62438, page 1-44 @ pdfready - mapdataweapons, armor, vehicles, ships and other products that respond...

44
ANSYS software helps increase efficiency in aluminum processing. New technology handles million-degree-of-freedom models with shorter response times on a single 64-bit workstation. CEI’s EnSight software brings high-end visualization to the desktop. Industry Spotlight Image courtesy of ISL.

Upload: others

Post on 04-Apr-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

ANSYS software helps increase efficiency in aluminum processing.

New technology handles million-degree-of-freedom modelswith shorter response times on asingle 64-bit workstation.

CEI’s EnSight software bringshigh-end visualization to the desktop.

Industry Spotlight

Imag

e co

urte

sy o

f IS

L.

Page 2: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community
Page 3: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

For ANSYS, Inc. sales information, call 1.866.267.9724, or visit www.ansys.com.To subscribe to ANSYS Solutions, go to www.ansys.com/subscribe.

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

Editorial DirectorJohn [email protected]

Managing EditorFran [email protected]

DesignersMiller Creative [email protected]

Art DirectorDan [email protected]

Ad Sales ManagerBeth [email protected]

Circulation ManagerElaine [email protected]

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Neither ANSYS, Inc. nor the editorial director nor Miller Creative Group guarantees or warrants accuracy or completeness of the material contained in this publication. ANSYS,ANSYS Workbench, CFX, AUTODYN, FLUENT and any and all ANSYS, Inc. product and service names are registered trademarks or trademarks of ANSYS, Inc. or its subsidiarieslocated in the United States or other countries. ICEM CFD is a trademark licensed by ANSYS, Inc. All other trademarks or registered trademarks are the property of their respectiveowners. POSTMASTER: Send change of address to ANSYS, Inc., Southpointe, 275 Technology Drive, Canonsburg, PA 15317 USA.

©2006 ANSYS, Inc. All rights reserved.

Editorial AdvisorKelly [email protected]

Editorial ContributorChris [email protected]

Solving Large ModelsFaster Using ANSYS forWindows XP x64New technology enables users tohandle million-degree-of-freedommodels with shorter response timeson a single 64-bit workstation.

Extreme Visualization for EveryoneCEI’s EnSight software brings newlevels of animation, realism anddetailed display of simulation resultsto the desktops of engineers andanalysts.

ANSYS Nonlinear TechnologyPowerful new capabilities are aimedat studying complex nonlinearbehavior in mechanical systems.

ContentsIndustry Spotlight

Features

Military and Defense SystemsEngineering simulation is an indispensable tool in developingweapons, armor, vehicles, ships andother products that respond to newthreats in an ever-changing world.

11

4

15

18

Departments

CAE CommunityJoin the ANSYS Workbench Community Forum . . . . 31

Simulation at WorkPredicting Thermal–Hydraulic Behavior in Nuclear Reactor Cooling Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Improving Efficiency of Vacuum Cleaner Fans. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27Leveraging Simulation throughout Product Development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Tech FilePractical Considerations in Using Beam Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37

Industry NewsAnnouncements and Upcoming Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Tips and TechniquesUsing the ANSYS Workbench Remote Solution Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39

EditorialHigh-End Visualization is Far More than Pretty Pictures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Increasing Efficiency inAluminum ProcessingANSYS Multiphysics with ANSYSCFX helps improve electrochemicalreduction.

21

Software UpdateTools from Harvard Thermal Now Part of the ANSYS Suite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33Guest CommentaryMaking Trade-Off Decisions in the Chaos of Product Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35

In developing military anddefense systems, researchersare working to improve per-formance of spinning artilleryshells. ANSYS software wasused on this projectile designto induce spin, reduce dragand improve range. Readmore in this issue’s IndustrySpotlight article beginning on page 4.

About the Cover

Page 4: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Editorial

2 High-End Visualization is Far More than Pretty Pictures

When you see high-end 3-Dvisualization in engineering andscientific applications, you’reimmediately struck by theincredible detail, color tones,smooth animation and photo-realistic quality that rival computer-generated scenesfrom a Hollywood movie. Thesevivid displays go far beyondbeing fascinating and pretty pictures, however. They provideinsightful perspectives into theoverall behavior of complexphenomena and show the

interaction of multiple parameters that, otherwise,would be difficult to grasp.

Engineering applications that utilize these capabilities cover a range of disciplines, including fluidflow, crash simulation, multibody dynamics, structuralmechanics, thermal studies, coupled-physics problems and other types of analysis that demanddetailed graphics displays of giant result files. Of course, routine work in these disciplines can beperformed just fine for most analysis problems. But conventional post-processors can choke on behemoth problems involving complex interactionsover time, with files containing hundreds of gigabytes(or even terabytes) of data representing some of thebiggest solutions in engineering and scientific analysis.

Complex problems are made-to-order for high-end visualization, which uses features such as sophisticated command languages, reading of multiple data sets, isolation of parts and data elements, and advanced animation techniques tocrunch through data and produce stunning graphicaloutput on ordinary workstations, desktop PCs and even laptops. For greater immersion into data, visualization software can provide parallel processing,distributed rendering and support for virtual reality displays such as multi-panel display walls.

As Bob Cramblitt explains in this issue’s article“Extreme Visualization for Everyone,” high-end visualization capabilities that once were available to a select few now are accessible to almost anyone, on any platform, for any simulation application. Solutions such as EnSight from Computational Engineering International (CEI) are particularly

By John KrouseEditorial DirectorANSYS [email protected]

Technology enables simulation results to be readily displayed and shared acrossthe extended enterprise.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

well-suited to coupled simulations since they canreadily combine results from multiple programs, even ifthey’re from different vendors.

In this way, high-end visualization has significantbenefits in enabling engineers and analysts who workon such problems to fluently study, interpret and interact with complex results. More broadly, results areeasily shared with others through CEI’s free geometryviewer called EnLiten, which runs on ordinary computers and requires no training. The viewer allowsanyone to see, analyze, manipulate and annotatecomplex visualization scenarios sent to them.

The capability for engineers and analysts to sharesimulation results with others has profound ramifica-tions in facilitating greater collaboration in productdevelopment. Colleagues a few desks away — oraround the world — can exchange data, brainstormalternative ideas, compare notes and generally worktogether more closely than might otherwise be possible. High-end visualization technology enablessimulation results to be leveraged beyond the boundaries of engineering and analysis. In cross-functional product development teams, people ingroups such as sales, marketing, inspection and production can better understand product behaviorand clearly see the impacts of changing a design, eventhough they have little familiarity with CAD or CAE.Likewise, results can be viewed by customers to gainbetter insight into product performance, or by suppliers to understand the functional requirements tobe met by their particular component or subsystem. In addition, simulation results now can be integral inbids, quotes, proposals and presentations.

In this way, high-end visualization has the potential to powerfully leverage engineering simulationacross the extended enterprise and bring significantbusiness value to companies at which simulation isalready an important part of design. By utilizing simulation results as a basis for collaborative productdevelopment and a foundation for communicatingproduct behavior throughout the supply chain, suchan approach enables companies to take full advantage of simulation data instead of leaving itlocked away with the group that created it.

Page 5: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

3

Industry News

Recent Announcements and Upcoming Events

ANSYS Completes Acquisition of FluentOn May 1, 2006, ANSYS announced that it has completedits acquisition of Fluent, which is now a wholly owned subsidiary of ANSYS, Inc. Fluent products utilize compu-tational fluid dynamics (CFD) principles and techniques toenable engineers and designers to simulate fluid flow, heatand mass transfer and related phenomena involving turbulent, reacting and multiphase flow. The company’sproducts are used by blue chip companies, small andmedium-size enterprises, and academic institutions andinstitutes around the world. Currently, ANSYS is focusingon unifying operations of the two companies.

As a result of the acquisition, ANSYS has significantlybroadened its offerings in the simulation market. The combination of ANSYS and Fluent software products andservices is expected to give ANSYS one of the most com-prehensive, independent engineering simulation softwareofferings in the industry, reaffirming and strengthening thecompany’s commitment to open interface and flexiblesimulation solutions that are driven primarily by customerdemand and choice. With more than 40 direct sales officesand 17 development centers on three continents, thecombined company employs approximately 1,400 people.

ANSYS Workbench Users Benefit from MatWeb IntegrationANSYS has partnered with MatWeb,® a leading provider oftechnical material data sheets, to provide seamlessaccess to material property data from within ANSYS Workbench. Increased material property data will allowANSYS users to more accurately simulate material behavior early in the design process, ultimately saving onmaterial costs of physical prototypes. ANSYS users canimport material properties from a growing collection ofmore than 55,000 material data sheets from MatWeb,including hundreds of various metals, plastics and ceramics.These data sheets include precise descriptions of material properties, the majority of which originate fromtesting by the manufacturers.

ANSYS and Granta Apply Materials Data to Empower Product DesignANSYS Workbench users now can easily import and applycritical materials data from GRANTA MI ,TM Granta DesignLtd.’s powerful “material intelligence” system. The

Upcoming Events

ASME PVP — 2006/ICPVT — 11 Conference Software Demonstration ForumJuly 25, 2006Vancouver, BC, Canada www.asmeconferences.org/PVP06/

5th World Congress of BiomechanicsJuly 29 – August 4, 2006Munich, Germanywww.wcb2006.org

Advanced Meshing Solutions for FEA and CFDAugust 3, 2006Derby, UK www.ansys.com/special/ris-admesh-derby/1.htm

31st International Symposium on CombustionAugust 6 – 11, 2006Heidelberg, Germanywww.combustion2006.org

CFD for Nuclear Reactor SafetySeptember 5 – 7, 2006Munich, Germanyhttp://www.nea.fr/html/nsd/workshops/CFD4NRS/index.html

Korean User ConferenceSeptember 14 – 15 Geongju, Koreawww.anst.co.kr

Schraubenmaschinen 2006 — Compressors, Superchargers, Engines, Vacuum PumpsSeptember 26 – 27, 2006

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

interoperability allows engineers in industries such as aerospace, defense, energy and medical devices to use thebest available information in product analysis and design,streamlining CAE collaboration. Customers in these industries realize multi-million-dollar benefits in reducedcost, enhanced performance, improved quality andspeedier design. GRANTA MI ensures data consistency,accuracy and traceability across the enterprise.

