Aa v2 i3 Ansys Advantage

ENERGYSPOTLIGHTPAGE 13

PREDICTINGVIBRATIONSPAGE 24

DESIGNING FOR ACOMPETITIVE EDGEPAGE 27

PAGE 8

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 S S U E 3 2 0 0 8

TM

GETTING CONNECTEDWITH MEMS

EDITORIAL

scenarios such as bomb-blast damage to buildings, ballistic impact of aircraft on power stations, stamping andother metal-forming manufacturing operations, or droptesting of cell phones and other consumer products.

The beauty of this breadth and depth of ANSYS struc-tural dynamics solutions is that they are not confined tospecialists intimately familiar with the technology who arerunning problems on supercomputers. Rather, analysiscompression techniques enable most problems to be runon conventional desktop machines. Moreover, analysismodels are far easier to set up through the ANSYS Workbench interface — with features for extracting geometry directly from CAD systems, building modelswith robust meshing tools, and setting up the analysis witha simulation tree that describes problems in user-friendlyterminology related to geometry and physical behavior.

In this way, engineers can easily perform structuraldynamics simulation as a routine part of development —fixing problems, refining designs and optimizing productperformance early in the cycle instead of with costly, time-consuming and usually less-precise physical testing. On acorporate level, forward-thinking manufacturers with thegood sense to invest in these tools enjoy the benefits ofhigher profitability and greater market share through well-engineered products designed with vibration, motion andother real-world behavior in mind. ■

John Krouse, Senior Editor and Industry Analyst

The Power of StructuralDynamics SimulationAnalyzing time-varying loads helps develop innovative products with vibration, motion and other real-world behavior in mind.

Companies increasingly rely on structural dynamicssimulation to study how products vibrate, bend, twist andotherwise move when subjected to loads that vary overtime. Whereas static analysis is traditionally used to deter-mine characteristics such as stress and deflection ofindividual parts under a constant load — such as a weighton the end of a beam — structural dynamics enablesdesigners and engineers to study product behavior ingreater detail. Such analysis could include determining thenatural frequency of a washing machine so the appliancedoesn’t jump around in the spin cycle, for example, or cal-culating the fatigue life of a car suspension to withstandyears of pounding by potholes and rough roads.

Structural dynamics is being implemented in anexpanding range of applications, as seen in some of thisissue’s articles. “Predicting Vibrations in High Power Burners” describes how engineers shortened developmenttime by five months by determining an assembly’s natural frequencies through modal analysis and refining the designearly to avoid these damaging displacements. “No MoreDropped Calls” covers the work of an engineering team atEPCOS NL that used multiphysics analysis to account forfluid, electrostatic and mechanical effects in simulating thetransient dynamic response of an innovative RF-MEMSswitch that promises to reduce the number of disconnectedcell phone calls and extend battery life. “Analyzing RandomVibration Fatigue” is about tools based on probability andstatistics used to study the damaging effects of highlyunpredictable arbitrary loads. There are also advanced toolsfor studying nonlinear dynamics where large, high-speedloads permanently deform structures. Applications include

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.

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

Executive EditorChris Hardee

Managing EditorChris Reeves

Senior Editor andIndustry AnalystJohn Krouse

Art DirectorSusan Wheeler

EditorsErik FergusonFran HenslerMarty MundyShane Moeykens

Ad Sales ManagerShane MoeykensHelen Renshaw

Graphics ContributorDan Hart

Editorial AdvisorKelly Wall

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, CFX, AUTODYN, FLUENT, DesignModeler, ANSYS Mechanical, DesignSpace, ANSYS Structural, TGrid, GAMBIT and any and all ANSYS, Inc.brand, product, service, and feature names, logos and slogans are registered trademarks or 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.

© 2008 ANSYS, Inc. All rights reserved.

About the CoverEPCOS NL and PhilipsApplied Technologies haveused simulation to developan RF-MEMS switch forimproving cell phone signalstrength.

Email the editorial staff at [email protected].

DesignerMiller Creative Group

Circulation ManagerSharon Everts

Cell phone © iStockphoto.com/michal koziarski; French wine valley © iStockphoto.com/katarzyna mazurowska

Spotlight on Engineering Simulation for the

Energy Industries13 The Global Challenge

As engineers and technologists rally to meet energy and environmental demands, they turn to engineering simulation to solve their most complex problems.

16 Blending Solar Panels with Roof ProfilesSimulation guides the design of innovative solar panel frames, reducing molding time, material and cost.

18 Coupling Analyses to Improve Nuclear SafetyCoupled thermal hydraulic and stress analysis of a CANDU feeder pipe helps determine integrity.

20 Reformers Getting ResultsSimulation pushes diesel-powered fuel cells on their way to early markets.

22 Harnessing Natural EnergyMultiple simulation tools are used as a cost-effective way to design reliable offshore wind turbines.

24 Predicting Vibrations in High Power BurnersEngineering simulation reduces development time for industrial burners by five months.

20

24

www.ansys.comANSYS Advantage • Volume II, Issue 3, 20082

CONTENTS

Table of ContentsFEATURES

4 SPORTS

Faster, Higher, StrongerEngineering simulation in sports comes of age at the

2008 Beijing Summer Olympics.

8 ELECTRONICS

No More Dropped CallsUsing true multiphysics incorporating fluid, electrostatic andmechanical effects, engineers are simulating the transientdynamic response of an innovative RF-MEMS switch for improving cell phone signal strength.

10 THOUGHT LEADER

Simulation-Based Innovation as a Competitive AdvantagePredictive analysis tools save time and money at Xerox and, moreimportant, enable top-line revenue growth and the competitiveadvantage that comes from developing winning products.

13

8

4

ANSYS Advantage • Volume II, Issue 3, 2008www.ansys.com 3

CONTENTS

3

34

SIMULATION @ WORK27 SPORTS

Picking Up Speed Speedbike designers use fluid simulation to gain a competitive edge.

30 AUTOMOTIVE

Electromagnetics Comes Through in the ClutchBorgWarner engineers meet a tight deadline in optimizing the design of a revolutionary variable-torque clutch for all-wheel-drive vehicles.

32 PARTNERS

Higher Returns on the Simulation InvestmentOptimizing Linux clusters for ANSYS Mechanical softwaredelivers fast turnaround on large problems.

DEPARTMENTS34 ACADEMIC

Driven to SimulationA teenage student helps improve the aerodynamic design of F1 race cars using ANSYS software and Windows high performance computing.

36 ANALYSIS TOOLS

Stretching Your Elastomer UnderstandingAccurate nonlinear analysis leads to a better material selectionprocess that enables innovation and faster time to market.

39 TIPS AND TRICKS

Analyzing Random Vibration FatiguePowerful ANSYS Workbench tools help calculate the damageof vibrations that lack straightforward cyclic repetition.

43 Extracting Solution-Dependent Regions in CFX-PostIdentifying and quantifying regions of reverse flow in the CFX-Post fluids post-processor.

45 OUTSIDE THE BOX

Bio-Inspiring EngineeringScientists use nature to advance technology.

The recent acquisition ofAnsoft Corporation by ANSYS,Inc. augments the depth andbreadth of the ANSYS port-folio of engineering simulation solutions for electronics by providing increased functionality,usability and interoperability. As a leading developer of high-performance electronic designautomation software, Ansoft isworld-renowned for expertise in

electromagnetic, circuit and system simulation. This tech-nology is highly complementary to leading-edge solutions fromANSYS in the areas of structural, fluids, thermal and electro-magnetic simulation.

Because electronics are now integral to many productsfrom automobiles to coffee makers to industrial equipment,true virtual prototyping must encompass all design aspects ofthose products. The fusion of ANSYS and Ansoft provides uswith a unique opportunity to address the convergence ofmechanical, fluids and electrical engineering that will take simulation to a new level — a level that will provide the truemultiphysics design solutions our customers need.

The next issue of ANSYS Advantage, to be distributed inDecember, will feature articles that introduce readers to thebroad range of capabilities added by the Ansoft product port-folio, for the design of products such as cellular phones,Internet-access devices, communications systems, integratedcircuits, broadband components, printed circuit boards, auto-motive electronics systems and power electronics.

We are excited about the state-of-the-art technologies thatAnsoft adds to the simulation software portfolio from ANSYS.The integration of these two companies and our technologieswill enable ANSYS to better serve our customers throughoutthe world by accelerating the delivery of powerful and compre-hensive, customer-driven engineering simulation solutions.

James E. Cashman IIIPresident and Chief Executive OfficerANSYS, Inc.

ANSYS Adds Leading Electronics Solutions to Its Portfolio

In the Olympics, hundredths of asecond can be all that separates amedal performance from no medal atall. When the margin of victory is thatsmall — whether running or rowing,cycling or swimming — elite athleteswill look for an advantage that providesa competitive edge. Dedicated traininghas always been the foundation for agold medal performance, but with onlyfractions of a heartbeat separating winners and losers, techniques such as psychology, nutrition, physical therapy, massage and meditation haveall become important tools for theOlympic athlete. With athletic perform-ance in many sports approaching the supposed limits of the human body, improved performance will moreoften come from dimensions other than fitness.

Engineering simulation, a valuabletool in certain high-tech industries forthe last three decades and currentlygrowing in its commercial and industrial

Simulation has also been used inthe two most recent Olympic Games —the 2004 Summer Games in Athens,Greece, and the 2006 Winter Games inTorino, Italy — but both its adoptionand visibility have been somewhat lim-ited to date. The recent launch of theSpeedo racing swimsuit in February of2008, though, has made quite a splashin the press. The timing of the productlaunch to coincide with the Olympics,the publicity that the rapidly fallingworld records has generated, and thesubsequent controversy about fair-ness that the suit has sparked havegenerated a high level of buzz. TheBeijing Games in all likelihood will be a turning point in the marriage of simulation and sport, accelerating itsadoption by teams and nations lookingto improve medal counts and shiningthe spotlight on simulation for specta-tors and the general public. It wouldn’tbe surprising if post-event office coolerconversations turn to the simulationcolor commentary.

While certain individuals in thesporting community have carried the

reach, is now also reaching into theworld of sports. High-profile examples— such as the 2007 America’s Cup andthe 2007 Formula One racing season— have conclusively demonstrated theeffectiveness of computer-aided engi-neering and the competitive advantageit can provide. In both examples, thewinners invested heavily in simulationanalysis.

Engineering simulation in sports comes of age at the 2008 Beijing Summer Olympics.By Chris Hardee, Executive Editor, ANSYS Advantage

Citius, Altius, FortiusFaster, Higher, Stronger

Image © Camera4/Thonfeld

Donald Miralle/Getty Images


FEATURE: SPORTS

Swimmers at the 2008 U.S. Olympic Trials wearing the Speedo LZR RACER suit

The German double flatwater kayaking team (foreground) at the 2007 World Championships


simulation torch for years, and certainteams and countries have alreadyadopted it wholeheartedly, others arejust beginning to experiment. The simu-lation-enabled medal count in Beijingwill likely provide the tipping point. Thewater sports — where hydrodynamicdesign and optimized technique can enhance speed — are all primecandidates for further analysis. Thisincludes swimming, of course, but alsoall of the boating events, such ascanoeing, kayaking, rowing and sailing.As with the America’s Cup yachtdesigns, which have benefited from asignificant investment in simulationanalysis, every detail of boat design,from the hull to the most inconsequen-tial fixture, can influence performanceand benefit from multiphysics analysis.All of the cycling events, with victorydependent on aerodynamic and struc-tural variables — such as helmet shape,wheel design, derailleur materials, andshifter mechanisms — can benefit fromfluids and structural analysis. In tennis,raquet design and material choiceshave already revolutionized the sport,increasing speed, power and control.And even in track and field events,where the equipment is simple — simply shoes in many cases — designand material advances can provideincreased bounce, a lengthened strideand the infinitesimal nose that can pro-vide the winning edge in a photo finish.

While simulation is likely to emergeinto the public spotlight in this year’sOlympic Games because of its impacton the events themselves, its impactand influence on the design of theevent venues — the stadia and sport-ing centers — while no less significant,will most likely remain out of the publiceye. The use of both mechanical andfluids analysis for the design of build-ings and the comfort and safety of thespectators, press and athletes who usethe buildings is a more traditional andentrenched use of the technology.

Image courtesy British Cycling

55

FEATURE: SPORTS

2008 Beijing Olympic Results

British track cyclist (center) at the 2008 World Championships

Olympic venues are the physical culmination of years of planning and construction, often preceded byprestigious architectural design compe-titions. The structures are innovativeand visually stunning and are meant notonly to engender pride from the hostcountry, but also to serve as iconiclandmarks for the future. While theseinnovative designs push the architec-tural envelope, simulation technologyhas played a significant role by validating that innovative designs, con-struction techniques and materials willwork with certainty.

People all over the world tuned in tothe Beijing Summer Games in August2008 and witnessed a stage that wascreated in part by the technologicalsophistication and power of computer-aided engineering technologies. Theyalso watched sporting events on thatstage in which the same technologyhad been brought to bear. Faster, Higher, Stronger — the Olympic motto— could as easily be a description of the benefits of engineering simu-lation. A paddle stroke, a flip turn, aheartbeat and hundredths of a secondwere the story in Beijing and often wereall that determined which athletes won a medal. It was interesting towatch how many simulation-enhancedathletes stood on the podium.

Swimmers wearing Speedo’s new LZR RACER® suit,designed using ANSYS software, won 94% of the goldmedals awarded and set 23 new world records.

The British track cycling team, supported by ANSYS technology, won 13 medals, including 7 golds in a total of 10 events.

With ANSYS simulation-enhanced boat designs, the German flatwater kayaking team won a total of 6 medals.

ANSYS Advantage • Volume II, Issue 3, 200866 www.ansys.com

FEATURE: SPORTS

The National Stadium –Bird’s Nest The Beijing National Stadium, or Bird’sNest, is the main stadium for the Beijinggames and was designed by Swiss archi-tects Herzog & de Meuron to incorporateuniversal qualities of beauty and to be sen-sitive to Chinese cultural traditions. Thisvenue hosted the opening and closing cer-emonies as well as the track and field

FEATURE: SPORTS

Table Tennis GymnasiumAs one of the most popular sports inChina, table tennis warranted a worthyvenue. The gymnasium’s roof, incorpo-rating a central transparent ball shape,was designed in keeping with both tradi-tional features and the modern Beijingstyle. Researchers at Tongji Universityused ANSYS software for a nonlinearanalysis of the pre-stressed, steel truss roof structure. In this analysis,researchers assessed the hybrid tension design for pre-stress, nonlinearbuckling and the ultimate bearingcapacity of the roof. The engineeringgroup applied BEAM188 elements andtook into account geometric, materialand support condition nonlinearitiescaused by the slide bearing. The teamdetermined that the structural validityof the roof depended on the strengthand rigidity of the support column,which is comprised of a central rigidring (5 meter by 2 meter) and roofbrace system. The analysis helped inthe creation of an optimum design.

Badminton ArenaBadminton — another popular sport inChina, like table tennis — was played inan arena designed in the shape of a shuttlecock. The building is a single-layer reticulated shell structure that is 62 meters long by 46 meters wide.Researchers at Beijing University of

Technology used ANSYS software toperform a nonlinear stability analysis. Theteam used BEAM188 elements — a 3-Dlinear finite beam element, based on theTimoshenko beam theory. The analysistook into account shear deformation androtational inertia effects. Utilizing ANSYSParametric Design Language (APDL), theresearch team constructed a virtualmodel of the arena. The simulationresults indicated that the structure isleast stable if the initial geometry dis-figurement ratio is approximately 1/250.

