


Simulation results from the open MRI case indicate hotspots under the armpits, a result that agrees with practicalexperience. Analysis also indicates resonant hot spots onthe legs, even though they are not directly under the coils inthe model. Given the frequency and material parameters ofthe body, the expected wavelength in the body is a little lessthan 1 meter, and resonances such as these are indeedpossible. The SAR is not quite symmetric; this is expected,as the excitations are not symmetric either. Entire scan protocols can be simulated in the software by moving thebody automatically through the scanner.

Another medical application in which human comfort isimportant is the design of wireless implants. Implants thatrequire directly wired power supplies can be uncomfortablefor the patient. But wireless power supplies that use low-frequency coupling require a bulky transmitter, reducing patient freedom. Wireless solutions that use higherfrequencies can potentially provide both comfort and freedom.One design challenge is to transmit maximum power to theimplant while also satisfying radiation and SAR regulations.

Simulations of wireless implants provide details thatotherwise are not easily obtained for several transmitter andreceiver locations. One important finding is that, in order toget accurate results, interior body components such asorgans, bones and fat tissue must be included in the simu-lation model. If not, the results can easily be off by morethan a factor two.

One final medical simulation example models an RFphased-array applicator for hyperthermia cancer treat-ments. In hyperthermia, a tumor is heated with RF powerand held at an elevated temperature for some time, such as15 minutes to 60 minutes. This weakens the tumor, whichhelps to make other therapies more effective. The challengeis to concentrate the hot spot in the tumor while minimallyaffecting healthy tissue.

The applicator consists of several dipole antennas printed on the surface of a cylindrical plastic shell thatmounts around the patient’s leg, the location of the tumorfor this case. The chosen frequency for the device, 138 MHz, is a compromise between hot spot size andpenetration depth. A higher frequency can provide a smallerhot spot, but it would be harder to penetrate deep into the tissue. Water cooling prevents skin heating during the procedure and is accounted for in the simulation model. A realistic tumor object, created using MRI data for thispatient, is inserted into the leg of the human body model.

By using the electromagnetic simulation capabilities inHFSS software, the applicator and its settings are optimizedto focus the hot spot in the tumor. Next, the power-lossinformation for every mesh element in the model is transferred automatically to the thermal simulation tool, ePhysics. The ePhysics product then computes temperature distribution as a function of time, taking

Model of a hyperthermia applicator and leg with tumor; in theimage, some applicator and water cooling system componentshave been removed for clarity. The green object is the tumor.Applicator design and tumor geometry provided by Duke University.

Sample of specific absorption rate that results on the bodywhen using the open MRI system, as simulated using theHFSS electromagnetic field simulation tool

The electric field (magnitude) that results when using a receiverimplanted in epidural space in conjunction with a wireless transmitter placed behind the back; the image shows a horizontal cross section of the torso and arms of a person,standing, using a wireless implant.

HEALTHCARE

Comparison of simulated and measured temperaturesin the tumor for a hyperthermia treatment caseMeasured results provided by Duke University.

6

5

4

3

2

1

00 5 10 15 20 25

XY Plot 2

Time [min]

30W

20W

Measurement

Simulation

15W 10W20W

17W 15W

Tem

pera

tire

Incr

ease

[deg

C]


HEALTHCARE

into account thermal material properties as well as water cooling, blood perfusion, air convection and thermal radiation.

Blood perfusion refers to blood flow through capillaryvessels in muscles and organs. This flow removes excessheat and must be included in hyperthermia simulations. Toinclude all the details of the capillary blood vessels would betoo complicated; therefore, a simpler model is used. It isassumed that a certain amount of blood enters a volume oftissue at a specified rate; it is also assumed that bloodassumes the tissue’s temperature and leaves the volume,taking a corresponding amount of heat with it. Perfusion forseveral tissue types can be found in literature [1] and isquantified in the simulation model as a temperature-dependent negative heat source. Overall, the simulationresults proved to be very sensitive to blood perfusion.

The input power to the applicator is varied over time forboth simulation and experiment. The outer layer of thetumor is assumed to have a higher perfusion rate than thecore, as is consistent with literature. Deviations betweensimulation results and experimental data in the early stagesare likely due to the fact that initial thermal conditions in thesimulation did not exactly match those in the experiment.

With these simulations, modeling software progressesbeyond device design into treatment planning. Finding theproper operating conditions through simulation relieves thepatient from invasive experimental procedures. To efficientlyoptimize conditions for a variety of patients in a hospitalenvironment, engineers must improve methods to translateMRI scan data into personalized human body models thatare ready for simulation.

Electromagnetic and thermal simulations are wellunderstood and used regularly for the design of medicalequipment and procedures. The next breakthrough isexpected when personalized human body models can begenerated efficiently and doctors use simulation for treatment planning. ■

The author wishes to acknowledge Philips Healthcare in the Netherlandsfor its work on MRI and Duke University in the United States for its workon hyperthermia.

References[1] Erdmann, B; Lang, J; and Seebass, M. “Optimization of Temperature

Distributions for Regional Hyperthermia Based on a Nonlinear HeatTransfer Model.” Ann. N. Y. Acad. Sci., Vol. 858, September 11, 1998,pp. 36–46.


ELECTRONICS

Keeping Coolin the FieldA communications systems company gains millions of dollars by using thermal simulation to bring tactical radios to market faster.By Patrick Weber, Mechanical Engineer, Datron World Communications, Inc., California, U.S.A.