Page 6: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

4

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Industry Spotlight

Engineering simulation is an indispensabletool in developing weapons, armor, vehicles,ships and other products that respond tonew threats in an ever-changing world.

By Steve PilzProduct ManagerANSYS, Inc.

For centuries, innovative military and defense equipment

that utilized the most advanced technology of the day has determined the

fate of nations and the course of history. As early as the seventh century,

the Chinese combined potassium nitrate, sulfur and charcoal to create

gunpowder. Western societies augmented the invention while developing

explosive-based weapons, including guns, rockets and torpedoes. Middle

Eastern and European states made the most of advances in materials to

create metal shields and stronger swords, which were less likely to break

in combat, giving soldiers a decisive edge on the battlefield.

Military andDefense Systems

Throughout history, technology and effective military defenseprograms have been closely tied. Recent advances are beingmade in the areas of materials (metals and composites, forexample), electronics, ballistics and aerodynamics, non-lethalcountermeasures and robotics (cave penetrators, unmannedaerial vehicles, etc.).

Page 7: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

5

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

As the millennium unfolds, the military anddefense industry continues to grow. Sources such asthe U.S. Central Intelligence Agency report that international military expenditures are approaching $1 trillion. In 2000, after a decade of level expendi-tures, spending began trending upward as many of theworld’s larger militaries restructured to meet expecteddemands of smaller, more regional conflicts; to createlighter and more flexible anti-terrorism forces; and tofulfill the increased emphasis on homeland security.Countries in all corners of the globe strive for the mostadvanced equipment, including aircraft,missiles, tanks, armored trucks and otherground vehicles, ships and submarines,explosive devices, body armor, firearmsand military telecommunications systems.

Engineers face difficult design challenges in developing new productsand upgrading equipment for military anddefense applications. Equipment often is made of non-conventional materials; it must withstand adverseconditions well beyond the range of most commercialproducts in terms of extreme temperature, humidity,dust and other contamination, explosive impact andhigh vibration levels. Moreover, budgets usually arestrictly regulated for individual development programs,and schedules must be compressed to meet rapidchanges in domestic threat levels, global situations andregional conflicts.

Simulation enables engineers to meet theserequirements by studying product performance and behavior in the early stages of development, evaluating alternatives, running what-if scenarios andoptimizing designs. This approach is considerablyfaster, less expensive, and much safer than buildingand testing multiple hardware prototypes. Simulationenables engineers to refine concepts, pinpoint problems, evaluate alternatives and optimize designslong before the first hardware prototype is built.

For military and defense systems, simulation applications are wide-ranging and include:

■ penetration depth of a projectile into tank armor■ fragmentation of a grenade■ stress levels in ship hulls and other structures■ vibrations in military vehicle chassis■ dynamics of landing gear and other

complex subsystems■ burn patterns in missile propulsion systems

A broad range of simulation toolsare used in performing these studies.Finite element analysis (FEA) studiesstresses and fatigue, and explicitsolvers analyze high-velocity impactsand large deformations. Computationalfluid dynamics (CFD) simulates fluidflow. Computational electromagnetics

(CEM) predicts electromagnetic properties such as radarsignatures. Two-way fluid structure interaction (FSI)solves problems in which fluid flow is perturbed bydeformation of a solid structure as a result of fluid pressure load, thermal stress or electromagnetic force.

ANSYS is a leader in providing such simulationtools in an integrated platform, enabling users to workon a single geometry model in performing numerousmultiphysics studies. By streamlining the complete simulation process and integrating these tools in a unified suite of software, ANSYS avoids the problems ofusing many separate tools from different suppliers(including lack of compatibility and difficulties inexchanging files between systems). Such integratedtools have become an indispensable part of military anddefense system development in responding to newthreats in an ever-changing world.

Page 8: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Industry Spotlight

Armor for Protecting Soldiers and Equipment

Simulation plays a key role in developing a wide rangeof protective systems for safeguarding soldiers andshielding delicate equipment from harm in extremelyhostile environments. Engineers at the Institute of HighPerformance Computing (IHPC), for example, usedANSYS AUTODYN to help design a helmet withgreater impact strength. Combat field data indicatethat the head and neck receive up to 25 percent of all“hits” during combat, even though they representabout one-tenth of body area. The new IHPC helmet,designed to reduce injury from such impacts, is optimized to protect the wearer from the impact of a9mm full-metal jacketed projectile fired at close range.

Using an advanced material model for hyper-velocity impact in ANSYS AUOTDYN, the KEVLAR®

helmet and the metallic projectile were modeled usinga Lagrangian mesh. The composite material modeltook into account a number of complex phenomenarelated to the high-velocity impact of anisotropicmaterials: anisotropic strength degradation and material anisotropy; melting, vaporization and decom-position; shock response; and coupling of volumetricand deviatoric response.

Researchers charged with investigating multiplehelmet designs found that by conducting simulationsinstead of physical prototypes, they could more quickly find the best design among those proposed.By combining simulation and experiment, IHPC drastically reduced the development time and overhead cost for the new helmets.

In another set of applications, researchers in several NATO member countries are using ANSYSAUTODYN to design and refine lightweight, vehicle-borne armor systems, in which a sympathetic detonation is used to disrupt projectile penetration.Called explosive reactive armor (ERA), the systembreaks up the mass and changes the direction ofkinetic energy projectiles; it also disrupts the formation

of the plasma jet of a conventional-shaped charge.When a projectile hits this ERA, the explosive contained in the reactive armor system detonates,causing a base plate to separate from the armor system and break the projectile into multiple pieces,thereby rendering it less effective. New types of reactive armor include those that can withstand multiple hits, as well as “smart armor,” which usesarrayed sensors and reactive elements to intercept aprojectile at a distance from the equipped vehicle.

Simulation and test results for a KEVLAR® helmet show the extent of damage from a ballistic impact.Image courtesy of Institute of High Performance Computing in Singapore.

Predicted damage

DDaammaaggee oobbsseerrvveedd

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

EExxpplloossiivvee rreeaaccttiivvee aarrmmoorr ssyysstteemm ddiissrruuppttss aa kkiinneettiicc eenneerrggyy pprroojjeeccttiillee..

6

Page 9: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Developing Military Vehicles and Aircraft

Some of the most complex military and defenseequipment are vehicles and aircraft, which are developed through extensive use of analysis and simulation. In one recent project, engineers at GeneralDynamics, General Electric and Honeywell teamed upto develop a cool-running gas turbine engine for theAbrams tank and Crusader self-propelled howitzer.The goals included making the engine lighter andsmaller with rapid acceleration, reduced noise and astealthy exhaust.

Using meshing tools from ANSYS ICEM CFD,engineers created 3 million-element models for conjugate heat transfer analysis as well as radiationheat transfer simulations. A combination of solvers(including those in ANSYS) then were used to study

engine heat-up, steady-state run conditions and criticalthermal transients that occur when the engine is shutoff and forced airflow stops.

On another project, engineers used explicit technology from ANSYS to assess the survivability of awheeled armored personnel carrier that runs over aburied land mine with its leading tire. Testing this type ofevent with a physical prototype would have been time-and cost-prohibitive, as well as extremely dangerous to personnel.

The survivability simulation gave researchers perspective into the behavior of the vehicle under suchconditions. Simulations revealed, for instance, theunexpected impact of surrounding rocks — or evensandy soil — in immobilizing a vehicle and breaching itsprotective hull. Such understanding has led to designsthat are more effective at defeating buried and surfacemines made from explosive, such as C4.

Simulation also has also been used in assessingpotentially catastrophic damage from what the militaryrefers to as foreign object damage (FOD). At highspeeds, a bird the size of a goose can rip through aluminum, steel and even titanium. In one study, engineers used ANSYS LS-DYNA to predict the impactof a slow-flying goose sucked into the high-speedrotating blades of a turbine engine, which damaged thefirst stage of the engine severely enough to cut efficiency about 30 percent. A different bird-strike studypitted a military transport plane flying at cruising velocity against a goose winging in the opposite direction. Using a combination of the Lagrangian andsmooth particle hydrodynamic solvers, the simulationshowed that relatively soft objects become very stiff and can cause significant impact damage at high speeds.

Fluid flow and subsequent conjugate heat transfersimulations helped develop a tank turbine engine.

ANSYS LS-DYNA predicts the damage caused by a goose (the football-shaped object) impacting a turbine engine.

Lagrangian and smooth particle hydrodynamic solvers indicate theextent of bird-strike damage to an aluminum aircraft wing.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

7

Page 10: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Industry Spotlight

ANSYS CFX shows surface and crossflow streamlineson a long-range spinning artillery shell. Courtesy of ISL.

Computational fluid dynamics and structural analysis software from ANSYS predict airflow around stored missilesalong with heat gained via solar energy. Courtesy of BAE Systems.