VENUES

Image © Sunmdm/dreamstime.com

Cancan Cho/Getty Images

STR/Getty Images

Angelo Cavalli/Getty Images

events. The 332 meter long by 296 meterwide elliptical structure has an unusual lat-tice steel design. One group of researcherssimulated the structure using ANSYS software for static and dynamic analysis.Another group used simulation to evaluatethe structural response to a spatially varying magnitude 7 earthquake in whichthere were multiple support excitations andmultiple natural frequencies in the 1 to 6 Hzrange. In this study, displacements of up to 0.9 meters were predicted and withinacceptable limits for the design.

ANSYS Advantage • Volume II, Issue 3, 2008www.ansys.com 7ANSYS Advantage • Volume II, Issue 3, 2008www.ansys.com 7

FEATURE: SPORTSFEATURE: SPORTSEvents

Flatwater Kayaking – The German National Team

Take a carbon fiber kayak no longer than 5.20 meters and no heavier than 12kilograms with an accentuated V-shaped hull and very low draft when loaded. Add

elite paddlers stroking at racing pace with double-bladed paddles, and you have theingredients for a very complex simulation problem. The Institut for Research and Develop-ment of Sports Equipment (FES) in Berlin performed this fluids analysis calculation usingANSYS CFX software. As the paddles grab the water, the trim of the boat and the waterresistance change constantly. The simulation, which involved two-phase flow around thehull and calculation of the boat’s changing trim, were verified through experimentation in atowing tank. The end result was an overall reduction in drag of up to three percent. Usingsimulation, FES helped design the German team’s entire fleet of flatwater racing kayaksfor the Beijing Games, as well as boats for the canoeing, rowing and sailing events. ■

Thanks to the following for their assistance with this article: for the Speedo swimsuit story: Keith Hanna,Leigh Bramall, Natalie Fieldsend and Helen Rushby, ANSYS UK, Ltd.; for the track cycling story: Rob Lewis,Total Sim; Scott Drawer, UK Sport; Natalie Fieldsend, ANSYS UK, Ltd.; for the flatwater kayaking story:Mathias Jirka, ANSYS Germany GmbH; Nicholas Warzecha, FES Berlin; for the Olympic venue stories: TonyHu and Angela Liu, ANSYS China (Pera Global Holdings, Inc.).

Swimming – The Speedo LZR RACER Swimsuit

From the starting gun to the final touch, there is nothing between anOlympic swimmer and a medal except water. Decreasing passive drag

was the engineering challenge that Speedo took on three years ago when itpartnered with a number of organizations — including ANSYS, NASA and sev-

eral universities — to create the world’s fastest swimsuit. In conjunction withresearch on fabrics and suit construction, as well as testing in water flumes,

fluid analysis using ANSYS software was a critical part of the project. Withthe analysis identifying the locations of greatest drag on the swimmer’s body, specialfabric panels were bonded to the suit in those regions and were also used tomold the swimmer into a more hydrodynamic shape. It has been calcu-lated that the suit has five percent less passive drag than theirprevious fastest suit, and world records have fallen at anunprecedented rate since the introduction of theLZR RACER swimsuit in February 2008.

Image courtesy Speedo

Image courtesy FES

Image © Camera4/Thonfeld

Poo

l pho

to ©

iSto

ckph

oto.

com

/Pur

due

9394

Track Cycling – The British National TeamIn competitive track cycling, aerodynamic drag is perhaps the most important adversary. Pedal-ing a single-speed, lightweight racing bicycle on a 250-meter banked oval track requires

attention to every detail in order to shave fractions of seconds.The bicycle–rider system, while seemingly simple, containsapproximately 250 components — all of them critical whenwaging a war on drag. Derailleur, sprocket, chain, wheels, helmets and suits can all be optimized, resulting in incrementalgains in speed. Working with UK Sport’s Innovation team, theBritish track cycling team has employed simulation technologysupported by TotalSim to a degree that few other Olympicteams have done to date. Following the 2004Athens Games, in which they won four

medals, the team has invested significant resources in researchand development, much of it on the computer. The resultsat the recent World Track Championships in March2008 had the British team winning nine outof 18 gold medals.

Image courtesy British Cycling Image courtesy United KingdomSports Council


FEATURE: ELECTRONICS

One of the most perplexing prob-lems for mobile phone users isdropped calls — those annoying andunpredictable disconnections when signal strength falls below a giventhreshold. Typically the problem is amismatch in impedance (resistance toelectromagnetic wave transmission)between the phone’s antenna andpower amplifier, causing signals to bepartially reflected back into the ampli-fier rather than transmitted into the surrounding open space. Suchimpedance mismatches usually arecaused by the presence of objectsadjacent to the antenna — the caller’shand, a car frame or building wall, for example — resulting not only indropped calls but also shortened talktime as battery power is drained trying to maintain signal strength.

An innovative solution to this prob-lem is an adaptive antenna-matchingmodule that senses the mismatch andautomatically changes the phone’simpedance by adjusting a capacitorvalue in a matching network betweenthe power amplifier and antenna. Thedevice is expected to reduce powerconsumption of mobile handsets by upto 25 percent and significantly reducethe number of dropped calls.

The heart of the module is a set of RF-MEMS (radio frequency-micro-electromechanical systems) switches,made with semiconductor manufac-turing techniques and materials. Thecompact size, sensitivity and speed ofMEMS devices are being leveraged in

large on–off capacitance ratio (1:20)needed to change impedance levels toan optimal value for better signal trans-mission. One of the major challenges indeveloping the device is ensuring thatthe switch actuates in 50 microsecondsor less — fast enough to shift imped-ance before a call is disconnected.

With conventional electromech-anical switches, such performance iseasily verified and refined through aseries of test and redesign cycles usinghardware prototypes. Semiconductorfabrication setup for MEMS is costlyand time-intensive, so engineeringsimulation is an indispensable tool inoptimizing MEMS designs early in devel-opment. Simulation is especially helpfulin predicting the complex MEMS per-formance, which typically is influencedby several interdependent variables andoften defies intuitive logic. Amazingly,the RF-MEMS switch in this applicationis small enough to fit on the head of a pin— approximately 250 microns squareand five microns thick, with a three-micron travel distance for the capacitiveswitching plate.

The engineering team used ANSYSMultiphysics software extensively in thedevelopment of the RF-MEMS switch.The solution was especially important indetermining switching speed, a criticalparameter that depends on three inter-related effects:

• Electrostatic force of a transducerthat actuates the opening andclosing of the switch when anelectrical voltage is applied

A single directly-coupled multi-field model of the RF-MEMSswitch containing elements accounting for three effects:fluid (blue), electrostatic (red) and mechanical (yellow)

an expanding range of applicationsincluding automotive manifold pres-sure sensors, ink-jet printer nozzles,pacemakers and industrial equipmentsystems.

This particular module is underdevelopment at component manufac-turer EPCOS NL, which recentlyannounced the acquisition of the RF-MEMS activities from NXP Semi-conductors. At specific points in thedevelopment, Philips Applied Tech-nologies — a contract research anddevelopment supplier — supported theRF-MEMS activities with their specificexpertise in finite element modeling.

RF-MEMS switches are well suited for this adaptive antenna-matching application because of their linearity and accuracy, and the

No More Dropped CallsUsing true multiphysics incorporating fluid, electrostaticand mechanical effects, engineers are simulating thetransient dynamic response of an innovative RF-MEMSswitch for improving cell phone signal strength.By Jeroen Bielen and Jiri Stulemeijer, EPCOS NL, Nijmegen

Sander Noijen, Philips Applied Technologies, Eindhoven, The Netherlands


FEATURE: ELECTRONICS

www.ansys.com 9

edge

mid

Capa

cita

nce

(pF)

Gap (µm)

Detail of the electromechanical model shows an example of the capacitance-gap relationship obtained from an independent electrostaticsimulation and used for the transducer elements shown in red. The twocurves for the edge and middle of the plate represent fringing effects ofthe electrostatic field on the capacitance-gap relationship.

• Mechanical configuration of com-ponents, including the residualgap between electrode platesthat determines the capacitanceof the closed switch

• Fluid behavior of the inert gas inthe hermetically sealed moduleas it is squeezed from the gapbetween the electrode plates asthe switch closes

Software from ANSYS accountedfor all three of these interrelatedeffects using the same directly-coupled multiphysics model, thusavoiding the delays and potentialinaccuracies of exchanging resultsbetween different models.

Parametric capabilities of thesoftware were especially helpful inmodifying the configuration of theswitch by merely changing a fewkey parameters rather than rebuilding the model from scratch. In particular,scripting features of the ANSYS Parametric Design Language (APDL)enabled the engineering team to implement an algorithm for readilydetermining the capacitance–voltage(CV) curve, including nonlinear snap-back instabilities that characterize thequasi-static behavior. APDL was alsoused to run design optimization andsensitivity studies, most importantly insimulating the potential instability of theswitch in its “almost closed” state.

The engineering group used specialized ANSYS elements to accurately represent switch behaviorfor the various stages of gap closure

between the switch’s elec-trode plates, especially in thecritical “almost closed”state. Nonlinear gap ele-ments were used to capturethe mechanical action of theswitch, including contact ofthe electrode plates atcomplete closure. Similarly,electrostatic transducer ele-ments provided high-fidelitysimulation throughout. A newnonlinear transient squeeze-film formulation capability of the FLUID136 element was used to accurately represent the air gap and fluid damping effects in the switch. EPCOS and Philips Applied Technolo-gies assisted in validationof this element for use inlarger pressure regimes.

Taking into account fluid,electrostatic and mechanicaleffects in a single model,ANSYS Multiphysics tech-nology accurately predictedthe switching time for the module andallowed engineers to refine the designfor optimal performance. The processenables the team to simulate numerousmodule configurations quickly, provid-ing fast turnaround for rapidly changingcell phone requirements for a widerange of phone models. Moreover, software from ANSYS is used in studying other aspects of moduledesign, including thermal–mechanicalsimulation to predict material creep and

plasticity for solder joint fatigue lifecalculation, or structural analysis todetermine stress and deformation forvarious packaging alternatives.

As a result of these capabilities ofANSYS Multiphysics software, an opti-mal design will be released to productionin the 2010 time frame, strengtheningEPCOS’s position in the competitivetelecommunications market with aninnovative product that meets a signifi-cant consumer demand. ■

Finite element model showing pressure (left quadrant) and displacement(right quadrant) overlaid on scanning electron microscope image of an RF-MEMS switch measuring 250 microns wide

Time (s)

Capa

cita

nce

(pF)

1 bar FE simulation

1 bar measurement

0.4 bar FE simulation

0.4 bar measurement

0.0+00 5.0E-05 1.0E-04 1.5E-04 2.0E-04 2.5E-4 3.0E-04

15

10

5

0

Simulation results of opening and closing transients at two ambientpressures agree closely with test measurements.

Acc.V Spot Magn Det WD Exp 100µm200 kV 2.0 200x SE 7.5 1 DIE16_JS70_C_5x5_right


Simulation-BasedInnovation as aCompetitive AdvantagePredictive analysis tools save time and money at Xerox and, more important, enable top-line revenue growth and the competitiveadvantage that comes from developing winning products.By Korhan Sevenler, Director Product Lifecycle Management, Xerox Corporation, New York, U.S.A.

© Xerox Corporation

Xerox Corporation was built oninnovation, defining an industry with theintroduction of the first plain paper photocopier using a new electrographicprocess it called xerography. Innovationremains at the foundation of the $17 billion company as the world leader in document duplication and manage-ment. In the face of fierce globalcompetition, Xerox provides the indus-try’s broadest portfolio of offerings.Digital systems include color and black-and-white printing and publishingsystems, digital presses and “book factories,” multifunction devices, laserand solid ink network printers, copiers,and fax machines.

Engineering simulation is vital indeveloping this expanding range ofproducts. Predictive tools ensure thathigh quality standards are met andreduce the number of prototype testingiterations, each costing tens of thou-sands of dollars and weeks of time.More profoundly, simulation enablestop-line revenue growth and the com-petitive advantage that comes fromdeveloping innovative, winning newproducts.

Design for Lean Six SigmaOne of the latest and most far-

reaching engineering initiatives is theXerox Design for Lean Six Sigma(DFLSS) strategy. The program wasinstituted in 2005 following the start of

the company’s Lean Six Sigma in manufacturing 18 months earlier. At theheart of DFLSS is the capability to per-form numerous design of experiments(DOEs) to study the sensitivity ofchanges in key product variables suchas part manufacturing tolerances, operating temperatures of the machinesor differences in print media. The idea isto arrive at a robust design — one thatdelivers defect-free performance inspite of these variations by taking theguesswork out of design and shiftingthe focus to optimization up front in development.

ANSYS DesignXplorer technologyholds great potential in the Xerox DFLSSefforts by enabling engineers to readilyset up these DOE studies, assessdesign sensitivities through responsesurfaces and quickly develop robustdesigns. The speed and ease of use of the software fits right in with the program of instilling DFLSS throughoutthe company’s engineering ranks andincreasing overall efficiency of engi-neering operations.

Simulation-Based Product DevelopmentAll Xerox products are developed

using leading-edge analysis tools.ANSYS Mechanical software is one ofthe primary analysis tools for advancedsimulation, particularly in multiphysicsapplications in which multiple physicalfactors must be evaluated. ANSYS CFX

Korhan Sevenler

FEATURE: THOUGHT LEADER

and FLUENT products, likewise, areused for computational fluid dynamicsstudies, and ANSYS DesignSpace tech-nology is an ideal tool in early productdevelopment for simulation-baseddesign to assess alternative conceptsand to optimize designs up front in thecycle. Xerox is moving toward greateruse of the ANSYS Workbench interface,which is consistent with the approach ofengineers performing their own analysis,rather than having a central group forthat function.

Software from ANSYS wasinstrumental in developing theiGen3 digital printing system,enabling Xerox to penetrate thecommercial printing market withimage quality “look and feel”comparable to offset presses —yet with a faster speed, greatereconomy for short-run press jobs and the ability to customizeeach page. More than $1 billionwas poured into the R&D project,which resulted in greater than400 patents on a product thatwas the most complex systemever developed by Xerox and that pushed the limits of speedand performance.

Structural analysis of numerousinterconnected parts and assemblieswas of critical value during the designof the iGen3. By shifting engineeringefforts up front as much as possible,engineers could more readily studyproduct performance, spot potentialproblems, evaluate alternatives andrefine the design to avoid problemslater. Using engineering analysis andDFLSS methods, Xerox brought theiGen3 to market on time while keeping

costs in line and maintaining qualityand reliability. The machine is nowregarded as one of the company’s premier flagship products and a majorsource of revenue.

Overcoming Organizational ObstaclesIn many respects, implementing

the tools and technologies for simula-tion-based product development iseasier than overcoming organizationalobstacles across a large, distributed

enterprise. One of the chal-lenges is in securing fundingand time in the developmentcycle for up-front simulation.Traditionally, engineeringgroups have been set up tocomplete designs as fast as possible, with incentivesbased on productivity and speed in performingthese tasks.

In contrast, the simu-lation-based design processfocuses on spending moretime early in the cycle to analyze and refinedesigns, thus saving timeand expense downstreamthrough reduced reliance on

Structural analysis of interconnected parts and assemblies was of critical value in developing the iGen3 digital printing system —one of the most complex systems ever developed by Xerox.