The communications systemsdesigned and built by Datron WorldCommunications, Inc. present majorthermal design challenges. The com-pany’s radios travel with today’s warfighters around the world in helicoptersand Humvees® as well as on foot. Thedevices are designed to survive in awide variety of environments, rangingfrom a sandstorm in the desert to amountain blizzard. These systems dissipate substantial amounts of heatyet must be sealed to the outside environment to prevent damage tointernal components — for example, if the radio falls into a creek, it still must work — and to prevent electro-magnetic interference.

Datron mechanical engineers facethe challenge of providing cooling management within a completelysealed radio cabinet in up to 60-degreeCelsius (C) ambient temperatures.Communication systems are designedwith heat sinks external to the cabinetthat use forced-air conventional cooling. Components with the highestlevels of power dissipation are mounted internally near those fins. Radios con-tain printed circuit boards (PCBs) forthe power supply, radio frequency (RF)filter, CPU and audio functions. ThesePCBs generate substantial amounts ofheat. In addition to keeping junctiontemperatures of board componentswithin specifications, Datron engineers

need to limit — for safety reasons —external temperature of the heat sinkto 15 degrees C above ambient.

Historically, thermal managementdesign was based on engineeringexperience and instinct. In order tounderstand the cause of any thermalproblems, engineers had to test a widerange of prospective solutions andcorresponding prototypes. The cost of developing, building and testingprototypes was high. But the resultingdelays in bringing each new product to market were even more costly.Datron engineers have improved thethermal design process by using thermal simulation.

The company now practices Simu-lation Driven Product Developmentand begins the thermal modeling earlyin the design process. Radios typicallygenerate 125 watts output and dissi-pate approximately 220 watts inside a 15-inch wide by 15-inch deep by 5.5-inch high box. Initial models aredeveloped based on very limited infor-mation, such as the size of the chassis,the RF output power and the expectedefficiency of the radio. Engineers selectprimitive objects, such as cubes, asbuilding blocks and parametricallyassign dimensions and material properties. Surface properties areassigned to the outside surface of the enclosure to represent the olivepaint that is typically used on the finalproduct. In the early design stages, the

CAD model of theradio chassis

Original radio design with ferrite core filters shows hot spots.

Fan Tray


ELECTRONICS

internal components are approximatedby a single component that dissipatesthe total amount of heat in the radio.

As the design progresses, moredetailed information on the PCBsbecomes available. Mechanical engi-neers model the different PCBs andcomponents within the chassis andevaluate the thermal performance.ANSYS Icepak macros are used toquickly generate models of standardpackages. Other macros are used togenerate heat fins from parametersincluding the number of fins, fin widthand fin spacing. The design team limitsthe model to approximately 1 millioncells by meshing smaller boxes aroundhot spots at higher densities.

In a recent project, early modelsshowed that junction temperaturesexceeded the typical maximum of 125to 150 degrees C. The original designspecified ferrite core filters that are relatively light but have a very low thermal conductance. Simulation usingthe ANSYS Icepak tool showed thatthe devices heated up the surroundingair to the point of overheating neigh-boring devices. Based on this insight,engineers replaced the ferrite filterswith aircoil filters that have a higherthermal conductance. This designchange was the key to significantlyreducing junction temperatures of highpower-dissipation components. Oncea working design was obtained, theengineers used parametric modeling

to optimize thermal management and acoustics.

Using this approach, Datron engi-neers improved the performance of thesoftware prototype until it met thermalrequirements within the required marginof safety. At that point, they ordered the first thermal hardware prototype. Testing showed that the thermal proto-type closely matched the simulationpredictions and also met all of the thermal design specifications. As aresult, no additional hardware proto-types needed to be built, and the radiowas brought to market substantiallyearlier than if the company’s originalbuild and test method had been used.

In other recent thermal design projects at Datron, ANSYS Icepak simulations showed that several powertransistors exceeded the junction temperature specification. By knowingthis early in the design process, it was

The Natural Convection ChallengeOne of the biggest challenges Datron engineers face is simulating natural

convection. This is inherently difficult and expensive to simulate because thebuoyancy forces are constantly changing. The Datron team developed a typical natural convection problem and compared the ability of all the leading thermal simulation tools to solve it. Several of the software packagestook 24 hours or more, while ANSYS Icepak software solved the problem inonly 20 minutes. Datron engineers liked the nonconformal meshing tools in the ANSYS Icepak product that make it possible to separately mesh — usually with a finer mesh than the rest of the model — critical areas within thesystem, such as high-dissipation components. Such a process increases theaccuracy in the critical areas without unnecessarily increasing computationaltime requirements.

ANSYS Icepak model shows the speed of the air from the fans along with temperaturecontours on the chassis. Blue indicates cooler temperature.

New design with aircoil filters shows that temperatures are reduced to acceptable levels.(The filter temperatures in degrees C have gone from the 200s to the 90s.)

possible to substitute other suitablecomponents with lower thermal resist-ances. If this problem had not beendiscovered until after the detaileddesign process, it would have requireda considerable amount of time andwork to correct. In addition, with thischange, engineers discovered thatthey could decrease the number of finsrequired, which provided more roomon the rear panel of the enclosure andmade it possible to reduce the overallsize and weight of the radio.

For Datron, simulation makes itpossible to validate and optimizedesigns much earlier in the develop-ment process, saving large amounts oftime and money. Engineering simu-lation has substantially reduced thetime required to bring new, improvedcommunications technology to themarketplace, and this can translateinto millions of dollars in revenue. ■

DesigningAgainst the WindSimulation helps develop screen enclosures that can better withstand hurricane-force winds.By Steve Sincere, President, Optimization Analysis Associates, Inc., Florida, U.S.A.