Analysis of Projectiles and Missiles

To avoid the danger and inaccuracy of testing projectiles through trial and error, analysis tools are used extensively in studying a wide range of scenarios. In one such project, researchers at the Institute of Saint-Louis (ISL) used ANSYS CFX insteadof building and testing hardware prototypes to investigate the aeroballistic performance of proposedsteerable projectiles. After multiple simulations, they learned that canting the guiding fins to induceprojectile spin reduced drag, giving it longer range.

In another computational fluid dynamics study,BAE Systems engineers worked with the U.S. Navy inusing ANSYS CFX and ANSYS Mechanical to analyzeconditions inside a shipboard multiple missile launchertube. The combined application of this softwareallowed engineers to model not only the airflow aroundthe stored missiles, but also heat gained by the system via solar energy. The combined fluid and thermal analysis produced temperature profiles of the launcher and missiles, and ultimately will help the Navy determine ship locations to safely mount the launchers.

Simulation has been a key part of improving gunbarrel designs, such as a program under considerationto use heat- and corrosion-resistant, lightweightceramic liners that could increase the design life of anAbrams tank barrel from its present 280 rounds to

8

Page 11: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Stresses predicted for O-ring and C-ring ceramic gun barrel liners

Explicit technology reveals the effects of a land mine exploding under an armored vehicle.

9

1,500 rounds. This would make a significant impact onfleet maintenance costs. In a study sponsored by theU.S. Army Research Laboratory, efforts focused onincreasing barrel life 50 percent with a sustained firingaccuracy, as well as increasing muzzle kinetic energy20 percent and reducing weight between 5 and 20 percent. Consulting firm Connecticut Research Tech-nologies (CRT) was able to develop a methodology forpredicting the strength and probability of failure forthese gun barrel liners through the use of ANSYS and proprietary software. In their approach, theCARES/Life algorithm predicts component reliabilitybased on stress results from ANSYS software andCRT’s WeibPar program (which uses statistical parameter estimation techniques) to account for failuredata and component geometry. The work originallyproposed would optimize the structural design of theceramic gun barrel intended for use in the Abrams M1tank, although smaller calibers were considered as theprogram advanced.

Simulation as an Indispensable Tool

In developing military and defense systems, organizations around the world rely on advanced technology from ANSYS. Engineers and analysts trustANSYS, as they have for decades, in providingadvanced analysis and simulation tools to meetdesign challenges.

Powerful advances in technology have elevatedthe role of ANSYS from a stand-alone analysis tool fortroubleshooting problems near the end of design tothat of a simulation approach for optimizing designsthroughout the product development process. Byusing simulation to refine products up front in designrather than using costly and time-consuming multiplephysical prototypes, companies improve designs,lower costs, maintain quality, facilitate innovation andgenerally are able to complete the developmentprocess faster. These capabilities are especially critical in military and defense applications, in whichadvanced engineering simulation has become anindispensable tool. �

The author wishes to thank Chris Reeves and Simon Pereira of

ANSYS, Inc. for their efforts and contributions to this article.

Page 12: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community
Page 13: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

demonstrated the following main constraints in dealing with production-size models:

■ Model size is limited by the amount of realmemory used by 32-bit FEM programs (especially in their internal databases) in a 32-bit computing solution.

■ Solution times are reduced using multiprocessor platforms.

■ Hardware architecture bottlenecks in memoryand storage subsystems increase elapsedtimes.

To address the first issue, Trenitalia investigated64-bit technology from the older Alpha-based solutions to the current Itanium®-based ones, and even to the latest AMDTM64/EM64T platforms. Bothshared memory and massive parallel architecturetechnologies have been evaluated, with efficiencyrequirements and economical constraints suggestinga scalable SMP architecture. Economical, technicaland maintenance needs led Trenitalia to carefully evaluate the new x86-64 technology of the latest PC-based AMD64/EM64T systems so that companystaff could determine if the new systems were readyfor reliable high-performance computing solutions.

11

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Trenitalia is Italy’s national railway agency responsiblefor managing the development, construction andmaintenance of the rail transportation system in thecountry. In this capacity, the Trenitalia Technical andResearch Department uses ANSYS Mechanical for thefollowing activities:

■ Design optimization for new equipmentrequirements on existing locomotives, coaches and wagons of the Trenitalia fleet

■ Structural stress–strength checks to comply with safety transportation rules for new vehicles

■ Maintenance engineering planning for bogiesand/or body frames deteriorated by fatigueand corrosion phenomena

The need for larger analysis models and shortercomputer response times led Trenitalia to evaluatenew calculation solutions. In looking for more computational power to improve mechanical stresssimulation capability, the Trenitalia Information Technology Department, in cooperation with the Technical and Research Department, investigated how finite element programs interact with modern operating systems. This research activity

New technology enables users to handle million-degree-of-freedom models with shorterresponse times on a single 64-bit workstation.

By Antonio GhelardiniTechnical and Research DepartmentTrenitaliaFlorence, Italy

Solving Large Models Faster Using ANSYS for Windows® XP x64

Static strength resultsfrom stress analysisof a bogie frame for a high-speed car

Page 14: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

12 The architectural core improvements in x86-64capable processors allowed Trenitalia to get very good integer and floating point mathematical results, andthe full-duplex star topology of PCI-Express modernworkstations gave sufficient bandwidth to move gigabytes of data to/from memory from/to storagesubsystem. These systems were less expensive thantraditional solutions and were going to be mainstreamin the market, but would they be able to accomplishtheir tasks?

Benchmarking New SystemsIn December 2004, two 64-bit operating systems weretested on the same hardware (an ordinary mono-processor 2 GB AMD64 personal computer) with the same 0.35-million-degree-of-freedom (DOF)test model:

■ The most recent version of Linux (kernel 2.6.9)with a native 64-bit ANSYS 9.0 got an elapsedof 270 seconds.

■ The Release Candidate of Microsoft®

Windows® x64 Edition with the 32-bit versionof ANSYS 9.0 got an elapsed of 180 seconds.

SMP scalability, greater efficiency in thread andmemory management, and maintenance constraintsled Trenitalia to select the new MS Windows XP x64Edition operating system. Several field tests wereplanned in 2005 to evaluate reliability, even while thex64 operating system was still in a beta phase, withthe current versions available of Win 32 ANSYS on a 4 GB two-way AMD64 platform.

Trenitalia found that the x64 operating systemwas indeed the best operating system for “/3 GB compliant” (that is, large addresses) Win 32 profes-sional programs (ANSYS 9.0A1 Win 32). ANSYS 9.0A1(this is the code version identity in output logs ofANSYS 9.0 SP1) finally was able to manage up to 3.7 GB of memory. Nevertheless, what Trenitalia still needed was a true, native 64-bit version of ANSYS forWin x64 able to address more than 4 GB of memory.At last, Trenitalia could define their ideal reference platform: a two-way Win x64 SMP workstation able tosupport at least 16 GB of memory with a low-latencymemory and storage subsystem. Due to platform anddrivers constraints with more than 4 GB of memoryinstalled, it took a long time (until summer 2005) andconsiderable tuning efforts to acquire the first fullworking prototype of the reference platform.

Real production models were used to benchmark(thanks to side-by-side trials) with the latest Intel®- and AMD-based platforms. Trenitalia selected theAMD64 solution due to its actual greater efficiency in executing 64-bit programs, improved scalability onSMP systems and better performances while solvingvery complex models. The latest dual-core CPUs also

were tested in a real production environment with thesame production test models.

The widespread availability of the new high-performance 64-bit operating system, the quickimprovements in quality in 64-bit developing tools, theinvestments in ANSYS parallel compliant high performance sparse solver (implicit SMP solver withlarge memory capability), the huge amount of memoryaddressable by 64-bit technologies, and an ANSYS management investment in emerging 64-bittechnologies allowed ANSYS to quietly publish a 64-bit native ANSYS 10 Win x64 beta product at theend of summer 2005. Trenitalia jumped on the newprogram files and began to test the beta product.

ANSYS 10 Win x64 beta was able to use consid-erably more than 4 GB of memory, and Trenitaliaimmediately started to develop some million-degree-of-freedom production models. While solving themwith the beta build (the ANSYS Win 32 version was noteven able to open them), Trenitalia demonstratedsome important issues in the same beta product,especially those dealing with the efficiency of thesparse solver. After having notified the local ANSYSoffice of this limitation, Trenitalia was allowed to report feedback directly to the appropriate manager atANSYS and, even when ANSYS had only a few weeksto improve their code before the scheduled officialrelease, they quickly identified the problems and fixedthem in time for the official ANSYS 10.0 SP1 Win x64product. This effort enabled the sparse solver to gain30 percent in efficiency from the beta version.

Compressing Analysis TimePresently, Trenitalia uses solid elements for meshingdirectly imported 3-D CAD models from the CAD systems, thereby significantly reducing engineering

First natural frequency from modal analysis of railwaycar body frame

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 15: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

13time needed to prepare models for the solver. Trenitaliaalso is able to solve large models typically using thehighly efficient and memory-hungry in-core feature ofthe sparse solver. In addition, they finally are able tobrowse through large post-processed result files inreal time and efficiently study larger models made withexpensive cluster solutions.

ANSYS 10.0 SP1 Win x64 allows Trenitalia towork with finite element models five times larger than was previously possible, with eight times fasterresponse times. And the new ANSYS Workbenchmeshing functions are significantly improving Trenitaliapre-processing work.

New computing platforms and information technologies are tremendously improving the depart-ment’s capability in dealing with more complex and realistic simulation models, simultaneously reducingthe department’s analysis-related response times.