ANSYS DesignXplorer study of a deformation of a polygon mirror used in a complex digital printer

2.41E-3

2.41E-3

2.41E-3

2.40E-3

2.40E-3

2.40E-31.96E+0

1.98E+0

1.99E+0

2.01E+0

2.02E+0

2.04E+0

4.90

E-1

4.94

E-1

4.98

E-1

5.02

E-1

5.06

E-1

5.10

E-1



prototype testing cycles and less last-minute troubleshooting. Over recentyears, Xerox has made great progressin moving to up-front modeling andsimulation in order to reduce physicalprototypes. Training for DFLSS hasbeen especially effective, beginningfirst with the enthusiasts and thenextending to those who recognized thevalue of DFLSS in their professionaldevelopment.

Communities of PracticeOne of the challenges for a large

organization is that professionals gen-erally have only limited opportunities toexchange ideas and share their knowl-edge. Xerox addressed this issue withCommunities of Practice: knowledgenetworks through which people withcommon specialties (engineers, sales,marketing, equipment repair, etc.) candiscuss best practices, experiences,tips and solutions to problems.

As part of this effort, Xerox holds anannual company-wide two-day forumfor engineers on modeling, simulationand DFLSS. Speakers from within thecompany, as well as outside experts, are invited, and ample free time andbreak-out are provided for informal net-working and information exchange.The company also hosts monthly“Lunch and Learn” sessions, in whichengineers meet to hear about and dis-cuss the latest simulation approachesand methods.

Driving Innovation with Simulation-BasedProcesses

One of the greatest values of Com-munities of Practice for Xerox is that

Xerox engineers use ANSYS Mechanical software in a wide range of analysis applications such as these studies of heat transfer (left) and contact analysis (right).

Particle trajectories in a printer emissions control subsystem were simulatedby Xerox engineers with ANSYS fluid analysis software.

dreaming up innovations, such as theincandescent electric light bulb; instead,he had a process of experimentation inplace and workers to carry out his directions.

Likewise, simulation technologytoday enables engineers to be their owninnovator, trying out different ideas effi-ciently to see what works and whatdoesn’t. By zeroing in on the good ideasand iteratively refining concepts withmultiple experiments, engineers todaycan leverage the speed and accuracy ofsimulation in driving product designinnovations, which are the foundation ofindustry-leading companies. ■© Xerox Corporation

engineers can discover how to best utilize simulation tools and techniquessuch as DFLSS and DOE in the productdevelopment process — not solely tosave time and money but also toexplore alternatives, try out differentideas and run through numerous what-if scenarios.

From that type of environmentcomes the stimulus for innovation needed to maintain a competitive edge. In this respect, simulation-basedapproaches today are analogous to whathappened in Thomas Edison’s lab in New Jersey, United States. Consideredone of the most prolific inventors in history, Edison didn’t just sit around

www.ansys.com1212


ANSYS Advantage • Volume II, Issue 3, 2008


CHEMICAL PROCESSING

www.ansys.com 13

The demand for energy, regardlessof rising costs, is driven by a com-bination of population growth, globalindustrialization and the desire forimprovements in quality of life in theremotest corners of the world. Theincreasing worldwide need for reliableenergy at a reasonable cost, combined with environmental concerns, hasbrought science and engineeringtogether into the spotlight. The collec-tive challenge is to utilize technology toanswer the call to find high-yielding bio-mass, deliver affordable oil and gas,develop clean coal, harness the sun’spower, capture the wind and “turn” the

The Global ChallengeAs engineers and technologists rally to meet energy and environmental demands, they turn to engineering simulation to solve their most complex problems.By Ahmad Haidari, Director of Industry Marketing for Process, Energy and Power, ANSYS, Inc.

tides. The potential benefits are there,but the projects are complex and theefforts, in business and human costs,can be enormous. For example, energyindustry engineers are being calledupon to:

• Drill deeper and in harsher environments with reduced environmental impact

• Develop technology to reducegreenhouse gases and the overall carbon footprint

• Make wind and solar power morecost-effective and scalable

ENERGY: OVERVIEW

1313

• Increase fuel cell reliability

• Build cheaper, safer nuclearpower plants

• Reduce energy consumptionthrough improved efficiency and retrofits

Energy companies and relatedindustries are applying engineering simulation technology at a high rate,indicative of an ongoing “energy revolution.” Companies both large and small and in many sectors deploy solutions from ANSYS, Inc. to optimizedesign and engineering approaches in

Roof photo © iStockphoto.com/Max HomendSimulation courtesy Stein DesignHouse image courtesy of Digital Vision/PunchStock


simulations. In addition, there is a needfor engineering software that modelsas much of the overall system as possible. ANSYS provides true multi-physics tools for applications in whichthere is little or no opportunity for simplification, of either the physics orthe detailed geometry. As an example,the unparalleled breadth of tools from ANSYS is ideal for studyingenhanced oil recovery (EOR):


applications ranging from energyproduction and processing to con-servation and end use. The broadacceptance of technology from ANSYSis, in part, due to the company’s unwavering focus on the energy indus-try along with continual investment inexpanding capabilities. For years, manycomponents of ANSYS technologyhave been applied to the energy industry. Turbomachinery applicationproducts are used in rotating equip-ment, including hydroturbine andpower system design. To meetnuclear regulatory concerns, ANSYSfollows NQA-1 quality classificationfor its ANSYS structural mechanicssoftware. Products in the offshore suiteare used for oil, gas and wind turbineoffshore installations as well as forensuring that installations are up toindustry-specific regulatory standards(code check). ANSYS pressure vesselcapabilities help engineers meet ASMEcode requirements and are used in stress analysis of tubes, boilers and storage tanks. Explicit analysiscapabilities are used for catastrophicevent simulation, including impactand explosion.

The undeniable demand from the energy sector is for a strong emphasis on the quality and reliability of

Engineering simulations are used to study micro-cracks and fatigue to evaluate turbine performanceat higher pressure and temperatures.Courtesy Siemens Power Generation

Technology from ANSYS is applied to flow assurance projects dealing with oil flow ability (addressing sand management, slug flow and gas-lift applications), ensuring more efficientand uninterrupted flow of oil in a given pipeline. This example shows slug flow in a gas-liquid pipeline.

14

ENERGY: OVERVIEW

exploring soil mechanics, rock fractionation, drilling, and thermallyenhanced oil extraction concepts in which engineers can combine an explicit study of rock fracture in asame-simulation environment as theyassess the transport of sand awayfrom the reservoir.

Another requirement is for engineering simulation tools to be scalable, available for use on an


CHEMICAL PROCESSING

www.ansys.com 15

engineer’s desktop, across the work group or for enterprise deploy-ment — based on an organization’s specific needs. Global energy companies leverage the flexiblearchitecture of solutions from ANSYSfor worldwide projects that address,for example, offshore platform design,building and retrofitting nuclearplants, designing and installing windturbines, and developing clean coal technologies.

Similar to other industries, espe-cially recently, energy projects andproduct design applications arerestricted by demanding time schedules. Engineers must get reliableanswers from their engineering simu-lation activities in a reasonabletime-to-solution. Features in ANSYStechnology, such as two-way CADassociativity and an intuitive workflowenvironment that reduces the timeneeded to set up problems, lead toefficiency. Organizations also are look-ing to reduce the engineering hoursneeded to run complex engineeringsimulations. Advances in high-performance computing (HPC), along with the exceptional performance of


ENERGY: OVERVIEW

ANSYS products on these machines,are proving to be valuable in simulatingreal-world problems in the energyindustry. Continual HPC improvementmeans that simulations with largecomputational cells are being modeledroutinely, benefiting from near-linearscalability of ANSYS software tools inparallel processing simulations.

There has never been a greaterneed for engineering simulation to helpconserve energy, to bring a plant onlinemonths or years ahead of schedule, to evaluate an innovative energy concept, or to improve equipment efficiency. Even small improvementscan reap big benefits, whether the end product is brand new or a retrofit. Today, engineers successfullyemploy simulation tools to design newequipment that uses less energy,whether power, steam or heat. Throughanalysis, development teams retrofit oiland gas, refining, and power-generationequipment, looking to improve through-put and reliability, avoid unacceptableinterruptions, and increase product life.This broad acceptance of engineeringsimulation tools puts ANSYS in aunique position to help power the world.

This industry spotlight highlightscustomer successes and details afew examples in which many of theANSYS solutions are used in engineering simulation related tothe energy industry. The selectedarticles capture the benefits cus-tomers have reaped in transformingleading-edge design concepts intoreality in nuclear, fuel cell, solarpower, wind turbine, and equipmentretrofit and design applications. ■

For environmental, green and sustainable design,engineers use simulation tools from ANSYS to evaluateretrofit options, optimize combustion and heat transfer,capture pollutants, design pipelines and sub sea equip-ment, look at safety and installation strategies, andincreased structural integrity to avoid spills and disruption,Vortex-induced motion for this truss spar is modeled tounderstand global and local forces on the spar streaks.Geometry courtesy Technip U.S.A.

Image © iStockphoto.com/Magumi Terao

Imag

e ©

iSto

ckph

oto.

com

/Han

s F.

Mei

er

Blending Solar Panelswith Roof ProfilesSimulation guides the design of innovative solar panel frames, reducing molding time, material and cost.By Matthew Stein, President, Stein Design, California, U.S.A.

One of the most efficient sources ofrenewable energy is rooftop photo-voltaic (PV) solar cells, which convertsunlight into electricity for homes andbusiness. Use is hampered, however,by high upfront costs as well as aesthetics, with most solar panelsmounted on unattractive brackets that do not blend well with house andbuilding designs.

Open Energy Corp. of SolonaBeach, California, has overcome thesedrawbacks with SolarSave® panels — a solar roof solutionunlike anything previously available in the industry. Panelsare designed to integrate and interweave with standard roofing tiles so as to blend in with the roof, an important con-sideration in subdivisions with strict homeowner bylawspertaining to roof profiles and solar panel installations. Theseintegrated panels are also cost-effective, as they areinstalled as tiling over part of the roof rather than as an add-on above traditional coverings. The lightweight panelsare warranted for 25 years, are easily handled, and can bewalked on, simplifying installation for roofing contractors andsolar integrators.

In their continuing efforts to improvethe cost-effectiveness and perform-ance of these solar panels, OpenEnergy commissioned Stein Design tocomplete a redesign of the panel withthe goal of reducing unit cost whileimproving strength and reliability. Thenew design was to be a four-foot-longPV panel to replace existing three-footmodels, cutting square-foot costs byreducing the number of electrical connections, related junction boxes and

other hardware. Analysis work was done exclusively usingANSYS DesignSpace software.

Stein Design started the redesign by first evaluating theexisting three-foot panel product. Three-dimensional solidCAD model assemblies were generated in SolidWorks® andthen imported into the ANSYS DesignSpace tool to performthe stress analysis. Two load cases were considered: (1) a300-pound per-square-foot pressure, satisfying at least 99percent of structural building code requirements across theUnited States and Canada for snow loads; and (2) a 400-pound load concentrated in a three-inch-diameter area,representing a concentrated heel-load of an installer on the

Open Energy SolarSave® panels are designed to integrate and interweave with standard roofing tiles so as to blend in with the roof profile and color.

Open Energy solar panels being installed


ENERGY: SOLAR

panel. Experimentally, this walk-on requirement is typicallychecked using an industry-standard 200-pound load transferred to the glass panel via a three-inch diameter steelpuck. Given the weight of some contractors and the heavyequipment they carry, 400 pounds was considered a bettertarget, providing more of a safety margin than the coderequirement of 200 pounds.

Analysis showed that the original three-foot plastic framewas strong enough to support both the walk-on and snowloads. However, further investigation was necessary for thenew four-foot panel, since analysis indicated that a 400-pound walk-on load (twice the industry standard) created stress that was 40 percent greater than the allow-able tensile stress of the solar panel glass (6,000 pounds persquare inch or psi). Moreover, there was concern that insome cases the bond between the glass panel and its plastic mounting frame might be compromised over timedue to temperature expansion differences in the materials,since the coefficient of thermal expansion for the solar glassis about one-tenth that of the frame’s polycarbonate plastic.This results in a one-quarter-inch linear expansion differenceacross the frame when subjected to a 200 degree Fahrenheittemperature change — from extreme heat in direct summersun to sub-zero nighttime temperatures in extreme winter climates. To account for these effects, the four-foot plasticmolded frame was split into two parts, and an interlockingexpansion joint, as well as 10 percent glass-fill to the polycarbonate frame material, was added. These changesreduced the thermal expansion difference between the solarglass and the four-foot plastic frame to one-third that of the original three-foot frame.

The next step in the redesign was to reduce thicknessesand reconfigure the frame walls and ribs to use less materialand shorten molding cycle time, thus lowering productioncost. The original three-foot panel design had nominal wallthicknesses of 0.210 inches and nominal rib thicknesses of0.150 inches, but some walls were as thick as 0.260 inches,resulting in a slow molding cycle time. ANSYS DesignSpacetechnology was used to verify the design as it progressedthrough multiple iterations in which nominal wall thicknesswas trimmed by 0.085 inches and rib thickness by 0.065

inches. Even though nominal wall thickness was reduced by40 percent and nominal rib thickness by 43 percent, themaximum frame stresses rose by only 33 percent overall,through improved rib design and placement. The maximumstress in the four-foot molded frame increased to approxi-mately 2,550 psi — from a level of approximately 1,700 psi inthe three-foot frame — well within the design target of a 3-to-1 safety margin for the 9,000 psi tensile strength polycarbonate material. The resulting four-foot panel frameuses less material than the original frame and can be injection-molded in two-thirds the time, yielding a finishedfour-foot assembly that costs the same to manufacture asthe original three-foot panels.

The use of ANSYS DesignSpace capabilities was criticalthroughout this entire redesign process and is part of thereason Stein Design can provide clients fast turnaround withdesigns that meet stringent requirements. Its ease of useenables engineers to get up to speed quickly, even if severalmonths may pass between analysis projects. Furthermore,the software interfaces seamlessly with SolidWorks mechan-ical design software, so part geometry can be readilychanged and analysis solutions regenerated quickly toinvestigate “what-if” scenarios throughout the developmentprocess. In this way, the technology guides the design to anoptimum configuration that satisfies multiple engineeringrequirements and enables projects to be completed muchfaster than would otherwise be possible. ■

Stress distribution and deformation for walk-on load on the three-foot panel frame

In this reconfigured design that reduced production costs, frame stress increased somewhat for both the walk-on load (top) and snow load (bottom), but remained well within targeted safety margins.


ENERGY: SOLAR


ENERGY: NUCLEAR

Coupling Analyses toImprove Nuclear SafetyCoupled thermal hydraulic and stress analysis of a CANDU feeder pipe helps determine integrity.By Myung Jo Jhung, Principal Researcher, Korea Institute of Nuclear Safety, Daejeon, South Korea

The ultimate goal of nuclear safetyregulation is to protect the public andthe environment from the radiation hazards that could accompany the production and utilization of nuclearenergy. The Korea Institute of NuclearSafety (KINS) develops and imple-ments nuclear safety programs, suchas safety reviews and inspections,development of regulatory standardsand monitoring of environmental radia-tion within Korea. In order to maintainand continually improve nuclear safety,increasing technology depth is requiredfor prediction, analysis, experimentaland remedial measures.

considered pressure and temperaturesimultaneously in generating the normal operating stresses. For the purpose of this study, a total time of180 seconds was considered forheat-up and cool-down. Assuming aninternal pressure of 10 MPa, the teamof investigators discovered that maxi-mum levels of equivalent stress andstress intensity were located in theintrados (inner curve) of the first andsecond bend. They also discoveredthat stress component variations alongthe circumference were more severealong the radius of the inner surfacethan along the outer surface.