One of the most popular residential structures in Florida is the screen enclosure (or screen room), consisting of an extruded aluminum frame covered withscreen. These structures are primarily intended to keepdebris and insects out of swimming pools and toincrease living space to include an outdoor environment.Even so, they must be designed to resist hurricane-forcewinds ranging from 100 mph inland to 150 mph in coastalareas, depending on building code requirements.

Recent hurricanes have revealed shortcomings in thesedesigns. Most are developed by contractors or enclosure fabricators based on oversimplified analytical assumptions.Components typically are sized without regard to the Aluminum Design Manual (ADM), Specifications and Guidelines for Aluminum Structures as specified by the Florida Building Code (FBC). Moreover, fasteners and fastening methods typically are selected for ease offabrication or accepted convention rather than suitability forthe high wind loads.

Using ANSYS Mechanical software, OptimizationAnalysis Associates, Inc. — an engineering consulting firmspecializing in mechanical analysis and design simulation —

performed analytical studies of existing screen enclosuredesigns using FBC wind loads. The company found thatthe simplified methods failed to accurately calculate forcesand moments. Thus, the complex interactions amongstructural members were not adequately accounted for inthe designs.

Finite element analysis (FEA) provides the most accurate method of determining such loads and inter-actions. Most engineers in the screen enclosure industrydo not have a background in FEA, however, and those withsuch expertise often forgo these studies due to time andcost constraints. The answer is an automated FEA-basedscreen enclosure design tool — one that is fast, is accurateand requires no FEA skills.

A perfect platform for this task is ANSYS ParametricDesign Language (APDL) — a scripting language forautomating common analysis tasks or even building models in terms of user-specified input variables. This adap-tive software architecture enabled Optimization AnalysisAssociates to create a web-based solution with a graphicalinterface through which screen enclosure designs could beconveniently specified and automatically evaluated.

© iS

tock

phot

o.co

m/a

kurt

z


BUILT ENVIRONMENT

Photo courtesy Richard Graulich/The Palm Beach Post.


BUILT ENVIRONMENT

APDL is used to automatically create, load and solve a full-frame model of a screenenclosure from parameters entered by the user describing the structure.

Users are required to enter only minimal input data,including basic geometry information of the frame, windload criteria, a sketch of the plan view (to provide x and ycoordinates for each corner), wall height, roof style, densityof structural members (number of columns to be used on awall, for instance) and sizes of the structural members.From this input data, three APDL macros then auto-matically perform an analysis, check results against guidelines and generate layout drawings — all completed inless than three minutes and requiring no user intervention.

The first APDL macro reads in the data to create, loadand solve the full frame model. Beam elements representthe structural members, which are coupled in the model to simulate hinged or rigid connections as necessaryaccording to the type of connections used. Shell elementsrepresent the screen in a proprietary method that deter-mines the load distribution on structural members.Solutions are obtained for the eight wind-load cases prescribed by the FBC.

A second macro performs all required checks definedby ADM criteria. This complicated process begins byaccessing external files containing section properties,material characteristics and other parameters associatedwith extrusions used in the design. Then a series ofnested APDL do-loops performs the ADM calculations for all nodes on every structuralmember for each load case. The macro entersthis data into arrays and sorts through them todetermine the limiting members. The limitingmembers are written to a summary report textfile, which is accessed by the web-based interface. The report provides a simple pass/fail outputwith percent overstress values (or interaction ratios).

If the user has a passing design, a third APDL macroproduces a layout drawing of the structure. This macrotakes advantage of the graphical capabilities of ANSYSMechanical software in generating annotation for dimen-sions and labels on screen enclosure 2-D layout drawings.

If the user does not have a passing design (or if the design istoo conservative), parameters may be revised and anotheriteration may be performed.

Optimization Analysis Associates has written programsfor more specialized work as well. A version of the model-building macro allows experienced users of software fromANSYS to create customized structures with nontypicalshapes and/or nonstandard bracing configurations. Anothermacro uses the ADM data to produce color contour plots ofinteraction ratios, a calculated value of allowable stress rationot existing in the results file. Locations of failure to meetthe ADM criteria give a quick visual indication of problemareas. In addition, these allowable stress ratio plots can beanimated with a modified version of the animation macroANCNTR.MAC and overlaid on 3-D models showingdeformed structural geometry.

One final specialized macro provides a cost estimate forthe construction of the design. This macro interrogates themodel to determine the length of each extrusion requiredalong with the square footage of screen and number of fasteners, brackets, etc. It accesses an external price list filefor each item, as well as factors for items such as labor,scrap, overhead and profit to determine the total cost. Thefinal output includes a complete parts list and a breakdownof all cost components.

The automation of the modeling and simulation-based evaluation using APDL provides a fast, easy-to-useand extremely accurate method of structural frame designs.The screen enclosure industry now has the potential toproduce hurricane-resistant structures, to significantlyimprove design productivity, and to improve cost estimatingand profit margins of contractors and fabricators who useengineering simulation for their designs. ■

Color contour plots of interaction ratios show locations’ potential wind-force failure in red.