Changing the Way Engineers WorkThe new ANSYS 10.0 SP1 Win x64 has changed theway Trenitalia engineers are accustomed to working.To quickly check the structural strength of a particularclass of vehicles, they formerly used manually optimized and simplified vehicle models that tookabout an hour to solve with the old 32-bit platforms.Thanks to the new 64-bit computing systems, thesame models take only about two minutes to solve, soa full day’s activity has shrunk to about two hours ofwork for modeling, solving and post-processing.

Trenitalia already has planned the move to Winx64 platforms for its heavy-duty technical users nowutilizing Win 32 and Linux 32 platforms. WB-ANSYS10.0A1 Win x64 enables users at Trenitalia to deal withmulti-million DOF models, with reliable, inexpensive(less than $10,000) and simple computing solutions.

Stress Analysis

ANSYS batch run telcond— Telaio4 TelCond-668meshfinale 1,5 MDOF 1 MDOF0,35 MDOF

Supermicro H8DCE 9.0A1: [143] 95 9.0A1: [521] 396 9.0A1: [8366] 5251AMD Opteron 250-E4 2x2,4GHz 10.0A1: [118] 71 10.0A1: [390] 287 10.0A1: [6566] 3995

Supermicro H8DCE 9.0A1: [174] 83 9.0A1: [693] 336 9.0A1: [10520] 4124AMD Opteron 280-E6 2x2,4GHz 10.0A1: [152] 54 10.0A1: [516] 239 10.0A1: [8445] 3089Num_Proc=4

Supermicro X6DA8-G2 9.0A1: [150] 156 9.0A1: [608] 651 N/AIntel Xeon64 2 MB L22x3,6 GHz no/HT

Modal Analysis

ANSYS batch run cassa_modif_01 caso_12a 1,7 MDOF0,8 MDOF –m = 2984 for 9.0A1

Supermicro H8DCE 9.0A1: [330] 272 9.0A1: [2383] 3532AMD Opteron 250-E4 2x2,4GHz 10.0A1: [249] 262 10.0A1: [1897] 2091

Supermicro H8DCE 9.0A1: [410] 245 9.0A1: [2307] 2572AMD Opteron 280-E6 2x2,4GHz 10.0A1: [309] 241 10.0A1: [2274] 2064Num_Proc=4

Supermicro X6DA8-G2 N/A 9.0A1: [2303] 5620Intel Xeon64 2 MB L22x3,6 GHz no/HT

Trenitalia Field Test ResultsANSYS 9.0A1 Win 32 and ANSYS 10.0A1 Win x64 with Windows XP x64 Edition on Dual-Processor Platforms

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Notes:■ Operating system: Windows XP Pro x64 Edition

fully patched by Windows Update■ 16 GB of memory PC3200 or PC2-3200, 16 GB swap file■ ANSYS working directory is on a 4 sata disks Raid 0 volume ■ Mechanical stress model: telcond — meshfinale

(0,35 MDOF) and telaio4 (1,5 MDOF)■ Mechanical stress model not linear: TelCond-668 (1 MDOF)

■ Modal analysis:cassa_modif_01 (0,8 MDOF) and caso_12a (1,7 MDOF)

■ NUM_PROC = 2, SIZE_BIO = 65536 (only for 9.0A1)■ MEMORY REQ. (MB)-m = 3072 for 9.0A1 and –m = 8192

for 10.0A1■ DATABASE SIZE REQ. (MB) –db = 2040 for 9.0A1

and –db = 2048 for 10.0A1■ Times are in seconds with the following format:

[CP Time] Elapsed Time

Page 16: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

The Technical and Research Department of Trenitalia, the national railway agency for Italy, uses ANSYS Mechanical for a range of applications, including design optimization of new equipment, structural stress–strength checks and maintenance engineering planning.

14

ANSYS Workbench and ANSYS for Windows® XP Professional x64

A powerful new wave of computer, operating system andsoftware performance is about to be available at affordableprices. Although beneficial to the general public, this technology is especially helpful for ANSYS customers in performing amazing analysis on the desktop.

The term x64 is used to describe the 64-bit architecturedeveloped by Advanced Micro Devices (AMD) and Intel to provide processors that are highly compatible with the x86 processors that have been the mainstay of personalcomputing for decades. The biggest difference between x64and other 64-bit processors is that x64 processors are compatible at the hardware level with 32-bit, x86 processors.So the x64 architecture, when combined with Windows XPProfessional x64 Edition, can run the thousands of 32-bit programs available today.

Currently, there are two basic x64 processor families:■ AMD’s AMD64■ Intel’s EM64T

Windows XP Professional x64 Edition is a near feature-complete version of Windows XP Professional that runs onx64 processors. Windows XP Professional x64 Edition supports up to 128 GB of RAM and 8 terabytes of virtualmemory address space for a 64-bit process, as compared to4 GB of both physical RAM and virtual memory addressspace for 32-bit Windows XP Professional. Additionally, Windows XP Professional x64 Edition supports 1 terabyte ofsystem cache.

How much memory will be addressable byANSYS Workbench or ANSYS?

In Windows XP Professional x64, the physical RAM limit is128 Gbytes, while the virtual memory limit is 8 terabytes.ANSYS Workbench and ANSYS are built as a native 64-bitapplication and can access the full virtual memory addressrange. In reality, the physical RAM you have installed willmatter the most. On Windows XP Professional x64, the 10.0

SP1 sparse solver in ANSYS is limited to a contiguous blockof 16 Gbytes (or 32 Gbytes for complex solvers). The itera-tive solver (pcg) can grow memory as large as the virtualspace will allow, although it is not practical to grow largerthan the physical memory available on the system.

For those old enough to remember the UNIX migration from32-bit to 64-bit, you might ask if there is a speed penalty forrunning 32-bit applications on Windows XP Professionalx64. Unlike the transition from 32-bit UNIX to 64-bit UNIX, inwhich the solver was typically about 10 to 20 percent slower due to the increased IO (double the amount), theinternal architecture of x64-capable microprocessors hasdoubled its internal register number and improved the DMAengine for its IO to alleviate this problem. In fact many scenarios run from slightly faster to significantly faster dueto more memory being allocated to the operating system file cache and the internal registry-related improved processor efficiency.

When will ANSYS Workbench and ANSYS beavailable for Windows XP Professional x64?

Currently both are available with 10.0 SP1. DistributedANSYS for Windows XP Professional x64 will be availableat release 11.0. �

By Raymond Browell / Product Manager / ANSYS, Inc.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 17: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

NASA used it to provide insight into complex flowphysics for Discovery’s redesigned external fuel tank,to interpret flight and wind-tunnel data, and to designtests with smaller-scale models.

CRAFT Tech® uses it to study terabyte-sized fluiddynamic and combustive problems associated withweapons delivery systems for supersonic fighter jets.

Researchers at the National Energy TechnologyLaboratory and Pittsburgh Supercomputing Centeruse it to better understand complex hydrodynamicsthat can lead to safer and more energy-efficient pulpand paper manufacturing.

Embraer uses it to drive fully immersive virtualreality (VR) for engineering analysis, simulation and virtual testing.

And DaimlerChrysler used it to help fuel its triumphant return to NASCAR’s Nextel stock car circuit after a 25-year absence.

It’s extreme visualization, and, thanks to a combi-nation of greater computing power at lower costs,

major software advances, compatibility with major CAE solvers and ability to run on a variety of computingplatforms, the technology is revolutionizing the wayengineers and scientists see, analyze, communicateand interact with their computational results.

The Democratization of Visualization

Over the last 10 years, a remarkable transformation hasoccurred in high-end visualization: Capabilities that once were available only to a select few now areaccessible to almost anyone on any platform for anysimulation application.

Animated visualizations can be generated for CFD,FEA, crash analysis or even coupled simulations combining results from different vendors’ simulationprograms. Displays can range from a standard colorscreen to stereo using low-cost glasses to immersivevirtual reality devices such as PowerWallTM and CAVEs.Computing can be done on anything — from a laptopequipped with a decent graphics card to a desktop PC

CEI’s EnSight software brings new levels of animation,realism and detailed display of simulation results to thedesktops of engineers and analysts.

By Bob CramblittCramblitt and Company

Embraer is stretching virtual realitybeyond its traditional role of displayingdigital mockups into highly collaborativevisualization applications for engineeringanalysis, simulation and virtual testing.

Appearing on a Screen Near You:

Extreme Visualization for Everyone

15

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 18: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

of any flavor (Windows, Linux orMac OS X), to a high-end work-station cluster with dozens of nodes orhigh-capacity shared-memory processors.

The power of visualization is no longer limited to analysts or visualization specialists. Free software such as EnLiten from ComputationEngineering International (CEI) enables scientists and engineers to share their work with colleagues to createa greater understanding of problems and solutions.

A Quiet Revolution

The revolution in visualization did not announce itselfwith fanfare like the dot.com boom, and it didn’t sufferfrom the bloated expectations that made the dot.combubble burst. Instead, it occurred through steadybuilding over the years, and the peak isn’t anywhere insight. Advances in visualization and growing user benefits are making strong progress year by year.

“People are awakening to the power of visualiza-tion,” says Kent Misegades, CEI president, “and with that awakening come new applications, new customers and greater penetration into the mainstream of large engineering, research and scientific organizations.”

As the technological and market leader inextreme visualization, CEI serves as a good bench-mark for progress in this field. Here are just some ofthe milestones that have been achieved over the lastsix years:

■ Los Alamos National Laboratories and CEIbreak the 1-billion-cell barrier for visualizationin 1999 and reach 11.5 billion cells a year later. Eyes are now on 100 billion cells, inconceivable a decade ago.