The team used ANSYS CFX fluidflow simulation software to model theflow of the heavy water coolant anddetermine the temperature distributionwithin the heavy steel pipe. The investi-gators set the initial conditions to be

ANSYS CFX model of pipe exterior (left) and interior (right)Transient thermal data representing a typicalheat-up and cool-down cycle of the pipe model

Time (sec)

Tem

pera

ture

(˚C)

Heat-up / Cool-down400

300

200

100

0

----–----–----–----

----–----–----–------–--–--–--–--–--

0 30 60 90 120 150 180

Because components in operatingnuclear reactor systems can be subjectto extreme forces and stresses that maythreaten their integrity, safety is a constant concern. Safety is ensured bypredicting conditions that would lead tocomponent failures using simulationsthat incorporate fluid structure inter-action (FSI) as a key technology.Simulations using FSI, for example, caninvolve taking results from a simulationof fluid flow with convective heat transfer and applying these results asloads in a structural simulation. In thepast, these fluid and structural fieldstypically were analyzed separatelydue to the limitations of computersoftware and hardware resources. But advances in both areas now permit unified and efficient multi-physics simulations that couple the combined effects of interrelatedphysical phenomena (physics or fields).

In this project, KINS researchersperformed a coupled thermal hydraulicand stress analysis of a pipe with twobends. They studied transient heat-upand cool-down of the feeder pipe that delivers the primary coolant to the nuclear fuel of a CANDU pressur-ized heavy water reactor. The researchteam then used the results of this simu-lation for fatigue analysis of the pipe.

The team developed a finite element model for simulation usingANSYS Mechanical software. For this structural analysis, the engineers

Image ©iStockphoto.com/Hans F. Meier


When integrating structural and fluidsanalyses, the intuitive interface of theANSYS Workbench platform enables designersand analysts to account for one-way or two-way fluid-structure interaction (FSI).

For example, if an ANSYS mechanical sim-ulation requires the results from an ANSYS CFXsimulation in the specification of a load, usersonly need to select the relevant surfaces andthe ANSYS CFX results file that contains thedesired load. The ANSYS Workbench environ-ment takes care of the rest, includingmanagement of files, extraction of data, inter-polation between meshes and application ofboundary loads.

User setup for two-way FSI only requiresselection of the surfaces at which informationsuch as temperatures or pressures areexchanged. The CFD and FEA solvers then run concurrently with robust implicit coupling onone or more machines connected by LAN, WANor even Internet. Load transfer between the twouses an advanced algorithm that is both profile-preserving and conservative. There is noneed for third-party software — all the dataexchange is handled automatically and inter-nally, using built-in socket-based inter-processcommunication (IPC).

There is no compromise in capability asthe ANSYS FSI solution uses the full power andfeatures of ANSYS CFX and ANSYS mechanicalproducts.

– John Stokes, Product Manager ANSYS, Inc.

Simplifying the FSI Processwith ANSYS Workbench

Temperature distribution in the pipe for typical heat-up and cool-down. From left to right: 10, 30, 70, 100, 130 seconds,with blue indicating lower temperatures and red higher temperatures

a stationary fluid and a temperature of20 degrees Celsius for both the fluidinside the pipe and the pipe itself. Asthe heavy water flowed through thepipe, the temperature of the pipeincreased due to the heat transferbetween the pipe and the fluid. Theteam assumed a constant referencepressure of 10 MPa, and in their simula-tions included the variations of materialproperties with temperature of both theheavy water and the pipe.

KINS engineers then used the ther-mal results from the fluids simulation asinput for a structural simulation thatanalyzed the resultant thermal stresses.They were able to obtain predictions of equivalent stress variations duringheat-up (30 seconds) and cool-down(100 seconds). Analyzing the resultsand comparing the heat-up with the cool-down phases, the KINS teamdetermined that the most severe axialand circumferential stresses arose atthe outer surface during heat-up and at

the inner surface during cool-down. Aswas seen with the pressure-basedstresses, maximum thermal stressesoccurred in the intrados of the bend.

The greatest thermal stressesfound during cool-down, combinedwith the pressure-driven stresses, wereused to determine the maximum equiv-alent stresses, which were quantified tobe approximately 19 MPa. The fatiguecurve for carbon steel [1] indicated a life of more than 106 cycles under thisstress — much greater than what thefeeder pipe is expected to see in oper-ation. Therefore, the KINS researcherswere able to conclude that the cumulative usage factor is almost infinite, and thermal fatigue of the pipedue to heat-up and cool-down over thetime considered is negligible for thisoperating scenario.

Software from ANSYS allowed the KINS team to successfully performa coupled thermal hydraulic-stressanalysis of the CANDU feeder pipe to verify integrity estimates. By per-forming a unified simulation, the combined effects of the interrelatedphysical phenomena could be investi-gated efficiently, reducing both the timeand the cost of independent simu-lations. At the same time, this approachprovided a more realistic picture of thebehavior of these components underthe given operating conditions. ■

References[1] ASME, ASME Boiler and Pressure Vessel

Code, Section III, Appendix I, The AmericanSociety of Mechanical Engineers, 2004.

Equivalent stress predictions in the pipe analysisassuming a constant interior pressure of 10 MPa. Theline indicates where calculations for stress variationsand stress variation components were conducted.

ENERGY: NUCLEAR


ENERGY: FUEL CELLS

ReformersGetting ResultsSimulation pushes diesel-powered fuel cells on their way to early markets.By Zdenek Pors and Ralf Peters, Institute of Energy Research – Fuel Cells, Forschungzentrum Jülich GmbH, Germany

In the transportation sector, fuelcells offer the potential of improving the energy conversion efficiency anddecreasing overall pollutant emissions.Polymer electrolyte membrane (PEM)fuel cells are especially suitable intransportation applications, for bothvehicle propulsion and as auxiliarypower units (APUs), because of theirgood dynamic characteristics.

Usually for logistical reasons, APUsuse the same fuel as the main engine;for example, this could be kerosene inaircraft or diesel in trucks. BecausePEM fuel cells are powered by hydro-gen, if they are to use a hydrocarbonfuel such as kerosene or diesel, catalytic reforming must be employedto extract the hydrogen. In a catalyticprocess called autothermal reforming,liquid hydrocarbon fuel reacts with oxygen and steam to produce a refor-mate product that consists mainly of hydrogen along with some carbonmonoxide and carbon dioxide. Reformingcommercially available diesel fuels hasa number of technical challenges,including complete fuel conversion,processing of aromatic compoundsand prevention of carbon formation,

A liquid fuel concentration isosurface colored by evaporation rate is plotted with the flow pathlines coloredby the concentration of evaporated diesel for the ATR-7mixing chamber.

Mixing device assembly for the ATR-7 autothermaldiesel reformer

also known ascoking. Theseeffects progres-sively deactivatethe catalyst and lead toreduced system durability.

The function of anautothermal reformer (ATR)mixing chamber is to supply uniformflow of a homogeneous mixture of air,steam and evaporated fuel to the catalytic reaction zone. In designing an ATR mixing chamber, importantprocesses for optimization include fuelinjection, atomization and evaporation.Poor quality of the reactant mixture hasa negative impact on the conversionefficiency, so it is important to maintainproper air–fuel and steam–fuel ratios toavoid hot spots. The fuel must also becompletely evaporated before enteringthe reaction zone to prevent catalystdamage due to coking.

To completely evaporate the fuel, a considerable amount of heat isrequired. One challenge of thisrequirement, especially for fuels with

high boiling temperatures, is that the temperature required for evaporationsometimes exceeds the fuel ignitiontemperature. Complete evaporationof the diesel fuel without coking thus represents the greatest technologicalchallenge in the reactant mixingprocess.

At Research Centre (FZ) Jülich inGermany, researchers have beendeveloping diesel fuel processing unitsfor PEM fuel cells, including ATRs,since 1998. They have used compu-tational fluid dynamics (CFD) as a tool for design optimization since 2003. From a modeling standpoint, of special interest to FZ Jülichresearchers was the mixing of diesel,


ENERGY: FUEL CELLS

air and water at the reformer inlet, alongwith the temperature and concentrationdistribution to individual sections of theceramic honeycomb structure thatcontains the catalyst. The researchgroup found this task too complex to be solved by “trial and error” designimprovements alone and chose FLUENT fluid flow simulation softwarefrom ANSYS to analyze the flow in the reformer mixing chamber.

To simulate the flow regime, theresearchers used the k-ε turbulencemodel, since its accuracy was sufficientfor the purpose. Based on the type ofnozzle in the reformer, they chose touse the pressure swirl atomizer model,one of the FLUENT software’s discretephase models, to study the fuel injection and atomization process.With respect to the complex physical phenomena of the atomization andevaporation processes, they chose the pure hydrocarbon n-tetradecane(C14H30) as the model fuel for diesel.

The analysis helped to identifyweak areas in the existing design andto optimize a new design. In the 5 kWe

power class, the previous state-of-the-art autothermal reformer type 5 (ATR-5)did not prove feasible because eventhough the mixture homogeneity was sufficient, the design could notguarantee complete fuel evaporation.

The CFD model of ATR-5 found unfa-vorable flow profiles near the fuelnozzle (for example, dead flow zones)and poor heat exchange from hotgases to fuel droplets. In a new con-cept, the ATR-7, the energy of thesuperheated steam was used to forceevaporation of the fuel while air wasthen added to the mixture downstreamof the evaporation zone. A specialdesign of the mixing chamber’s evapo-rator section guaranteed intensivemixing of the fuel spray and hot gases,thus completely evaporating the fueland preventing coking. Improvementsin the steam and air inlet designs led tostronger turbulence effects in the mix-ing chamber, which made for a bettermixture quality and an overall reduction

in the chamber size. The basic conceptof the ATR-7 was then further improvedin the next-generation ATR-8, whichenabled fuel distillation residue separa-tion from the reactant gas streambefore entering the catalyst zone.

This development confirmed for FZJülich researchers that FLUENT software was an important tool for optimizing mixing processes in reactorsand balance-of-plant devices such as heat exchangers. Worldwide, the ATR-7 and ATR-8 are the only knownreformers tested with commercialdiesel without measurable catalystdeactivation. The next developmentstep is to scale the design up from 5 kWe to 50 kWe. Analysis of the largermixing chamber concept shows a morecomplex flow field design. In the largerdesign, the incorporation of curvedblades in the air feeding area serves to intensify mixing and provides ahomogeneous flow profile behind theair mixing zone. A glass model of thisdesign has been fabricated for the purpose of flow experiments and isscheduled to be fully tested by 2010.With such reformers leading the way,cooperation between researchers andindustry could make it possible to introduce the first diesel- or kerosene-powered fuel cells to the market by 2015. ■

Comparison of hydrogen concentration during autothermal reformingof ARAL Ultimate diesel for the ATR-5 (non-optimized) and ATR-7(CFD optimized) designs

Glass prototype of a 50 kWe ATR

A 50 kWe ATR design incorporates curved blades in the air feeding area, whichresult in more agitated mixing and provide a homogeneous flow profile behind theair mixing zone, as seen by the pathlines depicted here.

ATR-5non-optimized

ATR-7CFD optimized

0 50 100 150 200 250 300 350 400 450 500

45

40

35

30

25

20

15

10

5

0

Time on stream [h]

H 2co

nc.i

n th

e dr

y re

form

ate

[vol

.%]

Structure of an offshore wind turbine and loads that must be considered for simulation

Turbulent Wind Fields

(Aerodynamics)

Waves and Currents

(Hydrodynamics)

Soil–Pile Interaction

Tower

Substructure

Foundation

Support Structure

Model

Rotor Nacelle Assembly

Rotor

Controller

Drive Train

Governments around the world are looking to offshorewind power because of its potential as a reliable source ofinexpensive, renewable energy. However, developing windfarms in a marine environment comes with a new set of engi-neering challenges. The support structures for these offshorewind turbines (OWTs), for example, must be designed tofunction effectively in deep water and with large turbines.Offshore projects also present challenges to design engi-neers, manufacturers and operators because storms, roughseas and saltwater subject the entire turbine and its associ-ated support structure to extreme stresses. Engineeringsimulation is a valuable tool for designing cost-efficient andreliable large-frame OWTs.

To evaluate OWT designs for life expectancy and certifi-cation, detailed analysis of critical parts of the turbine is very important in order to predict fatigue. The OWTs beingstudied consist of a turbine and tower that are attached to

a partially submerged substructure. The substructure is fastened to the ocean floor using foundation piles. In order toobtain accurate results when simulating the overall system, a number of effects must be considered simultaneously,including loads from turbulent wind fields, the turbine controlsystem, loads resulting from waves and currents, the elasticbehavior of the support structure, and the soil characteristicsof the local sea bed.

To carry out the complex simulation of OWTs withbranched support structures, engineers at the FraunhoferCenter for Wind Energy and Maritime Engineering (CWMT)used a special purpose aeroelastic software, ADCoS. Thistool relates the influence of the environment (wind effect,wave type and structure, sea state and behavior of theocean currents) and the soil-pile structural interactions, tothe structural capacity of the overall wind turbine. ADCoS,developed by Aero Dynamik Consult Ingenieurgesellschaft,

has typically been utilized for onshore windturbines and has the capability to conduct adetailed investigation of interacting loadsand the resulting dynamic response on anOWT. Extensive knowledge of all the loadsources and their interactions can helpimprove the reliability of OWTs and is vital for cost-effective operation of offshorewind farms.

Before analyzing a turbine’s reaction towave loading, the CWMT engineering teamneeded to develop a finite element model of the entire structure. Researchers usedANSYS Mechanical software to developand define the support structure as a parameterized beam model. Using theANSYS-to-ASAS translator, they trans-ferred the model to ANSYS ASAS software.

Within the ANSYS ASAS Offshoreanalysis tool, the engineering group used

ENERGY: WIND

www.ansys.comANSYS Advantage • Volume II, Issue 3, 20082222 www.ansys.com22

Harnessing Natural EnergyMultiple simulation tools are used as a cost-effective way to design reliable offshore wind turbines. By Fabian R. Vorpahl, Holger Huhn and Hans-Gerd Busmann, Fraunhofer Center for Wind Energy and Maritime Engineering (CWMT), Bremerhaven, Germany

Stefan Kleinhansl, Aero Dynamik Consult GmbH, Neuhausen/Stuttgart, Germany

Image courtesy REPower

ENERGY: WIND


ASAS-WAVE to calculate the wave loads on the supportstructure. With the aid of ASAS-WAVE, loads resulting fromlinear and nonlinear waves, as well as irregular sea statesand currents, were taken into account. The research teamcalculated the loads using Morison’s equation and thenexported the distributed member loads, as equivalent nodalloads, into a text file for further use.

For the next step, the team conducted the nonlinear sim-ulation of the foundation piles using the P-Y approach, asrecommended by the American Petroleum Institute. Thiswas done using SPLINTER, the soil-pile interaction tool, inthe ANSYS ASAS Offshore product. SPLINTER allows sim-ulation of single piles or pile groups, including groupinteraction effects via the soil medium. The outcomes of theSPLINTER analyses were linearized stiffness matrices foreach pile head.