ENVIRONMENT

Contours of solid particle concentration:During the suction phase, the solids werefound to become more concentrated alongthe bottom of the vessel, as shown by thered color in the early suction.

the top to the bottom of the vessel.This model solves a separate set ofNavier–Stokes equations for the fluidand solid phases. It accounts for thecoupling between and within thephases using exchange coefficients,the most important of which is for thefluid–solid interaction. The results madeit possible to determine whether themixing criteria were met under givenoperating conditions.

Each vessel in the plant has a different mixing criterion; however,most simply require that the solidsremain in suspension and are mixedwell enough for accurate sampling andtransfer to the next step of the vitrifica-tion process. Since pulse jet mixing is

StabilizingNuclear Waste Fluid simulation solidifies its role in the radioactive waste treatment process.By Brigette Rosendall, Principal Engineer, Bechtel National, Inc., California, U.S.A.

The nuclear site at Hanford, Washington, houses approximately 60 percent of America’s radioactivewaste. Near the Columbia River, thesite stores waste in 177 undergroundtanks as a combination of liquid,sludge and slurry. A vast complex oftreatment facilities is being constructedto convert this waste into a stableglass-like material using a technologyknown as vitrification, which involvesmixing the waste processed in thesevessels with hot glass formers such as rutile (TiO2) or silica. The mixture is then poured into steel canisters and cooled to solidify for permanentstorage. One of the major challengesin this process is keeping the solids inthe waste in suspension during itstime in the holding vessels before theseparation and processing stages.

Avoiding contact of any mechanicalcomponents with the slurry beingmixed during holding was crucial andled Bechtel National engineers workingon the project to select fluidic pulse jetmixers (PJMs). The action of the PJMsis carefully controlled by compressingair inside them to drive the slurry intothe vessel to create the mixing action. Only 80 percent of the slurry volumethat is suctioned up into each PJM isexpelled out of the mixers, which pre-vents air from escaping into the vessel.At that point, the compressed air isvented and a vacuum is applied to refillthe mixers. PJMs thus provide mixing

while keeping all mechanical compo-nents well away from radioactivematerials.

Because there had been little previous experience with PJMs in this mixing environment, it was critical that the engineering team be able toaccurately predict the ability of theunits to provide sufficient mixing foreach of the different vessels in which the wastes will be treated. Within thewaste treatment plant, each of the mixing vessels has substantially different geometries and processingrequirements. In addition, there is considerable variation in the character-istics of the mixture of fluid andparticles that will be processed in thedifferent tanks due to separation andconcentration of the radioactive com-ponents. The mixing performance of thePJMs is a function of the geometry ofthe vessel, number of PJMs per vessel,particle size, fluid characteristics, cycletime and other variables. It was impor-tant to validate the ability of the PJMsto keep the particles in suspension ineach tank.

To simulate the pulse jet mixingprocess, Bechtel engineers used theANSYS FLUENT fluid flow simulationpackage because of the software’sunique capability depth in modelingmultiphase mixing. The Euleriangranular multiphase model in ANSYSFLUENT software made it possible topredict the distribution of solids from Pulse jet mixer design

A B C D

PulseJet

Mixers

To Vent System

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air

Proc

ess

Air


At the end of the drive phase, higher concentrations arepredicted at the top boundary of the fluid domain whileconcentrations were reduced at the bottom as the solidswere pushed away from the jet nozzle exits.

Comparison of fluid flow predictions and experimental results for solid particle volumefractions averaged over tank radius and mixing cycle for a 140-inch-high tank

End-suction

Mid-suction

160

140

120

100

80

60

40

20

0

Elev

atio

n (in

)

0 0.02 0.04 0.06 0.08 0.1Volume Fraction

CFDExperiment

ENVIRONMENT

a turbulent process, Bechtel engineerschose ANSYS FLUENT software’s k-epsilon turbulence model based onthe results of a preliminary study. In thisstudy, computational fluid dynamics(CFD) specialists compared the resultsof various turbulence models toexperimental data to determine whichmodel was best at predicting thevelocity in scaled hydrodynamicstests.

The engineering group controlledtime-varying boundary conditions by auser-defined function that prescribed

the time-dependent velocities of eachjet and tracked the solids concentra-tion flowing through the nozzles and atthe top of the domain. This eliminatedthe need to track the free surfacesinside the PJMs and at the fluid–airinterfaces inside the mixing vessels,greatly simplifying the models.

The Bechtel team could performonly very limited physical testing due tothe high cost of building and testing the vessels and mixers. The companycommissioned the construction of a full-scale PJM vessel to perform experimental testing at Battelle Pacific Northwest National Laboratory. Fluidflow predictions of concentration andvelocity were then compared to themeasured data. The results showed thatthe ANSYS FLUENT simulations slightlyunderpredicted the solid-phase volume fraction, except at the higher elevationsin the tank. This difference was not

significant compared to the cyclic varia-tions in the concentration. At higherelevations, there were more significantdifferences between the experiment and simulation, with the simulation

predicting more uniform mixing than theexperiments demonstrated.

Even though the ANSYS FLUENTresults demonstrate slightly better mixing than the physical experiments,the results were close enough to giveBechtel confidence in the ability of thefluid flow model to provide pass–failjudgments in rating the performancesof the PJMs. Bechtel uses ANSYS FLUENT technology to model the many

different vessel designs and to determine whether or not PJMs couldprovide adequate mixing for each con-figuration. The use of fluid dynamics inthis application can potentially save asignificant amount of time and moneythat otherwise would be spent on additional physical testing prior tobeginning actual waste processing. ■

See also:

www.bechtel.waste2glass.com

www.hanford.gov


OPTIMIZATION

Topology Optimizationand Casting:A Perfect CombinationUsing topology optimization and structural simulation helps a casting company develop better products faster.By Thorsten Schmidt, Technical Director, Heidenreich & Harbeck AG, Moelln, Germany

and Boris Lauber, Application Engineer, FE-DESIGN GmbH, Karlsruhe, Germany

Engineers usually need to ensure both functionality andzero defects during component production. This often canbe achieved by simulating production processes and oper-ating conditions in the virtual world. Development teams inthe machine tool industry need not only to prove themechanical strength of components but also to take intoaccount rigidity and cost.