■ Free tools such as EnLiten enable complex 3-D models and animations to be viewed andmanipulated by anyone within the enterprise— even novices who don’t have any visualiza-tion application on their desktops.

■ CEI’s EnSight becomes available on 64-bitcomputers, enabling more complex visualiza-tions with multivariate data to be cached ondesktop systems.

■ Extreme visualization comes to the AppleTM

Mac OS X, enabling animations generated on high-end servers to be shared with Mac usersat home or on the road. CEI now providesinterchangeable visualization tools for everycomputing platform and operating system.

■ A parallel rendering compositor developed byCEI achieves a world-record rendering speedof 3.17 billion polygons per second on a cluster of 76 standard PCs.

■ CEI folds its record-breaking distributed-rendering technology into a product calledEnSight DR, the first commercial visualizationapplication to bring parallel graphics to theuser’s desktop.

■ New features such as ray tracing and multiplelighting sources increase image realism, making it easier to communicate complexconcepts to non-technical audiences.

Complementary Technologies

High-end visualization on EnSight is easily accessibleto ANSYS users through freely available interfacesprovided by CEI. As a result, the strength of ANSYSsimulation technology is complemented by the capabilities of EnSight.

“ANSYS products and EnSight scale very well onhuge problems that are typical of the markets weserve,” says Marcus Reis of Engineering Simulationand Scientific Software (ESSS), a leading distributor ofANSYS and CEI software that provides simulation andvisualization solutions for companies such as GeneralMotors, Petrobras, Embraer, Embraco (Whirlpool),Electrolux and others. “Our customers have verydemanding applications that require them to be able

Visualization was an important tool inDaimlerChrysler’s return to NASCARafter 25 years. The company helped theracing team to better understand therole of aerodynamics in determiningwhy some cars move to the front, whileothers fall back.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

16

Page 19: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

to quickly and easily visualize multiple results files withextensive transient analysis from ANSYS CFX andother software.”

A recent project initiated by Silicon Graphics Inc.(SGI), CEI, ESSS and Matthew Koebbe, a consultantwith GADAB Engineering, shows how extreme visuali-zation can be used to understand fluid flows that cannotbe accurately predicted by other testing methods.

Embraco of Brazil, a supplier of compressors andother products for Whirlpool, wants to better under-stand the flow within the suction muffler of a small hermetic refrigerating compressor used in householdappliances. The challenge undertaken by Embracoresearchers Fabian Fagotti and Celso Kenzo Takemoriis to capture pressure fluctuations, such as noise, overa broad range of frequencies. Although this mightsound relatively simple, just simulating internal pres-sures up to 10 MHz at various points could require amesh of nearly 5 million nodes and 80,000 time steps.

ESSS is starting with a more coarse mesh andrefining the model to see how far state-of-the-art computer technology can go in addressing Embraco’sproblem. ANSYS CFX was chosen as the solverbecause of its reputation for resolving fine-scale turbulent behavior. EnSight was selected as the visualization tool for a number of reasons, including itshandling of complex and transient simulations, abilityto read in results of multiple simulation cases for directcomparison and support of a scripting language for batch processing. SGI PrismTM hardware was usedfor the initial computing work.

Early results for the project are promising. Simulations showed pressure behaving as expectedbased on real-world use, and phenomena that couldn’t be determined by experiments have beensimulated in the computational realm. Quantified estimates of leakage, for example, have been determined by analysis of mass flows. Color-codedpathlines generated by EnSight are providing a betterunderstanding of mixing within the compressor.

ESSS currently is working with Embraco to determine future directions for the project, althoughthe initial stage already has fulfilled a major benefit ofextreme visualization: enabling customers to see whatthey couldn’t see before.

From Extreme to Pervasive

Although tremendous progress has been made invisualizing complex problems, in some ways the mostexciting developments are still to come. Capabilitiessuch as parallel processing, distributed rendering,photorealistic imaging and highly sophisticated anima-tions not only are making their way to the desktop, butare being implemented so the complexities are hiddenfrom the user. There will be more going on than everbefore behind the nice, simple-to-use graphical userinterface (GUI) — but the user will be blissfullyunaware of it.

Some of the things that CEI is working on include greater parallel processing automation, GUI customization that speeds access to commonly usedfunctionality, 2-D texture maps that increase imagerealism, lower-cost software for small shops and consultancies, and greater flexibility in compiling andediting animated videos.

“What we call ‘extreme visualization’ today will be commonplace, transparent and pervasive in thenear future,” says Misegades. “We are automating the process to such an extent that ‘post-processing’will be an extinct phrase — any engineer, scientist or researcher will be able to take results from practically any solver and easily turn out beautiful and revealing 3-D animations with little effort and fewspecialized skills.” �

Bob Cramblitt is principal of Cramblitt and Company(www.cramco.com), a technology communications firm. Hisarticles on computer graphics, CAD/CAM, IT and other technology subjects have appeared in trade journals and Websites worldwide.

Results from ANSYS Workbench are visualized with CEI’s EnSightto compare natural frequencies and excitation forces.Image courtesy of ESSS.

Petrobras uses a combination of ANSYS CFX 10.0 and EnSight Goldsoftware to visualize, analyze and communicate CFD simulations of hydrocyclones, the technology employed in the oil industry to classify and perform liquid separation analysis. This image showsvelocity levels distinguished by color.Image courtesy of ESSS.

Web sites with more information

� Computational Engineering International (CEI)www.ensight.com

� Engineering Simulation and Scientific Software (ESSS)www.esss.com.br

� Silicon Graphics Inc. (SGI)www.sgi.com

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

17

Page 20: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Nonlinearities are common in most real-world problems and represent some of the most challengingaspects of engineering analysis. To help users modeland solve these problems, ANSYS has a wide range of features and capabilities for handling the most common types of nonlinearity.

Changing status or contact nonlinearity —Many common structural features exhibit nonlinearbehavior that is status-dependent. Status changesmight be directly related to load, or they might bedetermined by some external cause. Situations inwhich contact occurs are common to many differentnonlinear applications. Contact behavior, such as separation and sliding with frictional effects, introduces nonlinearity into the analysis.

Geometric nonlinearity — If a structure experiences large deformations, its changing geometric configuration can cause the structure to respond nonlinearly. Geometric nonlinearity is

characterized by “large” displacements and/or rotations. Small deflection and small strain analysisassume that displacements are small enough that the resulting stiffness changes are insignificant. In contrast, large strain analysis account for the stiffnesschanges that result from changes in an element’sshape and orientation. The large strain feature is available in most of the solid elements (including all of the large strain elements) as well as in most of the shell and beam elements. ANSYS also handlestwo other types of geometric nonlinearities: stressstiffening and spin softening.

For thin, highly stressed structures, such ascables and membranes, the out-of-plane stiffness of astructure can be affected significantly by the state ofin-plane stress in that structure. Stress stiffness is thecoupling between in-plane stress and transverse stiffness. Spin softening softens the stiffness matrix ofa rotating body for dynamic mass effects. The adjust-ment approximates the effects of geometry changes

due to large deflection circumferential motionin a small deflection analysis. Spin softeningis used in conjunction with prestressing,which is caused by centrifugal force in therotating body.

Material nonlinearity — Nonlinearstress–strain relationships are a commoncause of nonlinear structural behavior. Many factors can influence a material’sstress–strain properties, including load history (as in elastoplastic response), environmental conditions (such as temperature) and the amount of time that aload is applied (as in creep response).ANSYS handles numerous material-relatedfactors that cause a structure’s stiffness tochange during the course of an analysisranging from anisotropic behavior, nonlinearstress–strain relationships, dependency ontime, rate of strain and certain coupledphysics effects such as piezoelectric andSeebeck effects, to name a few.

By Achuth Rao, Ph.D.Product ManagerANSYS, Inc.

Nonlinear history tracking option monitors results in real time during solution.

ANSYS Nonlinear Technology

Powerful new capabilities are aimed at studyingcomplex nonlinear behavior in mechanical systems.

18

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 21: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Robust Solution Techniques

ANSYS employs the Newton-Raphson technique tosolve the previously mentioned types of nonlinearities,in which the out-of-balance load (the differencebetween the restoring forces and the applied loads) isused to perform a linear solution. ANSYS checks for convergence based on force, displacement or other criteria. If convergence criteria are not satisfied, the stiffness matrix is updated and a new solution is obtained.

A number of convergence-enhancement andrecovery features are offered by default such as line search, automatic load stepping and bisection. For special cases such as nonlinear buckling, ANSYS offers an alternative iteration scheme, the arc-length method, to help avoid bifurcation points and track unloading.

Latest ANSYS Capabilities

Recent releases of ANSYS have seen further advancesin nonlinearity and solution techniques for handlingthese types of nonlinear behavior.

Manual rezoning — In a finite large-deformationanalysis, mesh distortion reduces simulation accuracy,causes convergence difficulties and eventually can terminate an analysis. Rezoning allows you to repairthe distorted mesh and continue the simulation.ANSYS offers a manual rezoning procedure that allowsusers to decide when to use rezoning and whatregion(s) to rezone, and then to generate a new meshon the selected region(s). During the rezoning process,ANSYS updates the database as necessary, generatescontact elements if needed, transfers boundary conditions and loads from the original mesh and mapsall solved variables (node and element solutions) to thenew mesh automatically. Analysis then continues onthe new mesh, with equilibrium achieved based on themapped variables.