Finally the CWMT researchers performed the aero-hydro-servo-elastic (nonlinear finite element) simulationusing the ADCoS software. In this step, engineers input themodel created in the ANSYS Mechanical software, the waveloads from the ANSYS-WAVE tool and the stiffness matricesfrom SPLINTER for the pile heads into ADCoS. The modelingof the support structure, calculation of wave loads and com-

putation of soil–pile interaction were executed using a singleWindows batch file. To date, some validation has been completed, and further validation will be performed underthe Offshore Code Comparison Collaboration project (OC3) within the International Energy Agency’s Wind Annex XXIII.

In researching the interaction of wind and water with awind turbine structure, CWMT engineers use ANSYSMechanical functionality for creating the structural modeland the ANSYS ASAS Offshore suite’s extensive options toaccount for hydrodynamic loads and soil characteristics.The adaptive architecture of these tools allows them to beconnected to ADCoS, which then enables the simultaneousaero-servo-hydro-elastic simulation of the OWT. Detailedload history information and resulting fatigue data — such asrain flow counts, load spectra and damage equivalent loads — can be derived from ADCoS. Equipped with thisknowledge, CWMT can perform in-depth investigations and optimization of critical parts, leading ultimately to even more reliable and cost-effective OWTs for future wind farm projects. ■

References

[1] Jonkman, J., Butterfield, S., Musial, W. and Scott, G., “Definition of a 5-MW Reference Wind Turbine for Offshore System Development,”NREL/TP-500-38060, Golden, CO: National Renewable EnergyLaboratory, February 2007.

[2] Kleinhansl, S., Mayer, M. and Mangold, A. “ADCoS — A NonlinearAeroelastic Code for the Complete Dynamic Simulation of Offshore-Structures and Lattice-Towers,” DEWEK – Proceedings, 2004.

[3] Vorpahl, F., Huhn, H., Busmann, H.-G. and Kleinhansl, S., “A FlexibleAeroelastic Simulation Approach for Offshore Wind Turbines,”European Offshore Wind Proceedings, 2007, www.eow2007proceedings.info/allfiles2/272_Eow2007fullpaper.pdf (11.06.08).

[4] Nichols, J., Camp, T., Jonkman, J., Butterfield, S., Larsen, T., Hansen,A.M., Azcona, J., Martinez, A., Munduate, X. and Vorpahl, F., “OffshoreCode Comparison Collaboration within IEA Wind Annex XXIII: Phase IIIResults Regarding Tripod Support Structure Modeling” (to be published).

[5] http://www.cwmt.fraunhofer.de

[6] http://www.aero-dynamik.de

Wave loads as simulated by ANSYS ASAS softwareSupport structure as a beam model in ANSYS Mechanical(left) and in ADCoS software (right)

Simulation process and data handling

ANSYS Solutions

Structural Simulation• Definition of parameterized

support structure

• Optimization of the structure

Offshore Simulation• Calculation of loads from

waves and currents on support structure

• Calculation of stiffnessmatrices for pile elements

ADCoS Software

Aeroelastic Simulation

• Parameterized model

• Deterministic andstochastic wind loads

• Loads from waves andirregular sea states

• Soil characteristics

Model used for ANSYSASAS simulation

Support structuremodel transferred toADCoS using a macro

Nodal loads transferred as timeseries in text file

Stiffness matrices for pile elements transferred as text files


ENERGY: BURNERS

24

Displacement at the first natural vibrational mode for the original Riello burner configuration

Industrial gas and oil burners are used for steam produc-tion in industrial processes and electric power generationplants. Reducing vibrations in these systems can increasecomponent longevity and reduce maintenance on boltsand flanges. A gas or oil burner is comprised of an intake system that draws air into the burner and an ignition area where fuel is introduced, mixed with the intake air and ignited. In the air intake system, an electric motordrives an impeller, which pulls in air and propels it along thevolute-shaped housing and into the ignition area. The ignitionarea contains a sleeve duct that encases a combustion head and attaches the entire burner to the associated combustion tube, or boiler. An industrial burner that is capableof producing more than five megawatts may exhibit structuralvibrations due to interaction between the combustion chamber and the flame initiating from the burner itself.

Experiments on these kinds of burners at Riello Burners Combustion Research Center showed that thefrequencies of the vibrations depended on the power ofthe flame, the dimensions of the combustion tube, andthe features and design of the burner assembly itself. In order to reduce vibration in the burner system, engineers performed a detailed study of their structural

Predicting Vibrationsin High Power Burners

Engineering simulation reduces developmenttime for industrial burners by five months.By Gianluca Argentini, Mathematical Modeler, R&D Department, Riello Burners, Legnago, Italy

properties, including natural modes of vibration anddependence on the geometry and the materials used inthe system’s components.

The usual mathematical description for problems ofthis type is based on the structural mass [M], stiffness [K]and damping [C] matrices of the system, which are related to the displacements of the structure {U(t)} by thestandard set of differential equations of motion:

While this equation can be manually solved for simple, linear, discrete systems, a numerical approachlike finite element analysis is necessary for complexgeometries such as in a burner-engine system. Engineersand researchers at Riello Burners have found that modalanalysis using ANSYS Mechanical software within theANSYS Workbench environment is extremely useful for a rigorous numerical treatment of alternate designs.

The engineering team applied a fixed constraint at the surface where the sleeve duct was anchored to the combustion chamber. For each body with a large mass, the engineers specified the appropriate physical constants,such as Young’s modulus and Poisson’s ratio, using the engineering data section of the ANSYS Workbench interface. When necessary, the team modified these valuesusing the material property form, which allows users tospecify suitable data in the physical characteristics field.

An initial simulation using the original burner design provided the same results as experiments with regard to the values of vibrational frequencies. In particular, as computed by the software, a value of 49 hertz demonstratedthe need to accurately balance the electric engine to avoidwhirling effects caused by a rotational speed of only 48 revolutions per second. Also, the smallest computed valueof 29 hertz is almost equal to the natural frequency of thecombustion tube (boiler). When designing and engineering ahigh power burner, the geometric and physical properties of the boiler — including the dimensions of the tube, thephysical characteristics of the materials and the water mass


flow — must be considered very carefully so thatthe boiler’s vibrations have a natural frequencythat is not close to that of the burner in order toavoid resonance problems.

To understand and include the influence ofcombustion on the burner’s mechanical vibrations,the engineering team considered and improved a mathematical model in which the flame’sperturbation and boiler’s geometry were combinedinto a unique Fourier series with frequenciesexpressed by:

where c was the speed of sound in the tube’s localenvironment (400 m/s), L the tube’s length (8 m)and m the natural frequency of concern. Using thisformula, f1 was calculated to be 25 hertz, a valuevery close to the smallest vibrational frequency of29 hertz computed by the software.

Using the harmonic response analysis module in the ANSYS Workbench platform,Riello engineers performed computations using sinusoidal loads with frequencies in the range of 5 to 100 hertz acting on the surface betweensleeve duct and combustion tube. This moduleallowed researchers to set values for dampingcoefficients to improve the accuracy of the simu-lation. The simulation results were confirmed by the data obtained from experiments for both

25

ENERGY: BURNERS

25

The geometry of a burner assembly: Air is taken in through the impeller and directed into the sleeveduct, which houses a combustion head (which initiates the flame).

Electric Motor

Sleeve Duct Impeller Volute

The study of structural dynamics is critical for understanding andevaluating the performance of any product. It is also essential for solving noise and vibration problems with existing structures. ANSYSmechanical solutions, including ANSYS Structural, ANSYS Mechanicaland ANSYS Professional software, provide all the necessary tools forsuch analyses.

Usually, vibration analysis begins with a modal analysis that estimates the natural vibration frequencies of a given structure. The frequencies of the structure can be determined from an unloaded state orfrom the loaded structure, as loads may shift the frequencies. Dependingon the environment of the structure, several advanced types of analysescan be performed: harmonic, spectrum, random vibration or transientdynamic analysis.

Harmonic response analysis is a technique used to determine thesteady-state response of a linear structure to loads that vary sinusoidally(harmonically) with time. The structure’s response is calculated at several frequencies and a graph of a response quantity (usually displace-ments) versus frequency is generated. Peak responses are then identifiedon the graph and stresses reviewed at those peak frequencies.

A spectrum analysis is one in which the results of a modal analysisare used with a known spectrum to calculate displacements and

ANSYS Software for Linear Dynamic Vibration Simulationstresses in the model. It is mainly used instead of a time-history analysisto determine the response of structures to random or time-dependentloading conditions such as earthquakes, wind loads, ocean wave loads,jet engine thrust and rocket motor vibration.

A random vibration analysis is similar to a spectrum analysis technique but is based on probability and statistics. It is used to analyzeloads that produce random time histories, such as acceleration loads during a rocket launch, that can be represented by a power spectrumdensity during each event.

Finally, a transient dynamic analysis is used to determine theresponse of a system under a given load variation over time.

ANSYS mechanical solutions allow the use of any of these tech-niques with various methods: direct analysis where the full matrices are assembled, mode superposition that reuses the results of a modalanalysis, or reduced methods that condense the problem to a smaller setof degrees of freedom. The last two options help reduce the compu-tational time. Another technique to reduce the computational time is the Component Mode Synthesis (CMS), which is used in reducing the size ofthe problem when large and complex assemblies are modeled.

– Pierre Thieffry, Product Manager, ANSYS, Inc.

Equivalent stress for impeller four-nodaldiameter mode at 141 hertz; at the edge ofthe hub’s central bearing surface, the localstress values are high and can be close tothe creep coefficient of the material.

Impeller four-nodal diameter mode associatedwith its first natural frequency; this value, 141hertz, is close to other vibrational frequenciesof the global system.


ENERGY: BURNERS

vibrational frequencies and displacement values. The peak displacement value for the mechanical system wascomputed at 30 hertz, confirming that the burner shows a structural resonance due to the first harmonic Fouriercomponent of the flame.

To eliminate the structurally dangerous lower vibrationalfrequencies, the team performed modal analysis for a set of possible design modifications to the structure of the components. After iterating through four new designs,the engineers reached a virtual configuration where the first natural frequency is sufficiently high that displacement ofthe global structure does not occur during normal opera-tion. In the end, by using the ANSYS Workbench platform,Riello Burners minimized the high cost for construction andtesting of intermediate prototypes, reducing the time todevelop an optimized model of the burner by approximatelyfive months. ■

References[1] Den Hartog, J.P., Mechanical Vibrations, Dover: New York, 1985.

[2] Strogatz, S.H., Abrams, D.M., McRobie, A., Eckhardt, B., Ott, E.,“Theoretical Mechanics: Crowd Synchrony on the Millennium Bridge,”Nature 2005, Vol. 438, pp. 43–44.

[3] Doria, A., “A Simple Method for the Analysis of Deep Cavity and LongNeck Acoustic Resonators,” Journal of Sound and Vibration 2000,Vol. 232 (4), pp. 823–833.

The new design of the volute showing reinforcement by ribs at upper and lowerparallel surfaces and by new material at the engine flange (modifications in green)


SPORTS

www.ansys.com 27

Speedbikes are the Formula One equivalent for humanpowered vehicles (HPV). They owe their ability to be fasterthan any other HPV to aerodynamics that are better than allother earth-bound vehicles. In order to achieve this state ofthe art, it is necessary to investigate both global and localaerodynamic effects and their interactions, in addition toanalyzing the human factor (cooling, breathing, vision fornavigation, safety). The similarities to automobile develop-ment are striking.

In 1994, the collaboration between Guido Mertens ofVRT-Speedbike e.V. and the Institute for Plastics Processing(IKV) at RWTH Aachen University (notably Johannes Dyck-hoff) led to the creation of the Speedbike Tomahawk 1. It wasdesigned to surpass the existing distance record over onehour. Its development revealed that speedbike design ingeneral had to address not only aerodynamics but alsoergonomics and driving stability. The results of this develop-ment process were several long distance records between1996 and 1999, including a record of over 82 kilometers inone hour set by rider Lars Teutenberg.

It became evident that, to further improve designs, eachcompetitor would have to decrease the frontal area, due toasymptotically improved drag values. The project Speed-hawk was launched as a cooperative effort between VRT-Speedbike e.V., ANSYS Germany and Adam OpelGmbH. The initial hull design for this new vehicle failed dramatically at the Speedchallenge 2004, which took placeat the Opel proving ground in Dudenhofen. This led to a significant redesign that used simulation to evaluate bothinternal and external factors.

The aim was to derive a new aerodynamic hull fromthe old one through the use of digitized point data. Theteam converted point data from a 3-D digitization thatwas performed at the Adam Opel GmbH Styling Centerinto regular surfaces with Autodesk® SurfaceStudio™.Parts without a direct effect on the air flow (redirectiongear under seat, chain and chain sheet) were neglected.

The simulation efforts that followed used FLUENT software. Researchers chose the RNG k-ε turbulence model because it offers a good compromise between

Picking Up SpeedSpeedbike designers use fluid simulation to gain a competitive edge.By Ralf Siber and Frank Werner, Adam Opel GmbH, Rüsselsheim, Germany

Guido Mertens, VRT-Speedbike e.V., Bergisch Gladbach, Germany and Marco Lanfrit, ANSYS Germany GmbH, Darmstadt, Germany

Speedbike designed in a cooperative effort between VRT-Speedbike e.V.,ANSYS Germany GmbH, and Adam Opel GmbH


SPORTS

computational accuracy, storage requirements and computing time. The mesh consisted of a five-tier prism surface layer on both the interior and exterior surfaces of thevehicle, while a hexcore mesh created in the TGrid tool filled the volume. Ground, tire, interior hull surface and riderclothing roughnesses were taken into account, while theouter hull was regarded as being hydraulically smooth.

Earlier tests showed that disk wheels themselves arevery aerodynamic, whereas in combination with wheel housings, the close proximity of these components makesthe parts act together like a friction pump. An enlargement ofthe wheel housings would disturb a large portion of the cleanlower hull flow. As a compromise, the physical vehicle wasdesigned using aero spokes.

For the correct computation of the complete model, theengineering team had to compute theexternal and internal flow in combination.This included simulation of the bow areas,the underbody and the rear of the vehicle.Researchers optimized the bow andunderbody profiles to minimize air con-gestion between the lower leading edge ofthe craft and the ground. The length of thetail was driven by the necessity to create agentle transition from the broadest part ofthe hull to the tail and to offer balanced control behavior in cross-wind conditions. In addition, the team simulated various windconditions to mimic real driving conditions.

An equally important emphasis for speedbikedevelopment was safety and comfort. While safety aspects can be considered by carry-overpreventive measures (Kevlar inlays within the

fiberglass/CFK skin or using the hood frame as a safetycage), taking rider comfort into account during developmentrequired a substantial effort. For this, air ventilation construc-tions had to be inserted and tested on a suitable test track atrunning conditions for each case. All information wasacquired subjectively from the rider.

The team designed the ventilation to occur passively,reducing interior cabin humidity and supplying cooling air tothe rider. In the Speedhawk, air flows into the ventilation system at a stagnation pressure point on the vehicle, is distributed through the interior and later escapes from thetail area. To simulate this effect, researchers computed the interior and exterior volumes in a coupled way so that both the flow resistance of the interior and the flow changeresulting from the addition of a passive ventilation system in

the front were accounted for simultaneously. In simulating the entire interior and exteriorof the Speedhawk together, the team

intended to significantly improve performance when compared to

their 2004 demonstration. The corrections made to the designhave resulted in a 10 percentimprovement in drag perform-ance so far, as well as muchmore significant driver comfort,boding well for the future performance of the vehicle.The molds for lamination will

be produced by Gaugler & LutzoHG in August 2008 and the team

looks forward to a finalized vehicle inSeptember 2008. ■

Revised ventilation designs (left) compared with earlier ones (right) led to greatly improved rider comfort; contour plots represent temperature.