Heidenreich & Harbeck AG in Germany was establishedin 1927 as a foundry for cast iron components. Today, thecompany’s range of capabilities has expanded to includemodern machine tools for finishing large, quality castingsthat have high accuracy requirements. The company’s in-house development department assists customers’designers and develops castings of complex machinestructures according to customers’ specifications.

The comprehensive software portfolio at Heidenreich &Harbeck contains several 3-D CAD tools, process simula-tion software for casting processes and numerical control(NC) machining, a sophisticated cost calculation tool basedon 3-D CAD models, and project-planning software. Inaddition, Heidenreich & Harbeck uses ANSYS Professionalsoftware for the simulation of mechanical properties. To provide optimal design proposals to accelerate thedevelopment of large castings, the company obtained

TOSCA® Structure software from German-basedFE-DESIGN GmbH. This product interfaces with ANSYSProfessional software.

In the past, the engineering team designed structuralcomponents with primary consideration to manufacturingrestrictions. But structural analysis of these componentdesigns often revealed weak points, especially for parts witha large number of load cases. Engineers then had to perform time-consuming iterations with alternating modifi-cations of CAD design and structural analysis in order tofulfill customer requirements.

Currently, the Heidenreich & Harbeck developmentprocess starts with modeling the design space, which usually is easy to define. Engineers import the design spacegeometries into ANSYS Professional software and thengenerate meshes. Boundary and loading conditions areapplied. Groups of volume elements that are required foroptimization are defined in ANSYS Professional technologyas components. The engineering team exports solver inputfiles from the ANSYS Professional tool and imports themdirectly into TOSCA Structure software with the latter’s user interface. Using this wizard-based technology, theoptimization setup can be executed with a few mouse clicksby re-using group definitions from ANSYS Professional to

Topology optimization of support arm for paper unwinderCourtesy Bielematik.

Model of original design,without optimization

Design space, as providedby customer with loadingdefinitions defined

Four GuidingWagons To Be Mounted

EccentricLoad

Meshed, optimized structure beforeincluding casting restrictions in theiterative design process


OPTIMIZATION

CAD System

Generation of geometry for design space

Structural Pre-Processing

• FEA model• Load cases• Components

ANSYS Mechanical

Optimization Pre-Processing

Optimization wizard

Structural Post-Processing

• Validate optimization results• Final evaluation

ANSYS Mechanical

Structural Pre-Processing

FEA model ofredesigned geometry

ANSYS Mechanical

CAD System

Redesign using extensive manufacturing knowledge

TOSCA Structure

TOSCA Structure

Batch Optimization Process

ANSYS Mechanical

Scheme of topology optimization using TOSCA Structure based on solver from ANSYS

define the design area, frozen areas, evaluation areas fordesign responses, and areas for the application of manufac-turing constraints. The optimization procedure is carried outin a batch process. TOSCA Structure software iterativelylaunches the ANSYS Professional solver for the analysis ofthe design space model and then launches the optimizationmodule that evaluates results and changes material proper-ties. Users who want to remain in the familiar ANSYSproduct environment may transfer the results produced bythe TOSCA Structure product back to ANSYS Professionalfor post-processing using a file containing the materialproperty values for the finalized optimization.

Heidenreich & Harbeck uses an optional module from FE-DESIGN called TOSCA Smooth to convert the optimization results into IGES or STL files containingisosurfaces and cutting splines based on the normalizedmaterial distribution.

For the design of castings, consideration of manufacturing constraints plays a very important role. It isessential to take into account demolding constraints forparts with low-cost restrictions. For a part that is loaded byan eccentric force leading to a torsional loading condition, anon-restricted optimization will generate a hollow sectionthat would lead to high torsional rigidity. By applying ademolding constraint in the TOSCA Structure tool, the engineer can obtain a design proposal that is less rigid but has no undercuts and cavities and may, therefore, be

manufactured without the use of cost-intensive cores in thesand mold. An automatic or user-defined parting plane maybe specified. For the design of stiffening ribs, the castingconstraints may be coupled with a wall thickness constraint.

A customer provided Heidenreich & Harbeck with thedesign space of a support arm for a large paper rollunwinder loaded with an eccentric force. The design with nocasting restrictions led to a hollow profile without access-ibility for fastening screws. A second optimization withcasting restrictions resulted in a two-beam structure. Thefinal design combined the benefits of both proposals(accessibility for screws along with hollow profile for cableand tube-laying, which the customer added to the specifi-cations after he became aware of the first design proposal).Due to topology, optimization rigidity was increased by 25percent, and weight was decreased 34 percent comparedwith the former two-piece design.

In another project involving a vertical lathe housing, thecustomer delivered two-dimensional sketches with theexpectation of final pattern drawings within only threeweeks. Using TOSCA Structure software, the rigidityrequirements were fulfilled with minimal material consump-tion, and time-consuming design iterations were avoided.This reduced development lead time by approximately 50 percent. ■

Visit www.huhag.de and www.fe-design.de for further information.