Nonlinear diagnostics — The nonlinear diagnosticstool in ANSYS can help you find problems in your modelwhen a nonlinear analysis has difficulty converging. Typically, nonlinear analysis fail to converge for the following reasons:

■ Too large a distortion

■ Elements contain nodes that have near-zero pivots (nonlinear analysis)

■ Too large a plastic or creep strain increment

■ Elements in which mixed u-P constraints are not satisfied

Tracking nonlinear residuals — As part of the nonlinear diagnostics, ANSYS allows tracking of the Newton-Raphson residuals during nonlinear iterations.Plotting the residual forces helps identify regions of highresidual forces. Such a capability is useful when you experience convergence difficulties in the middle of a loadstep, in which the model has a large number of contactsurfaces and other nonlinearities. Tracking the nonlinearresiduals allows one to focus on the nonlinearities in areaof interest, instead of having to deal with the entire model.Nonlinear diagnostics also allows one to identify elementsthat violate certain convergence criteria, such asplastic/creep strain increments and the like. The nonlinearhistory tracking option allows one to monitor results ofinterest in real time during solution. Before starting thesolution, you can request nodal data, such as displace-ments or reaction forces at specific nodes. You also canrequest element nodal data, such as stresses and strainsat specific elements, to be graphed.

Brake squeal analysis — The QR damped eigenvalue extraction method now can be used in problems with friction nonlinearities, in which an unsym-metric stiffness matrix may be produced. An example ofthis type of problem is brake squeal analysis, in which thecombination of ANSYS contact elements and theQRDAMP eigensolver provide an easy-to-use, efficient

Plotting Newton-Raphson residuals allows users to readily evaluate convergence difficulties.

19

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 22: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

means of determining unstable modes. ANSYS offersa two-step procedure in which the nonlinear unsymmetric stiffness terms due to frictional sliding ina static analysis are included in the eigensolution. Inbrake squeal analysis, the effect of the coefficient offriction (as well as other parameters) can be varied tosee the effects on different modes and the couplingbetween modes. This can help to determine whichmodes (frequencies) will be unstable and a source ofaudible discomfort.

Coupled physics — Due to interaction of variousphysics, coupled physics analysis is inherently nonlinear in nature. The interaction between variousphysics is typically either as a load or as a change inthe stiffness of the other physics. This type of inter-action makes the coupled system of equations nonlinear. ANSYS offers two types of coupled physicscapabilities: direct coupled physics and sequentialcoupled physics.

The direct method usually involves just oneanalysis that uses a coupled-field element type containing all necessary degrees of freedom. Couplingis handled by calculating element matrices or elementload vectors that contain all necessary terms. Anexample of this is a coupled physics analysis using thePLANE223, SOLID226 or SOLID227 elements. Userscan define material properties for these elements tomodel interaction such as piezoelectric, piezoresistive,Seebeck/Peltier effects and the piezocaloric effect.

The sequential method involves two or moresequential analysis, each belonging to a different field.The ANSYS Multi-field solver, available for a largeclass of coupled analysis problems, is an automatedtool for solving sequentially coupled field problems. It is built on the premise that each physics is createdas a field with an independent solid model and mesh. Coupled loads automatically are transferredacross dissimilar meshes by the solver. The solver isapplicable to static, harmonic and transient analysis,depending on the physics requirements. Any numberof fields may be solved in a sequential (or mixedsequential/simultaneous) manner. An application ofthe ANSYS Multi-field solver (MFX-Multiple codesolver) used for simulations with physics fields distributed between more than one product executable is the ANSYS Multiphysics and ANSYSCFX coupling for advanced FSI analysis. The solveruses iterative coupling in which each physics is solvedeither simultaneously or sequentially, and each matrixequation is solved separately. The solver iteratesbetween each physics field until loads transferredacross the physics interfaces converge.

In addition to some of the recent advances mentioned in this article, ANSYS continues toenhance its nonlinear capability. The next version ofANSYS will have further advances in areas of contactnonlinearity (line-surface contact, cohesive zonemodel using contact elements), material nonlinearity(Gurson’s material, anisotropic hyperelasticity), element or geometric nonlinearity (higher order shell,rebar elements) and convergence enhancementtechniques (stabilization). �

The author wishes to thank development and technical support personnel at ANSYS, Inc. and the various third-party solutions providers for their efforts and contribution tothis article.

Sequential analysis between ANSYS CFX and ANSYS Multiphysics provides for nonlinear coupled physics analysis of a MEMS micro-pump.

20

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 23: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

21

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Today’s aluminum industry is continually studyingways to increase the efficiency of reduction cells insmelters. Numerical simulation has become a highlyeffective tool for analyzing such complex processes.PCE Engenharia Ltda, S/S located in Porto Alegre,Brazil, has been using simulation for more than 10years. The industry consulting firm simulates equip-ment in the aluminum production chain and, in partner-ship with customers, develops new technological solutions for primary aluminum production.

Primary aluminum is obtained by a complexprocess of electrochemical reduction of alumina inHall-Héroult cells. DC current on the order of hundredsof thousands of amps flows from cell to cell in aluminum busbars. Inside each cell, current flowsdownward through the anodes, molten electrolyticbath, molten metal and cathode carbon block. Theelectrolytic bath floats on the top of the metal becauseof slightly different densities, and the two liquids do notmix. Undisturbed, the metal–bath interface would beflat and horizontal, but this is never the case in an

operating cell. The combination of the electric currentand the magnetic field generates volumetric forcesknown as Lorentz or electromagnetic forces. Theseset the metal and the bath in motion and deform themetal–bath interface.

Magnetohydrodynamics (MHD) is the sciencethat studies the effect of electromagnetic forces on

ANSYS Multiphysics with ANSYS CFX helps improveelectrochemical reduction.

By Dagoberto S. Severo, André F. Schneider,Elton C. V. Pinto and Vanderlei GusbertiPCE Engenharia S/S Ltda, Brazil

Increasing Efficiency in Aluminum Processing

Solid model of a 240kA cell

Page 24: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Magnetic model mesh with electric wireframe andsurrounding air box

22

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

problems, the ANSYS product has the magneticsscalar potential approach. Several options are available, but only general scalar potential is applicable to electrolysis cells to account properly forelectrical current going out through the steel shell. Theadvantage of magnetic scalar potential is that thesource conductors do not need to be part of the solidfinite element mesh, so that it was not necessary tomesh the entire smelter. Solid mesh is needed only inthe subject cell for a detailed electrical and magneticsolution [1]. At this point, the calculated magnetic field,the electric current density and the electromagneticforce are transferred from ANSYS Multiphysics toANSYS CFX computational fluid dynamics (CFD) software. A different mesh is required for the CFD simulation. Completely independent meshes can beused due to an interface program that was written tointerpolate and transfer data seamlessly. ANSYS CFXwas chosen because of the software’s robustness in

fluid flow. The resistive generation of heat in the bathlayer is proportional to the anode-to-cathode distance(ACD); therefore, this layer should be as thin as possible. However, the cell may become unstable at asmall ACD due to interface waves, resulting in currentefficiency loss. The minimum ACD at which the cellremains stable depends on cell design, busbar designand cell operation. Thorough understanding and control of cell MHD is the key to high current and energy efficiency.

The first step in MHD modeling is to obtain adetailed electromagnetic model. All relevant aspects of the cell are taken into account, such as external conductors (busbar arrangement), internal conductors(liquid layers, collector bars, anodes and cathodes)and the steel shell. ANSYS Multiphysics software was the natural choice for such a complex model dueto its flexibility with element types and physics phenomena representation. For magnetostatic field

Deformation of the interface between the bath and liquid aluminum

Magnetized shell colored by magnetic flux density (in Gauss)

Page 25: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Stream lines of the aluminum flow inside the cell

Vectors of the aluminum flow in the middle highof the aluminum

Vertical magnetic field (in miliTesla) at the aluminum

Slices showing the metal and bath distributioninside the cell

23

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

dealing with multiphase flow, high computational performance and easy-to-use data import ability.

Due to the nature of the flow system in an electrol-ysis cell, the fluids are expected to be completely stratified, separated by a distinct interface. The interface shape was modeled as free-surface two-phase flow. ANSYS CFX uses the homogeneousVOF (volume of fluid) method for interface tracking. In this method, each finite volume of the mesh is eitherfilled with metal or bath, or filled partially with metal andpartially with bath. The physical properties for eachfinite volume are weighted by the volume fraction ofeach fluid. The shape of the interface is supplied by thegeometric location of the finite volumes with 0.5 of volume fraction of each liquid. Steady-state modelsplay an important role in cell MHD design, since theygive relevant information about the mean values of flowpatterns, the metal bath interface deformation and theirsymmetry. The flow within a reduction cell has to satisfytwo contradictory requirements for the process efficiency. The alumina distribution and dissolution inthe bath requires high bath flow, whereas the currentefficiency suffers from high bath and metal flow. It is difficult or impossible to slow down the metal withoutslowing down the bath. In this situation, considering the importance of current efficiency, the criterion ofminimizing the metal flow should prevail.

Most competitive smelters around the world areable to work with current efficiency up to 95 percent,and an increase of 1 percent at this value means hundreds of Megawatts/hour of energy savings andmillions of dollars in increased aluminum production.Any modification in a live cell row demands hard andexpensive logistical work. Up-front simulation allowsfor virtual testing of many designs and operating conditions in a short time and at low cost beforeexpensive prototypes are constructed. A true multifield, coupled physics system involving fluids,electromagnetism and MHD in electrolysis cells is anarea in which simulation-driven design is strategicallynecessary. By using ANSYS Multiphysics and ANSYSCFX, PCE is improving its understanding of this complicated process and reducing time and costs. �

Reference:

1. D. S. Severo, A. F. Schneider, E. C. V. Pinto, V. Gusberti, V.Potocnik, “Modeling Magnetohydrodynamics of AluminumElectrolysis Cells with ANSYS and CFX,” Light Metals, (2005).