Pressure contour on exterior side (left) and top (right) of a speedbike, assuming a 4-degree diagonal flow

Lars Teutenberg fits tightly into the initiallydesigned Speedhawk prototype.Image courtesy Berndt Photography


CHEMICAL PROCESSING

www.ansys.com 29


AUTOMOTIVE

Electromagnetics Comes Through in the ClutchBorgWarner engineers meet a tight deadline in optimizing the design of a revolutionary variable-torque clutch for all-wheel-drive vehicles.By Chris Kurmaniak, Senior Design Release Engineer, BorgWarner TorqTransfer Systems, Michigan, U.S.A.

All-wheel automotive drive systemshave evolved dramatically, becomingincreasingly popular in an expandingrange of vehicles. The technologybegan several years ago with relativelysimple devices that would positivelyengage and disengage a vehicle’s sec-ondary axle with the main transmissionand driveline. Systems are now morecomplex, with sophisticated controllersthat continuously monitor vehicle condi-tions and actively adjust driveline torquebalance to enhance vehicle stability andhandling. These systems are alsointegrated seamlessly with the operationof the engine, transmission, anti-lockbrakes and a myriad of other vehicle powertrain and safety subsystems.

One of the most recentadvances in all-wheel-drivesystems has been made by BorgWarner TorqTransferSystems (TTS) — a leadingglobal designer and producerof transfer cases and torquemanagement devices for all-wheel-drive passenger cars, crossovervehicles, sport-utility vehicles and lighttrucks. The company recently devel-oped the electromagnetically actuatedNexTrac™ active all-wheel-drive system, which provides a slipping con-nection with varying levels of torquetransmission between the front-wheel-drive transmission and rear axle.

BorgWarner TTS engineers initiallyattracted customer interest in NexTracwith concept-level hardware. The challenge then was to provide a set ofproduction prototypes for a customervehicle in three months. This required

re-engineering and optimizing thedevice’s electromagnetic solenoidactuator, assuring reliability and properfunction for a wide range of operatingenvironments. In parallel, engineershad to provide guidance for the design,tooling and validation of the manu-facturing processes.

Using electromagnetic capabili-ties in ANSYS Multiphysics software, engineers were able to achieve thesegoals, quickly evaluating design alter-natives, optimizing device operationthrough simulation, and studying theimpact of different material properties,

Based on electrical current from an electronic control unit (ECU), the armature of an electromagnetic solenoidactuator moves laterally to compress clutch plates separated by an organic friction material, thus applyingrequired torque to the rear axle. The stator (containing the coil winding) is bolted inside an aluminum castingfixed to the vehicle’s rear axle.

Armature

Electromagnetic Solenoid

Stator

Electronic Control Unit

manufacturing processes, operational environments and other designvariables. This minimized hardware pro-totype testing and enabled BorgWarnerto subsequently land the contract andquickly launch a new robust productwith market-leading performance.

NexTrac communicates with thevehicle’s electrical bus and is modulatedby a BorgWarner-supplied electroniccontrol unit (ECU) containing propri-etary control algorithms. According tothe level of electrical current provided,an electromagnetic solenoid actuatorcompresses clutch plates. This action

Clutch Plates

www.ansys.com 31

AUTOMOTIVE

structures. Software from ANSYSenabled BorgWarner TTS to moveforward quickly by including B-Hnonlinear material property datainto analysis models and com-paring results with linear andnative “reference steel” results.

The efficiency and interactivity ofthis process enabled analysts to gothrough multiple what-if iterations incomparing behavior of alternativematerials.

Another valuable feature of thetechnology from ANSYS was the ability to analyze the design’sresponse in the time domain.Through the design permutationprocess, BorgWarner engineers eval-

uated actuator reactions to typicalcontroller inputs and used themas a tool for optimization. Thisenabled them to balance staticdesign metrics with metrics ofdynamic response. This feature isparticularly important for prod-

ucts controlled in real time.Through successful analysis and

the ability to provide design direc-tion to its sub-suppliers, BorgWarnerTTS was able to meet customerdeadlines and deliver a highly refinedactuator design for maximum force capacity, improved linearity,reduced response time and reduced power consumption. In this way,ANSYS Multiphysics softwareenabled BorgWarner TTS to quicklydeliver next generation NexTractechnology and meet the demands ofa growing all-wheel-drive market. ■

applies the required torque to therear axle, enabling smoothengagement, exceptional torqueaccuracy and quick performanceresponse at a relatively low cost.Engineers from BorgWarner usedsoftware from ANSYS in studyingthe many design constraints andnonlinearities associated with the stator and armature that comprisethe device’s electromagnetic solenoidactuator — the core element ofNexTrac operation.

The engineering team used simulation in designing the variouselements of the actuator to meet arange of performance criteria. Funda-mentally, the actuator must achieve aspecified maximum force (whichdictates the maximum clutch torqueof the system) within the size andpower consumption constraintsspecified by the vehicle manufac-turer. It must also reach itsmaximum force within a mini-mum amount of time and proceedto maximum force bounded by aforce-to-electrical-current linearityrequirement. The assembly must bedesigned robustly with respect to temperature and manufacturing vari-ations. It must also survive a multitudeof durability test schedules and vehiclelevel validation testing, and accomplishall objectives at minimum weight andcost. The ANSYS Multiphysics tech-nology enabled BorgWarner engineersto quickly encompass all theseimportant design constraints andnonlinearities.

During the simulation process,engineers meshed and analyzed theconcept-level actuator to study electro-magnetic performance, identify keydesign deficiencies and guide develop-ment toward an optimal design. Theythen selected important geometry features, which were automatically iterated and meshed by custom script programs generated within the ANSYSMultiphysics software environment.

One major advantage of the customscript feature was that it enabled coilwinding characteristics to be inte-grated into the design permutation.

The ability to study the balance ofmagnetic circuit design (primarily fluxdensity and the number of coil windings that would fit in the coilpackage) was extremely valuable. Theengineering team also employed custom scripts to adjust temperatureand material properties in validating the design under vehicle and manu-facturing conditions. In addition, keydimensions — known for problems ordifficulty in manufacturing — werepermuted to study their effect on sys-tem performance before prototypeswere produced.

The flexibility of the materialproperty specification processproved to be very helpfulthroughout the project. Like many simulation prob-lems, accurately representing nonlinear material behavior isvitally important to producing relevant results. This is particu-larly true in the estimation of forcedeveloped in electromagnetic

NexTrac active all-wheel-drive system provides a slippingconnection with varying levels of torque transmissionbetween the vehicle’s front-wheel-drive transmission and the rear axle.

Front Axle

Rear Axle

Driveshaft

Front-Wheel-DriveTransmission

NexTrac Variable-Torque Clutch

BorgWarner engineers used electromagnetic capabilities inANSYS Multiphysics software in calculating the magnetic flux density of the armature.



In many organizations, design engineers perform FEA(finite element analysis) simulations using desktop systems.With this approach, high-powered systems are required eventhough they are typically only fully utilized for short periods oftime due to workflow interruption. Alternatively, solving morecomplex problems on shared back-end compute systems ismore efficient. Offering excellent price/performance ratio,Linux®-based clusters have become a common platform for these back-end computations. However, the concept ofachieving high performance through interconnected systemsintroduces performance and manageability challenges.

One of the biggest obstacles to a quick return on a cluster investment is the initial cluster setup. After deploy-ment, the system configuration on compute nodes needs toremain consistent. Even minor discrepancies — a misseddriver update on one system, for example — are hard to troubleshoot. In the case of a cluster upgrade or system failure on a compute node, the compute nodes need to bereprovisioned, leading to maintenance downtime. Moreover,it is difficult to identify and control processes in a cluster ascompared to a single system.

To better understand optimal methods for installing andrunning ANSYS Mechanical software on a Linux cluster, aseries of benchmark tests was performed by Penguin Computing and run on an Intel®‚ Cluster Ready certified cluster using Penguin Computing’s Relion® 1600 seriesservers, equipped with two dual-core Intel Xeon® 5160CPUs. Scyld ClusterWare™ 4.20, a cluster managementsolution from Penguin Computing, was installed on the cluster [1]. Scyld ClusterWare is designed to make the management of a Linux cluster as easy as the managementof a single desktop system and is fully compatible with Red Hat® Enterprise Linux®. With ClusterWare’s lightweightprovisioning, compute nodes boot over the network from amaster node into local memory. Avoiding a local operatingsystem (OS) installation on every compute node guaranteesconfiguration consistency, allows for easy node replacementand ensures cluster scalability. ClusterWare also provides aunified process space: All processes running in the clustercan be directly controlled from the master node.

Before an ANSYS Mechanical model can be solved inparallel, it must be decomposed so that the computationscan be distributed. A sparse matrix is generated when performing the structural analysis and has a resulting sizethat depends upon the number of degrees of freedom in themodel. Ideally, this matrix can be stored entirely in memory.In this case, the solver is run in the in-core (IC) mode. If the simulation run cannot be executed in-core, the ANSYSMechanical software writes the sparse matrix to a file, passing through the disk input/output (I/O) subsystem. Inthis latter case, the simulation is run out-of-core (OOC), andthe read/write I/O speed to the local scratch space greatlyimpacts solver performance.

Figure 1 shows performance results for ANSYS Mechanical benchmarks, tested using three memory config-urations. As expected, memory configuration had thebiggest impact on performance for the larger models —benchmarks 7 and 8. Significantly longer runtimes for thesetwo models with the 8GB memory configuration occurredbecause they were solved out-of-core. Moving from an 8GB to a 16GB configuration resulted in a 32 percent

PARTNERS

Higher Returns on theSimulation InvestmentOptimizing Linux clusters for ANSYS Mechanical software delivers fast turnaround on large problems.By Joshua Bernstein, Software Engineer and Arend Dittmer, Director Product Management, Penguin Computing, California, U.S.A.

Figure 1. In-core vs. out-of-core performance comparison

Figure 2. Relative performance of SATA vs. SAS disks

1 2 3 4 5 6 7 8

Runt

ime

(sec

)Ru

ntim

e (s

ec)

Benchmark Number

Single, Dual and Quad Disk Configuration

8GB16GB32GB

SATA

SAS

3000

2500

2000

1500

1000

500

0

3000

2500

2000

1500

1000

500

0

Figure 3. Distributed solver scalability

Number of Processes

Ethernet

Infiniband

1200

1000

800

600

400

200

01 2 4 8 16

Runt

ime

(sec

)

Figure 4. Performance comparison of round-robin vs. sequential core allocation

Number of Processes

Sequential

Round-Robin

Runt

ime

(sec

)

1200

1000

800

600

400

200

01 2 4 8 16


PARTNERS

performance increase as the problem was now solved in-core. While the benefit from solving FEA solutions in-coremay seem obvious, this characteristic is often overlookedwhen managing back-end compute systems.

To illustrate the influence of different local storage con-figurations on performance for an analysis model that runsout-of-core, benchmark 7 was run on a node configured with8GB of RAM. Figure 2 presents observed runtimes for twodifferent disk configurations. Serial attached SCSI (SAS) diskconfigurations performed up to 18 percent better than serialadvanced technology attachment (SATA) configurations.SAS drives spin considerably faster than SATA drives —15,500 RPM vs. 7,200 RPM — but their better I/O perform-ance is partially offset by their higher cost and smallercapacities. Moving from a single disk configuration to a two-disk redundant array of independent drives (RAID0)configuration yielded performance gains of 10 percent for SAS disks and 18 percent for SATA disks, indicating asignificant advantage for the parallel I/O provided by theRAID0 configuration.

Distributed ANSYS Mechanical technology spreads thecomputational workload of a single solver run across multi-ple systems. For benchmarking solver scalability, distributedbenchmark BMD-4 was chosen. Each node in the clusterwas configured with 8GB of RAM and used a RAID0 diskconfiguration consisting of four SAS drives.

Figure 3 illustrates the scalability of the BMD-4 bench-mark. The job scales well on Ethernet and Infiniband®

fabrics. Marginal performance improvement is noted for theInfiniband fabric relative to Ethernet results. The cores usedfor this set of benchmark runs were allocated round-robin:Each process was launched on one core on a different system. After four cores on four systems had been allocated,the algorithm wrapped around and allocated the next coreon the first node in the set, and so forth.

A second set of tests, shown in Figure 4, was performedusing a node packing method for distributing processesonto cores. When scaling across an increasing number ofcores, up to four processes were started on one node beforemoving to the next node. Better performance is achievedwith the round-robin allocation method, as round-robin max-imizes the amount of memory available for each process.

The presented benchmark results are useful for deter-mining the optimal set of conditions for running an individualjob. This does not consider the more realistic case in whichmany simultaneous ANSYS Mechanical runs need to beexecuted in parallel. Another option is to optimize for totalthroughput rather than for performance of individual jobs.When optimizing for throughput, the performance of individ-ual jobs of the same priority has to be balanced against thenumber of simultaneous jobs running on the cluster. Bestperformance for high-priority jobs is achieved with round-robin allocation of cores on nodes that are dedicatedexclusively to running one high-priority job at a time. ■

References[1] These benchmarks are described at http://www.ansys.com/services/

hardware-support-db.htm.

[2] To request the benchmark models, contact Shane Moeykens,[email protected].

Many users of ANSYS Mechanical software want to get thevery best performance out of their software and hardware combi-nation. Others wish to make the best hardware purchase decisions.Fortunately for both groups, ANSYS has put together a white paper on “Obtaining Optimal Performance in ANSYS 11.0.” Thisperformance guide provides a comprehensive resource for ANSYSMechanical users who wish to understand the factors that impactthe solution performance of this software on current hardware sys-tems. The guide provides information on ANSYS Mechanicalcomputing demands, hardware considerations, memory usage,parallel processing and I/O considerations. The guide also includesgeneral information on how to measure performance in the ANSYSMechanical solvers and an example-driven section showing how tooptimize performance for several ANSYS Mechanical analysis typesand equation solvers. The guide provides summary informationalong with detailed explanations for users who wish to push thelimits of performance on their hardware systems. Both Windows®

and UNIX® or Linux® operating system issues are covered through-out the guide. This guide can be downloaded from the ANSYSCustomer Portal.

— Ray Browell, Product ManagerANSYS, Inc.

How to Obtain Optimal Performance?


ACADEMIC

What began as a hobby for enthusiasts who pioneeredroad racing in France in the 1890s has evolved into a pro-fessional sport attracting millions of followers around theglobe. Grand Prix racing now takes place in Europe, theAsian Pacific rim, the Middle East, South America and theUnited States, with one of the most successful series beingFormula One (F1). The passion associated with F1 racinghas grown beyond the teams, drivers and technicians whoare personally involved with the sport. Broadcast to coun-tries worldwide, each of the 17 races in the 2007 seasonwas watched by an average of 597 million viewers.

Milad Mafi, a teenage student in Germany, has followedFormula One racing since childhood, and is much morethan a regular fan. Although only 16 years old, he is alreadyskilled at optimizing the aerodynamic design of F1 race carswith the aid of high performance computing (HPC) on theWindows® platform.