ANSYS Professional simulation results, whichare evaluated during the optimization process

Simulation of the casting process Final component design

Fires onboard ships are notuncommon and pose a danger to bothcrew and equipment. It isvital to develop effectivemethods to extinguishthese fires. At the sametime, international agree-ments such as the MontrealProtocol on Substances thatDeplete the Ozone Layer havebeen signed. These agreementslimit the use of firefighting agentssuch as Halon that, though effective,come with a high environmental price.In order to find an alternative to Halon,the U.K. Ministry of Defence (MOD)completed a comprehensive researchprogram that looked at alternative firesuppression technologies for use onRoyal Navy vessels. The work led to thedevelopment of a low-pressure watermist system, or fine water spray (FWS).This new FWS system combines salt water from a ship’s high-pressuresalt water (HPSW) system, which typi-cally operates at a pressure of 7 bar,together with a 1-percent-concentrationaqueous film-forming foam (AFFF).

As part of this program, MOD validated and used simulation as a toolto assess the performance of the FWSsystem, with and without additive,when fitted onboard a ship. Thisanalysis decreased the need for expen-sive fire testing for future assessmentsand design of fire control measures.The United Kingdom ANSYS officedeveloped a fluid dynamics modelusing ANSYS CFX software, validatedit blindly against MOD’s full-scaleexperiments, and demonstrated itsapplication to a real vessel.

Because of the complexity of theapplication, the simulation involved alarge number of software models that included existing capabilities,existing models that required somespecial functionality extended throughFORTRANTM, and some models thatwere implemented entirely throughFORTRAN. The simulation modelswere validated against data from alarge-scale experimental rig.

Measurements of the FWS dropletinitial conditions, in air and without fire,were commissioned at South BankUniversity (SBU), London, using high-speed photography. This providedinformation at a specified, small radialdistance from the nozzle, for velocity(predominantly radial) and mass flowfor each of a group of droplet-sized bands, as a function of azimuth and elevation. SBU performed

measurements for two working fluids:water and water with 1 percent by volume AFFF. The university alsomeasured to ascertain whether theadditive affected the terminal speedof a droplet with a given mass. TheSBU measurements were employed inthe initial conditions for the particletransport model.

To determine how the fire becomesextinguished, the combustion modelcalculates the fuel evaporation ratefrom the heat delivered to the fuel bythe fire. The model then predicts whereand how rapidly fuel vapor is burnedand heat is released exothermically. Asthe fire cools after spray initiation andradiation is attenuated by the spray,soot, and gaseous products (as well asthe foam film when that is present), theheat returned to the pool of liquid fuelis diminished and so is the evaporation

Temperature isosurfaces and droplet trajectories before fire extinction is completed in a ship’s machinery space


MARINE

Fighting Firewith SimulationThe U.K. Ministry of Defence uses engineering simulation to find alternatives to ozone-depleting substances for fire suppression.By Michael Edwards and Michael Smerdon, U.K. Ministry of Defence, Bristol, U.K.

Yehuda Sinai and Chris Staples, ANSYS, Inc.


MARINE

rate. If the spray system is appro-priately designed, then extinction isachieved when combustion processceases. Fuel vapor usually vanishes ashort while after the fuel evaporationrate falls to zero.

The MOD and ANSYS researchteams validated the fluids model bycomparing it to data from a MODexperimental rig. The rig was largescale with a volume of 1,080 cubicmeters. Inside the experimental rigthere were mockups of the largeequipment — diesel generator andgas turbine enclosures typically foundwithin a Royal Navy (RN) machineryspace. The FWS comprised 16 GWLoFLowTM K15 nozzles fixed on a 3-meter grid near the ceiling. Bucketsat the floor were used to measurecumulative water delivery. Additionalinstrumentation was added to thespace to enable validation of themodel. Liquid fuel (F-76, which is acommon fuel for shipboard diesels,gas turbines) was provided in one oftwo rectangular trays, having areas of

Simulation model of the rig geometry and temperatureisosurfaces before spray inception

Droplet trajectories and maps of water vapor molefraction after spray inception

3 and 1.5 square meters, respectively.The teams validated the simulationagainst two separate tests: waterspray for the larger tray and waterspray with additive for the smaller tray.

The results of the validation weregenerally encouraging, and the pre-dicted extinction times and method ofextinguishment were reasonably pre-dicted. There were some noticeablediscrepancies, and there was evidencethat building leakage (the effects of

which had been studied in previousresearch by ANSYS) was an importantfactor in this regard. Other influenceson the results of the model were identi-fied: The fuel model used heptanerather than F-76; the coefficient ofrestitution was set at zero for waterdroplets so that when they hit struc-tures they were removed from themodel; and positioning of the mockupstructures, fuel trays and nozzle posi-tions represented a worst case.

The technology from ANSYS that can be applied tofire propagation, fire suppression and smoke manage-ment for ships, airplanes, trains, cars and trucks is alsoused for ventilation and thermal modeling in the builtenvironment industry. These comprehensive multi-physics capabilities, which address safety and comfortconcerns, are frequently used upfront during the designand construction of buildings.

In order to provide information for design improve-ment, design optimization and energy efficiency in thebuilt environment, predicting conditions such as airvelocity, temperature, relative humidity, thermal radiationand contaminants is extremely important. The simulationmust also take into account ventilation, heat loss andsolar radiation effects on the structure walls, roof, floors,windows and doors, as well as the presence and activityof people and equipment in these areas. Simple air flowmodeling assists engineers and architects in quantifyingand simulating the impact of structural and equipmentdesign modifications on the thermal comfort of a space’soccupants.