Page 26: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

24

Simulation at Work

ANSYS CFX simulates large temperature gradients in light water reactors.

Predicting Thermal–HydraulicBehavior in Nuclear ReactorCooling Systems

Gesellschaft für Anlagen- und Reaktorsicherheit (GRS)is Germany’s central scientific–technical expert organization for all issues related to nuclear safety andnuclear waste management. The Nuclear Researchand Consultancy Group (NRG) is the Dutch center ofexcellence, developing knowledge, products andprocesses for safe application of nuclear technologyfor energy, environment and health. The Plant Perfor-mance and Technology unit provides consultancy andservices for safe and efficient use of nuclear reactorsand non-nuclear industrial installations.

Among the activities of GRS and NRG are projects on nuclear power plant life extension, severe

accident management and new reactor conceptassessment. In all these areas, an accurate predictionof the thermal–hydraulic behavior of light water reactors (LWRs) has become ever more important.Since traditional system codes do not predict thethree-dimensional behavior of these flows well, theinterest in the use of computational fluid dynamics(CFD) for such applications has increased significantly.However, in order for utilities and the regulatory agencies to have full confidence in CFD predictions,validation with high-quality experimental data isrequired. This need for validation has been addressedin a number of recent projects, including the ECORA(Evaluation of Computational Fluid Dynamic Methodsfor Reactor Safety Analysis) project in which GRS,NRG, ANSYS Germany and others are partners. Partof the ECORA project is the comparison between CFDand experimental results from the full-scale UpperPlenum Test Facility (UPTF).

By Martina ScheuererGesellschaft für Anlagen- und ReaktorsicherheitGarching, Germany

Sander WillemsenNuclear Research and Consultancy GroupPetten, Netherlands

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 27: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

25

One serious threat to the lifespan of a reactorpressure vessel is the occurrence of pressurized thermal shock (PTS) during an emergency core coolant(ECC) injection of cold water during a loss-of-coolantaccident (LOCA). The hot water present in the systemmixes with the cold ECC water in the primary coolantpiping as it flows toward the reactor vessel. When thecold water comes into contact with the pressure vesselwall, it may lead to large temperature gradients andconsequently high stresses in the wall of the vessel.The challenge for the CFD model is to correctly predictthe complex turbulent mixing phenomena occurringduring the ECC injection.

Within ECORA, GRS and NRG made simulationsof single-phase and two-phase flow UPTF test cases.

In both cases, grids were generated with ANSYSICEM CFD, and the flow and heat transfer were simulated using ANSYS CFX, primarily using the SSTturbulence model. ANSYS CFX was chosen becauseof its strength in two-phase flow modeling combinedwith the robustness of the coupled solver.

In the single-phase flow test case, the system initially was filled with stagnant hot water, and coldECC water was injected into a single cold leg. Thecold leg is connected to the reactor vessel at the top,in which the flow is immediately directed downwardinto an annulus past the reactor vessel wall (the so-called downcomer). Once the water reaches thebottom of the lower plenum, it turns upward to thenuclear core. While high-frequency flow oscillations at

Layout of the Upper Plenum Test Facility, a full-scale experimental rig of the primary cooling system of a four-loop 1300 MWe Siemens/KWUpressurized water reactor (PWR). The test vessel internals and the primary coolant piping were exact replicas of the reference plant. Otherimportant components, such as the core, coolant pumps, steam generator and containment, were replaced by simulators that mimic theirthermal–hydraulic behavior during a large break loss-of-coolant accident (LOCA).

Temperatures on the vessel and cold leg walls (left) and at cross-section through the middle of the cold leg with ECC injection (right)for the single-phase simulations

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 28: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

26

Simulation at Work

the connection of the cold leg to the downcomer werenot picked up in the current simulations, the thermalstratification in the cold leg is predicted accurately.Also, the most important factor for determining the severity of the thermal shock, the minimum temperature in the cold leg, is predicted to within 3 percent of the experimental value by ANSYS CFX.

The two-phase flow test case involvessteam–water flow in the intact cold legs and in thedowncomer. The coolant water of the primary systemflows rapidly through the break, and a significant fraction of the water flashes to steam. The pressure inthe primary system decreases as the blow-down progresses. When the pressure has reached a threshold value, the accumulators begin to injectemergency core coolant (ECC) water into the coldlegs. This is the point at which the test cases begin,with the water level in the cold leg just above the coldleg centerline. This two-phase flow simulation uses the

The left image (an isosurface of a constant velocity of 0.3 m/s, colored by temperature, 50 seconds after injection) shows ECC water at 304Kinjected into the horizontal cold leg and spreading in both directions — to the left toward the downcomer, and to the right toward the pump. The ECC water rapidly mixes with hot water, which has an initial temperature of 461K. The right image shows how water also flows backtoward the injection nozzle and pump. The warmer water found just below the free surface flows with similar velocities in the reverse direction. The large velocity gradients in the transition layer generate turbulence and enhance mixing.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

free surface modeling capability of ANSYS CFX. The basic functionality was verified using models of only one-quarter of the full UPTF geometry in order to reduce simulation times and quickly performquality checks. The results of the two-phase flow calculation compared favorably with experimentaldata for the temperature distribution along the coldleg. The vertical temperature distribution and the free-surface position also are in good agreement with data.

The largest discrepancies in both the single- andtwo-phase calculations are observed for quickly fluctuating flow phenomena. Although no definite conclusions can be drawn at this stage, it is suspected that the use of statistical turbulence models is the cause for this discrepancy. As a result,GRS and NRG currently are investigating the use ofinnovative turbulence modeling approaches, such as the scale-adaptive simulation (SAS) model developments in ANSYS CFX. �

Results of the ANSYS CFX reference calculation (left) and UPTF experiment (right). The numbers indicate the height in the cold leg (1 = topand 6 = bottom). The stratification in the cold leg is accurately predicted by ANSYS CFX. The calculated lowest temperature in the cold leg,which is the most important factor for determining the severity of the thermal shock, is within 3 percent of the experimental value.

Page 29: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Improving Efficiency of VacuumCleaner FansANSYS CFX is an integral part of turbomachinery performance studies focused on radial fan design.

At the Institute of Fluid Mechanics (Lehrstuhl fürStrömungsmechanik, LSTM), a research group hasbeen established to conduct research and develop-ment work in the field of turbomachinery. This workincludes turbomachinery design based on combinedanalytical design considerations, numerical perform-ance studies, rapid prototyping, and experimentalinvestigations of prototype impellers and diffusers aswell as complete fans. Extensive research and development has been done in the field of radialcompressors with an emphasis on radial fans forvacuum cleaner applications. ANSYS CFX 10.0 computational fluid dynamics software is an integralpart of the design process.

The flow through centrifugal turbomachines isprincipally radial in the region in which the energy fromthe rotating impeller is transferred to the air flow. Foraxial flow fans, the rotating impeller is passed in anaxial manner in the region in which the energy transfertakes place. Work on axial blowers also is being conducted at LSTM-Erlangen.

Centrifugal fans are similar in many respects toboth centrifugal pumps and centrifugal compressors.The principal distinction between these machines isthat pumps propel liquids that are practically incom-pressible, but compressors handle gases under suchconditions that a clear change in the density of theflowing fluid results. Usually, fans are devices that handle compression ratios below 1.1, whereas blowers operate between 1.1 and 4, and compressorsoperate above the compression ratio of 4.

In the first step of the development process, thegroup at LSTM-Erlangen carries out performancestudies of existing fans that later will be redesigned toreach higher efficiencies. These performance studiesare carried out numerically utilizing CAD data of thereal geometry as a basis for the grid generation.ANSYS ICEM CFD software provides excellent toolsfor fast grid generation for complex geometries.ANSYS CFX provides all the multiple frame-of-reference tools needed to perform computations for

By Philipp Epple and Caslav IlicInstitute of Fluid MechanicsFriedrich-Alexander University of Erlangen-Nürnberg, Germany

CAD model

27

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 30: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

rotating machinery. With ANSYS CFX, it is quite easyto conduct performance studies of the radial fans usedin vacuum cleaners. A CFD simulation with ANSYSCFX can encompass the flow from the test rig throughthe impeller and the diffuser into the guide vanes,through the motor and out into the space behind it.Very detailed information can be obtained from thecomputational results, and these can be employed tocompare to corresponding results obtained duringexperimental studies. A complete test rig is availableat LSTM-Erlangen so that combined numerical andexperimental investigations of radial blowers can becarried out.

LSTM-Erlangen has specialized in compressorsused in vacuum cleaners that have very high rotationalspeeds, ranging typically from 30,000 rpm up to50,000 rpm.

After measuring and simulating the actual fan, theimproved design goals are set. These goals usuallyinclude better efficiency at the same pressure and flowrate. When a complete new design is targeted, a fullsystem inverse mean line design is performed, consid-ering impeller, diffuser and deswirl vanes as a unit andnot separately. In addition, the system in which the fanwill operate is considered, so that the fan will be perfectly matched to the operating conditions of thevacuum cleaner. Verifying the geometries generated

with the inverse full system mean line design on theCFD simulation with ANSYS CFX, it is possible to iter-atively improve the design of the fan to specifications.In this way, it was possible for LSTM-Erlangen toachieve improvements of more than 10 percent in efficiency of vacuum cleaner fans. One valuable feature of ANSYS CFX is the power syntax, which permits the writing of post-processing scripts usinginline Perl commands. We used this method to getprecise information on the flow inside the machineneeded to validate and to improve the inverse full system mean line design.