Aerodynamic ComputationsMilad started watching Formula One TV broad-

casts as a child and loved racing maneuvers suchas passing. Over time, however, he realized likemany others that these thrilling moments werebecoming rarer, and today are close to extinction.In modern F1 racing, changes in position most

often take place during pit stops, which have become partof the race strategy.

A reason for the the decreasing number of passingmaneuvers can be found in safety regulations imposed bythe governing body, the Fédération Internationale de l’Auto-mobile (FIA), and the resulting changes to the aerodynamicsof modern F1 race car designs. In the 1980s, road grip wasderived primarily from wide tires. With a shift to narrower tires, aerodynamic grip — which dependson several factors including the intensity of airturbulence — has become a much moreimportant factor than the mech-anical grip provided by the tires. The external aero-dynamics of modernFormula Onecars producea downwardforce of

Driven to SimulationA teenage student helps improve the aerodynamic design of F1 race carsusing ANSYS software and Windows high performance computing.By Eric Tierling, Freelance Writer, Germany and Shane Moeykens, Strategic Partnership Manager, ANSYS, Inc.

Pressure contours on the surfaceof a classic 1980s F1 race car

Image © Afby T1/dreamstime.com


ACADEMIC

roughly 25,000 Newtons (N), which corresponds to a gravitational force of 2.5 metric tons.

Maintaining laminar air flow across the wings and body of an F1 car is an ideal that is not realized in practice. Turbulence and eddies form around bluff components of acar body and trail behind the moving car, reducing efficiency.The wings of an F1 car may fail to fulfill their design functionin some highly turbulent flow conditions, producing little orno downward force while creating significant resistance. This necessitates that drivers reduce their curve speed andcurve acceleration, making it nearly impossible to execute apassing maneuver on the track.

More Passing ExcitementDriven by his disappointment in the decreasing number

of passing maneuvers, Milad started to research the problem. Where other teenagers prefer to play football, heexamined the design and aerodynamic behavior of F1 racecars — an exciting challenge for a boy who programmed hisfirst computer-aided design (CAD) solution when he wasjust 11 years old.

There are two primary approaches for exploring theaerodynamic behavior of an F1 race car: using a wind tunnelor performing fluids simulation. Efficiency, visualizationcapabilities and accuracy — not to mention the financial savings — make simulation analysis an excellent option.With the aid of Microsoft and ANSYS, Milad turned to an industry standard — PC servers for high performancecomputing. He was able to investigate turbulence effectsand the design modifications necessary to make passingmaneuvers on the track easier.

Microsoft’s Compute Cluster Server 2003 (predecessorof the current Windows HPC Server 2008) provided aneffective HPC software platform capable of performingcomplex flow simulations. Working with a Microsoft HPCpartner, Milad received access to a Windows HPC Servercluster having more than 250 CPUs — all tied together withthe Microsoft HPC solution. For the flow analysis, he used

An optimized F1 race car designbased on Milad’s research

The interaction between the rear wing and diffuser leads to a decrease in total pressurebehind an F1 race car.

FLUENT software from ANSYS, the same technology thatmany Formula One teams also rely upon.

By using this combination, Milad had a powerful simula-tion platform for his aerodynamic experiments. In addition,he took advantage of the comprehensive support providedby Microsoft and ANSYS, allowing him to obtain answersquickly and concentrate on his work. With the WindowsHPC platform and FLUENT software, he could analyze theproblematic components of F1 race cars and avoid cost-intensive and time-consuming physical tests.

Milad had the discipline to first investigate the compo-nents of interest — wings, diffuser, bargeboards, etc. — in2-D before moving to 3-D calculations. After weeks ofintense computations, his results indicated that the wingletsand the lower rear wing element, which interact with the diffuser, produce significant turbulence. Although it mightseem simple, this observation could influence the aero-dynamic design of modern F1 racing cars and yield betterpassing capabilities.

Windows HPC Enhances Racing ExcitementTrying to keep F1 racing exciting without compromising

safety, the FIA constantly makes rule changes aimed atincreasing the ability of cars to overtake each other. Oneapproach for achieving better passing capabilities mightbuild on Milad’s research and the resulting design optimiza-tions he performed on the front wing, rear wing, diffuser andbargeboards.

Given the fact that F1 race teams make significantinvestment in improving performance, it is not sur-

prising that Milad has already been incontact with several Formula One

teams. As a result, the nexttime you watch an F1 race,you might see car designsthat have been influencedby Milad’s observations. ■

3-D analysis of an O-ring elastomer seal employed in a valve cavity

Stretching Your Elastomer UnderstandingAccurate nonlinear analysis leads to a better material selection process that enables innovation and faster time to market.By Siddharth Shah, Product Manager, ANSYS, Inc.

Elastomers, or rubbers, are a category of engineering materi-als that are used in many critical applications and have propertiesthat are very distinct from commonly occurring solid materials.They exhibit a highly elastic nature, allowing them to be stretchedto many times their original length and, upon release, quicklyreturn to their original shape. This ability to significantly deformand, as a result, conform between complex adjacent surfacesmakes them very attractive for use in seals, sealants, gasketsand shock-absorbing applications.

For a good seal, the elastomeric part needs to maintain sufficient pressure against the sealing surface so that a leak isprevented. Since they are often expected to function at extremeconditions, it is critical to determine whether sufficient pressurecan be maintained.

Elastomers have the following characteristics:

• Ability to undergo large deformations and sustain strains in the range of 500 percent

• Highly nonlinear load-displacement or stress–strain relationship

• Nearly or fully incompressible — can undergo very little volumetric change under stress or cannot be compressed significantly under load

• Exhibit high energy absorption under cyclically varying load, providing excellent damping properties

• Highly dependent on temperature, operating frequency and duration of use

Elastomers have a wide range of applications and come inan even wider range of material types. Diverse and often conflicting design requirements make material specificationsdifficult. Selecting a material is often a complex process, withthe final choice dependent on a series of trade-offs and intangible factors. A material is often selected based on familiarity and experience that takes years to develop. As a

Large deformation analysis of an elastomeric-based automotive door seal

Imag

e ©

iSto

ckph

oto.

com

/Jam

es P

auls


ANALYSIS TOOLS

result, as newer elastomericmaterials are made available, acontinuous learning process is necessary.

Simulation can help aug-ment the understanding of an elastomer’s performance byproviding deep insight thatmay not be available throughphysical testing. With softwaretools, it is possible to studymany prototypes rapidly, accel-erating the understanding of thematerial. In this way, engineersare provided with a means tomake better material choicesand develop a more effectiveselection process.

SimulationWith any finite element analysis, the accuracy of

the material properties used is critical. However,because of the highly nonlinear and nearly incom-pressible attributes of elastomers, their mathematicalcharacterization assumes a central role in ensuring thequality of any analysis. Complex mathematical models,often referred to as hyperelastic material models, arerequired to accurately describe elastomer behaviorunder loading conditions.

Most elastomeric specimens need to be tested in alab to extract their stress–strain behavior. The goal isto acquire the stress–strain curves of the material inthe desired operating state and then find the matchingmaterial model to mimic that behavior. It is highly rec-ommended that more than one set of test data — suchas uniaxial tension, biaxial tension and shear test data— be used to identify the correct model for the material.

ANSYS Mechanical TechnologyTo fully rely upon a simulation tool for the material

selection process, the software needs to be accurate andhave an established record of excellent correlation withexperimental results. The mechanical suite of softwarefrom ANSYS has repeatedly proven to have all of the nec-essary features to perform quick and accurate simulationsof elastomeric components.

In ANSYS software, there is a wide choice of materialmodels backed by robust element technology sufficient tocover all possible combinations of natural and syntheticelastomers. To further enhance accuracy of simulations —such as predicting the damping behavior of elastomers— the hyperelastic material models can be freely combined with any of the viscoelastic material models.


ANALYSIS TOOLS

www.ansys.com 37

compression

unloading 1

unloading 2

unloading 3

unloading 4

unloading 5

Uniax ExpBiax Exp

Shear Exp

Uniax FitBiax Fit

Shear Fit

1.81

1.5

1.25

1.

0.75

0.5

0.25

0.

Stre

ss M

Pa

Strain mm/mm

0. 0.1 0.2 0.3 0.4 0.5 0.6 0.679

Screenshot of curve fitting setup (top) and samples of uniaxial,biaxial and shear fitted data using the Yeoh 3rd order curve fitting model available in the Engineering Data tool in the ANSYS Workbench platform (bottom)

Pressure vs. deflection loading curve for a typical gasket material

Material models available withANSYS mechanical tools

Hyperelastic

Neo-Hookean

Mooney-Rivlin

Polynomial

Ogden

Arruda-Boyce

Gent

Yeoh

Ogden (Foam)

Blatz-Ko

Anisotropic

User


ANALYSIS TOOLS

Cook Compression designs and manufactures inlet-dischargevalves and capacity-control equipment for industrial reciprocating gascompressors in refineries and petrochemical plants. In the industry,they have built a reputation for long life and efficiency. Historically,metals have been the primary material of choice for valve elements.In the early 1970s, Cook Compression pioneered engineered thermo-plastics as valve element materials.

Customers continue to demand a longer mean time between failures (MTBF) for their valve elements.This has been the main drivingforce behind the investigation of alternative materials. Today, Cook Compression is analyzing elastomeric materials for use in reciprocat-ing compressor valves using nonlinear FEA.

A compressor valve must open and close with each stroke of thecompressor (300 to 1500 rpm), forming a gas-tight seal when closedand allowing gas to flowthough the valve whenopen. Since elastomericmaterials have no strength,applying them in environ-ments in which differentialpressures exist is challeng-ing. To make things morecomplex, these valves alsooperate at differential pres-sures that cycle between zero and some value. Since compressorvalves are aerodynamic devices, shapes that promote efficient gasflow are desirable in order to reduce pressure losses as well as theload/power on the driver equipment.

Meeting the needs of dynamic differential pressure loading usingan aerodynamic shape made from a material with no inherent strengthis not an easy task. First, designers model the parts in 3-D with SolidEdge™ software and evaluate them for ease of manufacture. At thisstage, designers then conceptualize shapes that promote efficient flow,with consideration of negative parameters relevant to compressor

“The nonlinear capability of ANSYS Structural software has proven to be an invaluable tool in quickly

evaluating shapes and elastomeric polymers for use in compressor valves. Time to market is reduced, and the

accuracy of the FEA results provides the necessary confidence to spend money on prototype production.”

Kevin Durham, Director of Valve EngineeringCook Compression

valve design — such as fixed clearance, ease of repair in the field andthe robustness of the design to handle plant process upset conditions.

Once shapes are determined, elastomeric materials are selectedon the basis of their mechanical properties and their resistance tochemical attack. Candidates passing this last criterion get evaluatedusing nonlinear analysis in ANSYS Structural software to provideinsight into the deformation and stresses at operating temperatureand pressure. Shapes can be adjusted based on the analysis output,and the design evolution continues until a shape is deemed worthy ofa field trial.

There are millions of polymers that can be evaluated. Physicallytesting them all would be costly at best and impossible at worst.Being able to create a systematic method for polymer selection andthen having the capability to perform nonlinear FEA provides insight

into how the polymerbehaves under operat-ing conditions, which inturn provides feedbackabout how to improvethe selection process.In short order, many polymers and polymerfamilies can be elimi-nated, leaving only the

most promising candidates. Having a reliable simulation modelmakes analysis fast and accurate.

In the Cook Compression analysis, physical measurements of theprototypes matched the simulation-predicted deflections within 0.002inches. As a result, the world’s only compressor valves with elas-tomeric valve elements are operating successfully in a number oflocations. More work is being conducted to expand the operatingenvelope, and elastomeric designs are greatly increasing the MTBFwith lower valve pressure drops than their thermoplastic and metalcounterparts.

To aid in identifying the right material model forstress–strain data, software from ANSYS provides a veryefficient and lucid curve fitting tool. This tool, available inPrep7 as well as in Engineering Data, can account formuch of the experimental stress–strain data of the mate-rial under consideration and then quickly comparedifferent material models. The tool then automaticallymakes available the various mathematical constants thatcan be used by the material models.

Additionally, it is critical for simulation to account forcontact between elastomeric components. Often the com-ponents contact themselves as well as adjacent surfacesthat are made of materials other than rubber. The robust,automatic, surface–surface contact capability with mechan-ical software from ANSYS accounts for this nonlinearity. Thiscapability is not only robust, but is automatic, quick andstraightforward to use. ■

Cook Compression manufactures completecompressor valve assemblies such as this one.

Deformation contours for the prototype compressor valve

Cook Compression: World’s First Compressor Valve with Elastomeric Elements

Analyzing RandomVibration FatiguePowerful ANSYS Workbench tools help calculate the damage of vibrations that lack straightforward cyclic repetition.By Santhosh M. Kumar, Technical Support Engineer, ANSYS India

Determining the fatigue life of parts under periodic, sinu-soidal vibration is a fairly straightforward process in whichdamage content is calculated by multiplying the stressamplitude of each cycle from harmonic analysis with thenumber of cycles that the parts experience in the field. Thecomputation is relatively simple because the absolute valueof the vibration is highly predictable at any point in time.

Vibrations may be random in nature in a wide range ofapplications, however, such as vehicles traveling on roughroads or industrial equipment operating in the field wherearbitrary loads may be encountered. In these cases, instan-taneous vibration amplitudes are not highly predictable asthe amplitude at any point in time is not related to that at any other point in time. As shown in Figure 1, the lack ofperiodicity is apparent with random vibrations.

The complex nature of random vibrations is demon-strated with a Fourier analysis of the random time–historyshown in Figure 2, revealing that the random motion can berepresented as a series of many overlapping sine waves,with each curve cycling at its own frequency and amplitude.With these multiple frequencies occurring at the same time,the structural resonances of different components can be excited simultaneously, thus increasing the potentialdamage of random vibrations.

Statistical Measures of Random VibrationBecause of the mathematical complexity of working

with these overlapping sine curves to find instantaneousamplitude as an exact function of time, a more efficient wayof dealing with random vibrations is to use a statisticalprocess to determine the probability of the occurrence ofparticular amplitudes. In this type of approach, the randomvibration can be characterized using a mean, the standarddeviation and a probability distribution. Individual vibrationamplitudes are not determined. Rather, the amplitudes areaveraged over a large number of cycles and the cumulativeeffect determined for this time period. This provides a morepractical process for characterizing random vibrations thananalyzing an unimaginably large set of time–history data formany different vibration profiles.

Figure 1. Random vibrations measured for vehicle on a rough road showing perio-dicity for single, dual and quad disk configuration

Figure 2. Random time–history can be represented as a series of overlappingsinusoidal curves.

0.2

0.1

0

-0.1

-0.2

Ampl

itude

Ampl

itude

Ampl

itude

Time

Time

Time

24

12

0

0

0

1

2

3

Day 4

5

6


TIPS AND TRICKS

PSD Analysis Sample Problem To illustrate how power spectral density analysis is

used in calculating the fatigue life of a part undergoing random vibration, consider a cantilevered aluminum beam(Al 6061-T6 [E=68.9 GPa, γ=0.3]) that is 150 mm long by 15mm wide by 7mm high, as shown in Figure 5. This systemhas an overall damping ratio of 5 percent. An instrumentassembly of weight 2N is mounted on the tip of the beam,and its movement is restricted to only the vertical direction.The assembly must be capable of operating in a white-noiserandom vibration environment with an input PSD level of0.475 g2/Hz (from 20 to 200 Hz) for a period of 4.0 hours. Thechallenge is to determine the approximate dynamic stressand the expected fatigue life of the assembly.