Engineering solutions from ANSYS provide a cost-effective and accurate means of designing efficientsmoke management and detection systems. The unpar-alleled breadth of solutions across multiple disciplinesprovides the ability to quantify the behavior of materialssubjected to fires or extreme heat and possible structural

deterioration during catastrophic events. These can be analyzed in detail using explicit dynamics and structural modeling. Solutions from ANSYS allow forthe analysis of events ranging from explosions thatencompass blast waves (in the context of homelandsecurity) to deflagrations in combustible mixtures.

— Thierry MarchalIndustry Marketing Director Materials and Consumer Care, ANSYS, Inc.

For more information, visit www.ansys.com/industries/hvac.

Engineering Simulation for the Built Environment

Courtesy SOLVAY S.A.


MARINE

Descriptions of Models Used in the Simulation

Model Implementation Purpose

RANS turbulence modeling (SST) Existing model Determines turbulent transport

Laminar flamelet combustion modeling (Peters)

Extended current model

To include combustion modeling of heptane fuel and evaporated watervapor, with reduced set of species

Soot modeling (Fairweather et al.) Implementednew model

Assesses impact of soot on infraredradiation and visibility

Transient Lagrangian particle transport model

Existing model Assesses the impact of water spray on fire and fuel, with two-way coupling of mass, momentum, convective heat and radiant heat

Multiple droplet size groups Existing model Determines penetration since largerdrops are better at penetrating keyregions directly, small droplets evaporate quickly and can reach key regions by entrainment

Coupled fuel evaporation Implemented new model

Calculates fuel burning rate

Subgrid droplet–congestion interactions

Implemented new model

Estimates direct removal rate of droplets by subgrid congestion

Soot scavenging by water droplets Implementednew model

Determines how scavenging affectsinfrared radiation and visibility; also predicts delivery of scavenged substances to boundaries

Additive effects on water spray and fuel evaporation rate

Implementednew model

Predicts attenuation of radiant heatarriving at pool surface

After completion of the validation,the model was successfully applied toa real machinery space aboard an RNship. MOD is proposing the use ofFWS in its future vessels for fire sup-pression that was validated by theexperiment [2] and this work. ■

References[1] Sinai, Y., Staples, C., Edwards, M., Smerdon,

M., “CFD Modelling of Fire Suppression byWater Mist with CFX Software,” Proc.Interflam 2007, Vol. 1, 2007, pp. 323–333.

[2] Hooper, A., Edwards, M., Glockling, J.,“Development of Low Pressure Fine WaterSpray for the Royal Navy: Results of FullScale Tests,” Proc. Halon Options TechnicalWorking Conference, 2004.

AcknowledgmentsThis work was a team effort. The authors wishto thank Dr. J. Glockling of the Fire ProtectionAssociation, Dr. G. Davies and Prof. P. Nolan ofSouth Bank University, as well as P. Guilbert, P.Stopford, H. Forkel and P. Everitt of ANSYS, Inc.for their contributions.

© British Crown Copyright 2009/MOD.

Published with the permission of the Controllerof Her Britannic Majesty’s Stationery Office.

Reusing Legacy MeshesANSYS tools enable users to work with finite element models in various formats for performing simulations as well as making changes to part geometry.By Sébastien Galtier, Software Developer and Pierre Thieffry, Product Manager, ANSYS, Inc.


When designs from past projects must be analyzed, or when amodified version of the geometry must be evaluated, the starting pointis generally the original CAD model. In some cases, however, onlylegacy finite element models are available that cannot be importeddirectly into the user’s current simulation software. These includeNASTRAN®, ABAQUS® and ANSYS FLUENT models, for example, as well as many text-based archival versions of ANSYSmodels. Fortunately, tools in the ANSYS Workbench environmenthave been developed so users can easily convert these models for use in creating new simulation models of the original design and also in modifying the original shape to meet new designrequirements.

Legacy models such as the mesh for a connecting rod, shown in Figure 1, can be read into ANSYS FE Modeler, locatedin the Toolbox section of ANSYS Workbench version 12.0. Once imported, the model is handled by the Skin Detection tool in FE Modeler to provide a proper segmentation of the model’s facets.The quality of the segmentation is key to the process — especiallywhen modifying the shape of the model — and the procedure consists of grouping the external faces of finite elements so theyaccurately represent faces similar to a geometric model. Edges andvertices of the model will then be naturally derived from these faces.

Several methods can be used to identify the faces: detection byangles (between the normal orientations of neighbor elements),detection by curvatures, or employment of facet groups defined bythe user. This last method helps in creating specific areas in whichloads and boundary conditions can be applied. Figure 2 shows theresulting geometry generated based on curvature detection in FE Modeler from the legacy mesh. A mechanical simulation systemfrom ANSYS then can be linked to FE Modeler to apply loads andboundary conditions, as shown in Figure 3, and the model then canbe solved to determine the resulting stresses and deflection (Figure 4).

After such an analysis, the model may need to be modified because the existing design does not meet current technicalrequirements. For this purpose, FE Modeler provides capabilities tomodify the geometry through a feature called the ANSYS Mesh Morpher. A so-called target configuration is created by duplicating theinitial geometry. Then transformations such as offsets, translations orrotations can be applied to the geometric entities. Figure 5 shows howoffsets can be used to enlarge or shrink the holes.