An important aspect of such a statistical representa-tion is that most random processes follow a Gaussianprobability distribution. This distribution can be seen in a frequency-of-occurrence histogram (sometimes referred to as probability density function), which plots the num-ber of times random acceleration peaks reached certainlevels in small frequency segments called bins. The his-togram shown in Figure 3 represents a random signalmeasured for 10,000 seconds and indicates that thisrandom signal follows a classic bell-shaped Gaussianprobability distribution.

Representing the random signals in this manner issometimes called a zero-mean Gaussian process, sincethe mean value of the signals centers at zero of the his-togram, as do the random signal responses, which areusually described in terms of standard deviation (orsigma value) of the distribution. Figure 3 shows how theGaussian distribution relates to the magnitude of theacceleration levels expected for random vibration. Theinstantaneous acceleration will be between the +1σ andthe -1σ values 68.3 percent of the time. It will bebetween the +2σ and the -2σ values 95.4 percent of thetime. It will be between the +3σ and the -3σ values 99.73percent of the time. Note that the Gaussian probabilitydistribution does not indicate the random signal’s frequency content. That is the function of the powerspectral density analysis.

Power Spectral DensityThe usual way to describe the severity of damage for

random vibration is in terms of its power spectral density(PSD), a measure of a vibration signal’s power intensityin the frequency domain. Looking at the time–historyplot in Figure 4, it is not obvious how to evaluate theconstantly changing acceleration amplitude. The way toevaluate is to determine the average value of all theamplitudes within a given frequency range. Althoughacceleration amplitude at a given frequency constantlychanges, its average value tends to remain relativelyconstant. This powerful characteristic of the randomprocess provides a tool to easily reproduce random signals using a vibration test system.

Random vibration analysis is usually performed over alarge range of frequencies — from 20 to 2,000 Hz, forexample. Such a study does not look at a specific frequency or amplitude at a specific moment in time butrather statistically looks at a structure’s response to agiven random vibration environment. Certainly, we wantto know if there are any frequencies that cause a largerandom response at any natural frequencies, but mostlywe want to know the overall response of the structure.The square root of the area under the PSD curve (greyarea) in Figure 4 gives the root mean square (RMS) valueof the acceleration, or Grms, which is a qualitative meas-ure of intensity of vibration.

Figure 3. Gaussian distribution (right) of random signal (left)

Figure 4. Random time–history (left), power special density (PSD) of a random time-history (right)

Figure 5. Problem sketch of aluminum beam with a weight at the tip undergoingwhite-noise random vibration

Time (sec)

Time

Ampl

itude

PSD

(G2 /H

z)

Frequency (Hz)

Acce

lera

tion

(g)

No. of occurrence

3

2

1

0

-1

-2

-3

3

2

1

0

-1

-2

-30 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 0 100 200 300

Acceleration (g)


TIPS AND TRICKS

150

15

7 X

20 Frequency (Hz) 200

0.475

PSD(G2/Hz)

Analysis of the assembly under this white-noise environment results in a bending stress contour plot shownin Figure 6, which shows a maximum 1-σ bending stress of55.4 MPa (see accompanying table).

Reponse Power Spectral Density (RPSD)Figure 7 shows a response power spectral density plot

(new in Workbench 12.0) of a node at root having maximumbending stress at the system’s first natural frequency of ~56 Hz. The integration of the RPSD curve (the area underthe curve) yields variance of bending stress. The square rootof the variance is 1σ value of the bending stress.

Fatigue AnalysisFor fatigue life calculation in the sample problem, root

mean square (RMS) stress quantities are used in conjunctionwith the standard fatigue analysis procedure. The followingprocedure explains how to calculate the fatigue life using oneof the most common approaches: the Three-Band Techniqueusing Miner’s Cumulative Damage Ratio [1].

The first step is to determine the number of stresscycles needed to produce a fatigue failure. When the root ofthe beam is connected to the other parts of the structurewithout any fillet, the computed alternating stress has toaccount for stress concentration effects. The stress con-centration factor K can be used in the stress equation or indefining the slope b of the S-N fatigue curve for alternatingstresses. The stress concentration should be used onlyonce in either place. For this sample problem, a stress con-centration factor K = 2 will be used in the S-N fatigue curveas shown in Figure 8, where slope b = 6.4.

The approximate number of stress cycles N1 required toproduce a fatigue failure in the beam for the 1σ, 2σ and 3σstresses can be obtained from the following equation:

where:N2 = 1000 (S1000 reference point)

S2 = 310 MPa (stress to fail at S1000 reference point)

S1 = 55.4 (1σ RMS stress)

b = 6.4 (slope of fatigue line with stress concentration K = 2)

The number of cycles to fail (N) under dynamic stress iscalculated as follows:

Figure 6. 1-σ bending stress distribution

Standard Deviation Bending Stress Percentage of Occurrence

1σ stress 1x 55.4 = 55.4 MPa 68.3%

2σ stress 2x 55.4 = 110.8 MPa 27.1%

3σ stress 3x 55.4 = 166.2 MPa 4.33%

Figure 7. Response power spectral density of bending stress distributionfor aluminum beam

Figure 8. S-N curve for 6061-T6 aluminum beam with a stress concentration of 2

3.2591e+82.5e+8

1.e+8

5.e+7

2.5e.7

1.e+7

5.e+6

2.5e+6

1.e+6

5.e+5

2.5e+5

1.e+5

50000

2500017785

Response PSD -Normal Stress

56.426

2. 2.5 5. 10. 25. 50. 100. 200.

Frequency

K=1

b=6.4

310

Stre

ss (M

Pa)

K=2

103

51.7

103 N, Cycles to fail 108

PSD


TIPS AND TRICKS

41

damage is generated by the 3σ level, even though it actsonly about 4.33 percent of the time. The 3σ level generatesmore than two times as much damage as the 2σ level,which acts about 27.1 percent of the time.

The above fatigue cycle ratio shows that about 95.71percent of the life of the structure is used up by the four-hour vibration test. This means that 4.29 percent of the liferemains, with the expected life of the structure obtainedfrom the following calculation:

Used life + remaining life = 4.0 hrs + [(4.0) x (0.0429)] = ~4.17 hrs

While fatigue life evaluation under a random process ishighly complicated, Miner’s Rule provides a reasonablygood prediction. In the example, the safety factor of 2 calculated from structural stress values is not adequate toensure fatigue life of the beam for the chosen environment.When it comes to design for manufacturing, it is recom-mended that the beam design be changed to provide afatigue life of approximately 8 hours, amounting to a safetyfactor of 2 on the fatigue life. ■

Reference: [1] Steinberg, D.S., “Vibration Analysis for Electronic Equipment,”

John Wiley & Sons Inc., 2000.

The author would like to thank Eng Hui Khor, ANSYS, Inc., for his technicaladvice and editorial assistance.

Miner’s RuleMiner’s cumulative fatigue damage ratio is based on

the idea that every stress cycle uses up part of the fatiguelife of a structure, whether the stress cycle is due to sinu-soidal vibration, random vibration, thermal cycling, shockor acoustic noise.

Miner’s fatigue damage cycle ratio calculation is as follows:

The actual number of fatigue cycles (n) accumulated during four hours of vibration testing can be obtained fromthe percent of time exposure for the 1σ, 2σ and 3σ values:

An examination of the above fatigue cycle ratio showsthat the 1σ RMS level does very little damage even thoughit is in effect about 68.3 percent of the time. Most of the


TIPS AND TRICKS

4242

ANSYS Advantage • Volume II, Issue 3, 2008 434343www.ansys.com 4343

TIPS AND TRICKS

Units and Algebraic Operations in CELCEL is an algebraic expression language that follows standard

unit arithmetic. It allows CFX-Post (and CFX Solver) users to createpersistent, algebraic expressions to evaluate integrated quantities,new variable fields, complex material properties and boundary conditions. CEL variables include many common constants, all solution variables and any other expressions you have created,regardless of the order in which they are created. All variablesinclude appropriate dimensions. The CEL subsystem ensures thatquantities are converted to the same units before being added andflags errors if a user tries to add values that equate to differentdimensions. For instance, you can add 10 meters/second to 30 kilo-meters/hour but cannot add 10 meters/second to 30 kilometers.The same holds true for other available algebraic operations.

Reverse flow is indicated by the light blue contour region.

Extracting Solution-Dependent Regionsin CFX-PostIdentifying and quantifying regions of reverse flow in the CFX-Post fluids post-processor.By Robin Steed, Senior Fluids Application Specialist, ANSYS, Inc.

When evaluating computational fluid dynamics (CFD)results, engineers are often faced with the need to quantifyaspects of the flow based on specific behaviors in the solu-tion fields as opposed to geometric locations. One exampleis the need to visually identify and quantitatively evaluateregions of reversed air flow in an automotive vehicle cabin.This will be demonstrated using CFX-Post to evaluate a CFDsolution obtained from FLUENT software.

Visualizing Reversed FlowUsing the solution from the cabin airflow simulation, a

vertical plane through the vehicle was colored by velocity inthe x direction (velocity u). Deep blue regions in this contourplot suggest low speed or reverse flow. It’s difficult to tell,however, at exactly what point the flow is actually reversedbecause of the smooth color gradient. To locate regionswhere the velocity is in the negative x direction, you can sim-ply create a contour plot on the same plane, manuallyspecifying the contour boundaries at -100, 0 and 100 metersper second (m/s) in order to clearly delineate forward andreverse flow regions.

Quantifying Reversed FlowAlthough visualization is helpful in understanding the flow

field, it is also necessary to quantify the results for the purposes of evaluation, comparison and optimization. Youcan use the CFX Expression Language (CEL) to quantitativelyevaluate the proportional area of the reverse flow region, andthen evaluate the average temperature of forward andreversed flow on the plane.

Flow field on a plane down the middle of the cabin, colored by velocity in the x direction

In the example, to calculate the area proportion, you need to know the area of the plane (Plane 1) and the area of reversed flow. For this, you would use the area()@<location> function:

Plane area = area()@Plane 1

To calculate the area of reverse flow, you must first create a User Surface object and define it using the From Contour method, picking contour level 2 on Contour 1. This area is called User Surface 1. Using a similar approach, User Surface 2 is defined by pickingcontour level 3 on Contour 1. This provides you with alocation for your expression.


TIPS AND TRICKS

www.ansys.com44

The area of reverse flow is returned by the expression:

Reverse flow area = area()@User Surface 1

The area proportion is obtained by dividing the twoexpressions:

Reverse flow area proportion = Reverse flowarea / Plane area

Next, you would obtain the area weighted average tem-perature on the reverse and forward flow regions and thedifference between the two in a similar manner using theareaAve()@<location> function. The area averagefunction returns the area weighted average of a variable orexpression at the specified location.

Tave reverse = areaAve(Temperature) @User Surface 1

Tave forward = areaAve(Temperature) @User Surface 2

Tdiff = Tave reverse – Tave forward

Object PersistenceThe calculated objects in our example (Tave reverse,

Tave forward, Tdiff, etc.) have the dependencies on post- processing objects or geometries (in this case, the contourplot and the plane that the contour resides on). But whathappens if the plane is moved?

The real power of CFX-Post lies in the persistence of itsdata — meaning the calculated object data is preserved

Contour and user surface object geometry details

Reverse flow regions (in blue) at new plane location

Tdiff

Reverse Flow Area Proportion

Expressions

Plane

Tave Reverse

Tave Forward

Reverse Flow

Plane Area

Contour 1

Point Cloud

Vector 1

Streamline 1

User Surface

User Surface

when the input changes. If the plane is moved to a new loca-tion, the dependent object states are instantly updated,including to the level of the calculated expressions.

As you post-process your results, you can add figures,tables and comments to an HTML report, which can option-ally include 3-D viewer files [1]. Post-processing objectscreated during the session, including the report, can besaved to a CFX state file.

The CFX-Post state file is a text file containing only thepersistent data (parameters of objects), not the results. TheCFX-Post state allows a user to restore a post-processingsession or apply an existing post-processing state to newsimulation results without the need for scripting or journalfiles. This feature allows engineers to reduce time spent onpost-processing, compress the analysis process andincrease productivity. It is also the basis for the automaticextraction of quantitative results from ANSYS CFX softwarein a CFD simulation using ANSYS DesignXplorer software. ■

References[1] Free viewer download available at http://www.ansys.com/

products/cfx-viewer.asp

Quantitative data at original and new locations

Expression Original Location New Location

Plane area 1.588 [m2] 1.505 [m2]

Reverse flow area 0.972 [m2] 0.666 [m2]

Reverse flow area proportion 0.612 0.443

Tave forward 286.300 [K] 287.708 [K]

Tave reverse 286.861 [K] 290.798 [K]

Tdiff 0.561 [K] 3.091 [K]

Dependencies for user-created objects and expressions.

OUTSIDE THE BOX

Bio-InspiringEngineeringScientists use nature to advance technology.By ANSYS Advantage Editorial Staff with the assistance of Matevz Dular, University of Ljubljana, Slovenia

Animals are naturally engineered to be highlyefficient. Fireflies emit a light that essentially produces no heat. Cockroaches and crickets areable to traverse complex and uneven terrain, a traitworth imitating if you’re developing a robot thatneeds to navigate unknown and rough territory.Studying natural solutions to the physical world is not new and can often lead to important and surprising technological advances.

Within the study of locomotion, observations inthe animal kingdom have led to a number of engineering developments that may have other-wise been overlooked. The rough textures found onshark skin would intuitively introduce extra draginto the swimming process. When swimsuits aredesigned with similar surfaces, however, they provemore efficient than other alternatives. In anotherexample, penguins provided engineering insightwhen researchers at the Massachusetts Institute ofTechnology developed a marine propulsion systemthat uses two oscillating foils to create thrust in the water. The system boasts significantly higher efficiencies than the traditional propeller system.

In order to fully understand any propulsion system, familiarity with the fluids that are involved isessential. For Matevz Dular, who was collaboratingwith the Marine Biology Station Piran in Slovenia,simulation was an effective tool to employ whenwanting to learn more about jellyfish locomotion.As Dular puts it, the animals are usually not verykeen to cooperate experimentally, and simulationsolved this problem. Using FLUENT simulationsoftware, he was able to help researchers visualizethe flow patterns related to the animals’ move-ments, gaining insight into the dynamics of the bellcontraction and relaxation.

Dular utilized a user-defined function (UDF)to describe the movement of the jellyfish bell, adynamic mesh capability to describe the meshmotion, and another UDF to calculate the forcesthat the bell imposed on the surrounding fluid asit moved. With the resultant force, he was able tocalculate the acceleration of the jellyfish in eachtime step. The jellyfish acceleration was then used

to assign the velocity boundary condition for the fluid moving past the animal. By using two phases of liquids in the simulation — both with identical properties — Dular was able to introduce a virtualcolor tracer in order to visualize the vortices that wereproduced by the jellyfish locomotion.

From an engineering point of view, jellyfish propulsion is interesting because it was perfectedover millions of years. With simulation, we can further our understanding of how it is that animalsare so efficient and try to learn from those natural designs. So what other natural engineering solutions are waiting to be discovered? Humming-bird-inspired new high-tech flight techniques withaerospace application? Insect-inspired drug deliverymethodologies within the pharmaceutical industry?Time will tell. ■


Image © iStockphoto.com/Klaas Lingbeek-van Kranen

Documents

Aa v2 i3 Ansys Advantage