Once the geometry has been modified, ANSYS Mesh Morpher willtransform the initial mesh to conform to the target configuration.These transformations are parametric, with each geometric feature

Figure 1. NASTRAN finite elementmodel of a connecting rod

Figure 2. Geometry created fromsegmentation based on curvaturedetection

TIPS AND TRICKS

Figure 3. Loads and boundary conditionsapplied for analysis with mechanical simulation software from ANSYS

Figure 4. Total deformation results fromthe analysis

Figure 5. Deformed geometry, inwhich the hole on the left has gottensmaller while the other two havebeen enlarged


TIPS AND TRICKS

48

Figure 6. Raw result of conversion in Parasolid format, with all faces NURBS representations

Figure 7. New design after sewing all faces together andmodifying geometry with ANSYS DesignModeler software

affected by a parameter that is used as a way to control theamount of morphing between the initial and target configu-rations. Changing the shape of an existing member can beachieved with projection to a new CAD shape. In this case,the faces or edges created by the skin detection processare projected onto an imported CAD model.

It is important to note that mesh morphing modifies onlythe node coordinates, and no remeshing occurs during theprocess. Once the mesh has been morphed, the model canbe used in the mechanical simulation exactly as it was done with the original model. Since the geometry topologyremains the same, all loads and boundary conditionsapplied to the initial model are still valid, so the analysis can proceed as before.

In this example, changes to the model geometry did notaffect the general shape of the model too heavily: No holeswere added, for example, and the topology remained thesame. The FE Modeler application used in conjunction withANSYS DesignModeler software provides all necessarytools to allow for such changes.

To make more significant changes to the model, the initial geometry must be converted to a Parasolid® model.The result of the conversion is a set of surfacescorresponding to each of the faces obtained from the SkinDetection tool. The surfaces can then be sewn together in FE Modeler to create volume bodies. Figure 6 shows the raw result of this conversion, and Figure 7 illustrates the new design after sewing all faces and modifying the geometry with the standard features of ANSYS DesignModeler. In this way, these tools in the ANSYS Workbench framework allow legacy models to be reusedin a process that is not only faster but also less error-pronethan manually recreating meshes from scratch. ■


Heart disease, often caused by partially blocked coro-nary arteries, is the most common cause of death in theworld. Stenting has become one of the most popular formsof treatment to open plaque-encrusted atheroscleroticcoronary arteries, with hundreds of thousands of such pro-cedures performed in the United States each year. However,according to the American Heart Association, about one infour stent patients will experience restenosis, a repeatednarrowing of the stented artery, less than six months afterthe procedure. Some patients with restenosis must undergoa second stenting procedure to alleviate the subsequentblockage, while for others a full bypass operation is the onlysolution. A team from the Virginia Military Institute (VMI) iscombining simulation with animation software from Compu-tational Engineering International (CEI) to help identify apossible cause for restenosis and to find solutions thatmight help reduce the risk of developing it.

The team at VMI hypothesizes that restenosis may bethe result of arterial injury incurred during the stenting procedure itself. During this procedure, the medical teaminserts a balloon, sheathed by the stent, into the artery andinflates it. Once the stent expands, the balloon is deflatedand removed, leaving the stent in place.

The engineering team at VMI identified one possible reason for injury: end flare, which is caused by balloon over-hang at the end of the stent. This exerts increased pressureon the arterial wall and may scrape it during inflation, whichcould stimulate uncontrolled cell growth in that area.

The balloon’s mechanical properties vary dramatically during the expansion process. Though it begins as a highlyflexible material, the balloon eventually expands in a nonlinearfashion as it nears the stent’s final diameter, making the prob-lem numerically unstable. A factor that is critical to accuratelysimulating the problem is how the structure of the balloon, thestent and the artery are meshed.

The team used HarpoonTM, from Sharc, Ltd., to generatea complex mesh designed to follow the balloon, stent andartery through the expansion from a 1 millimeter diameter toa 3 millimeter diameter geometry. Once the mesh was established, the data was exported to ANSYS Mechanicalsoftware to provide information about stresses and geometry changes that occur during expansion.

The team used EnSight® to turn the simulation data intoanimations that depict the inflation process. The resultingimages allow the medical research team to visualize theprocess for the entire assembly or to focus on the individualcomponents — options that are impossible during thestenting procedure itself. By using simulation and visualiza-tion tools together, manufacturers may be able to redesignand numerically test stent designs and procedures, arrivingat a very clear picture of how each variable affects theoverall issue — all without a patient. ■

In this image, the stent and balloon are hidden, and the remaining plot depicts onlythe artery after stent inflation. The contours represent arterial stress. The red ring,which occurs at the location of highest stress, aligns with the location at which endflare occurs during stent inflation.

The stent expansion process, with the stent shown in light gray, the balloon in darkgray and the artery colored by arterial stress

ACADEMIC

Expanding Stent KnowledgeSimulation provides the medical industry with a closer look at stent procedures.By Matthew R. Hyre, Associate Professor of Mechanical Engineering, James C. Squire, Professor of Electrical Engineering,

and Raevon Pulliam, Virginia Military Institute, Virginia, U.S.A.

© iStockphoto.com/cinoby and © iStockhpoto.com/fasloof

ANSYS, Inc.Southpointe275 Technology DriveCanonsburg, PA U.S.A. 15317

Send address corrections to [email protected].

Documents

ANSYS Advantage Volume 3 Issue 1