Finally, with the design validated and improvedwith a careful and detailed ANSYS CFX simulation,there is almost no risk in building expensive prototypes.

Experimental investigations are then carried outat LSTM-Erlangen’s test rig, and the results are compared with specifications of the improved fan andwith the CFD computations. The detailed full systemsimulation agrees very well with measurements at thetest rig.

If further development is needed, modificationsare carried out. After careful design and CFD validationand improvement with ANSYS CFX, it is seldom necessary to make additional modifications.

Grid generated with ANSYS ICEM CFD

Comparison between measurement on the LSTM test rigand simulation with ANSYS CFX

Detailed simulation of a vacuum cleaner fan, including impeller, diffuser, deswirl vanes and motor

28

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Simulation at Work

Page 31: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

EMD used ANSYS software throughout developmentof the SD70ACe diesel locomotive.

ANSYS Mechanical helps Electro-Motive Diesel launch a newtrain locomotive on schedule and within budget.

Electro-Motive Diesel (EMD) is the world’s largest supplier of diesel-electric locomotives for commercialrailroad applications, including freight hauling, intercitypassenger, commuter, switching, industrial and mining. Since the early 1930s, the company has produced more than 58,000 locomotives for customers in more than 70 countries worldwide. EMDis also a global provider of diesel-powered engines formarine propulsion, offshore and land-based oil welldrilling rigs and stationary power generation. Headquarters, engineering and parts-manufacturingoperations are located in LaGrange, Illinois. Finalassembly is conducted in London, Ontario, from whichproducts are shipped to customers around the world.

One of the company’s major new products is theSD70ACe locomotive, which uses an alternating-current drive system to develop continuous tractiveeffort (pulling force) of 157,000 pounds for pullingheavy freight trains as well as higher-speed intermodaltrains. A direct-current drive model also is available foroperations requiring somewhat lower traction levels.

Meeting Difficult DemandsTo maintain its strong leadership position, EMD wasunder intense pressure in the development of theSD70ACe to meet difficult demands of performance,reliability, fuel economy, crashworthiness and operatorcomfort in the highly competitive locomotive market.Locomotives must operate safely and economicallyfor decades while pulling heavy loads under harsh conditions with a minimum of downtime. Most unitslog more than a million miles during the first six years

of operation and have a useful life of nearly 30 years,with some major components lasting over 50 years in the used-equipment market. Durability ofcomponents undergoing repeated fatigue cycles frominherent dynamic loads is a major concern.

Achieving these goals while shortening the development cycle to generate sales as quickly aspossible is particularly challenging for locomotivemanufacturers because of the significant time andcost associated with running physical tests on theselarge, complex machines. Individual components andassemblies are massive, requiring heavy-duty materialhandling equipment just to move them around andgigantic rigs to conduct tests on components. Forexample, the 70x10x2-ft. steel underframe structureconnecting the wheel assembly to the locomotivebody weighs in at 90,000 pounds. Near the end of thedevelopment cycle, complete vehicles are design validated on Department of Transportation test tracks,for which time is expensive and must be scheduledwell in advance. Problems uncovered during thesetests can mean major modifications and test iterations,with the potential to increase costs exponentially, significantly delay product launch and seriously jeopardize product profitability.

The Simulation StrategyEMD met these challenges with engineering simulationthroughout the design/development process from theearly conceptual stages of the cycle. Parts andassemblies were analyzed to refine the designs andverify that engineering specifications were met. Theentire locomotive was modeled to study overall product behavior to avoid unforeseen problems surfacing during final validation tests on the test track.By using simulation as a verification and validation toolthroughout design, EMD minimized physical testingwhile shortening development time and optimizingstructural characteristics of the SD70ACe.

ANSYS Mechanical software is a key element in this simulation strategy. Structural analysis wasused in evaluating stress, deflection and harmonics of

29

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 32: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Simulation at Work

components, especially in guiding designers in sculpting the basic shape and topology of parts earlyin development. Before the underframe was built, forexample, EMD used ANSYS Mechanical to verify that the structure met design specifications for withstanding tension and compression “buff” loads of 1 million pounds while minimizing the weight of this large component. The software also was used in detailed analysis of suspension dynamics, isolation mounting of the cab and other subsystemsfrom vibration, as well as in the development of the crankcase, cylinder heads, piston rods and other reciprocating parts of the locomotive’s high-performance diesel engine.

Throughout the development cycle, designersand analysts worked collaboratively on the SD70ACe,with simulation and drawings completed simultane-ously. During the process, engineers were able to per-form first-pass analysis on individual parts early in development to study alternatives and perform what-ifevaluations. Detailed simulations of final designs andcomplex assemblies were performed by the analysisgroup. Routine analysis work was computed on HP desktop X-class computers, while assembly models of the entire locomotive were run on HP J6000UNIX workstations.

Staying on Schedule and within BudgetThe use of analysis throughout product developmentwas instrumental in helping EMD launch the SD70ACeon schedule and within budget. ANSYS softwareenabled analysts and designers working on the product development team to optimize the design ofindividual components and assemblies as well as theoverall locomotive structure. By studying alternativeconfigurations, the team was able to make informeddecisions on the trade-offs between stiffness, weightand natural frequency requirements, for example.

Structural analysis of both the engine and the full locomotive played a key role in increasing fuel efficiency while maintaining high durability and operator comfort with minimized vibrations throughout the structure. The new product builds on the record-setting reliability performance of EMD’s locomotivesthat railroads around the world use to pull heavyfreight trains.

EMD has now delivered 20 SD70ACe locomo-tives to CSX Corporation and is nearing completion ofan order for 115 locomotives to the Union Pacific Rail-road. Furthermore, BHP Billiton Iron Ore Pty. Ltd. is

acquiring 14 of the locomotives for mining operationsin northeast Australia. In the United States, 16 units arebeing purchased by Montana Rail Link (MRL), whoseheavy-haul train tests confirmed that the SD70ACeprovides fuel savings greater than 20 percent. Five ofthe new locomotives will replace eight older units oncoal and grain trains that MRL operates over the continental divide in western Montana.

Importance of Analysis“We absolutely could not design locomotives competitively without extensive use of analysis fromconcept through release to manufacturing,” explainsR. Thomas Scott, manager of noise vibration andstructural analysis at EMD. “Physical testing on theselarge, complex structures must be kept to a minimum,and we use simulation to guide the design, verify thatengineering specifications are being met and help validate the final design.”

According to Scott, the robust capabilities ofANSYS software enable the company to perform critical structural, vibration, modal and durability analysis throughout the development cycle. Moreover, the integration of modeling, solution andpost-processing capabilities within a single packagehelps team members study design alternatives, perform what-if simulations and exchange ideasquickly. “This facilitates collaboration and breaksdown organizational and functional barriers,” saysScott, “allowing design and analysis groups to worktogether toward our company goal of continuing to develop and build the best locomotives on the market today.” �

The entire locomotive was modeled to study overall product behavior and avoid unforeseen problems during final validationtests on the test track.

ANSYS Mechanical verified that the underframe structure metdesign specifications for withstanding tension and compression“buff” loads of 1 million pounds.

30

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 33: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

This portal is the central repositoryfor information and discussion onANSYS Workbench, helping usersget the most from this simulation tool and CAE framework.

By Pierre ThieffryTechnical Solutions Specialist ANSYS, Inc.

The ANSYS Workbench Community Forum is a valuable, collaborative resource for longtime Workbench users as well as beginners. This portal is the central repository for information and discus-sion on the software and is dedicated to helping customers use ANSYS Workbench more effectively.The forum offers information on Workbench as a pre- and post-processing tool for ANSYS and as a CAE framework for engineering knowledge capture. For access to the ANSYS Workbench Community Forum, go to the ANSYS Customer

Join the ANSYS WorkbenchCommunity Forum

The corporate area of the forum has numerous examples of ANSYS Workbench use, including sample fileswith links to documents and demonstration files that show how to use specific features of the software.

Portal (www.ansys.com/customer) and click on thebottom-left item of the menu.

The forum is divided into two areas: corporateand public. The corporate area has numerous examples of ANSYS Workbench use, including sample files. Have you ever wanted to post-processstatistical results in ANSYS Workbench, use cyclicsymmetry or reuse all your existing APDL commandswith this application? The Migration Series helps toanswer your questions and gives you the ability todownload detailed examples.

CAE Community

31

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

For access to the ANSYS Workbench Community Forum, go to the ANSYS CustomerPortal (www.ansys.com/customer) and click on the bottom-left item of the menu.

Page 34: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

The next question follows: “When I change anything in my command block, do I need to solve themodel again? Could I evaluate results without theneed for a new solve that takes 10 hours?”

The answer is that you don’t need to solve again.To discover how this can be done, join the ANSYSWorkbench Community Forum and have a look at theposting “WB Migration: Post-Processing” located inthe corporate area of the forum, under ANSYS Work-bench FAQs/Tips and Tricks Forum section “FromANSYS to Workbench.” There are many more entriesto read in this section about cyclic symmetry, transientanalysis and meshing techniques.

Ever thought of customizing ANSYS Workbenchto automate your daily work or creating scripts forWorkbench? Take a look at the ANSYS WorkbenchCustomization Forum or visit the Customization section of the FAQs/Tips and Tricks Forum, which has many examples to get you started. There are sample scripts that you can download and start using immediately in your particular application.

If you cannot find what you’re looking for in thecorporate area, then the public area is the right placeto go. This is where users can post and exchangeinformation with other members and ANSYS staff, askquestions about using the products in the DiscussionForum and give feedback to ANSYS in the Sugges-tions Forum. Member requests range from basicmodel setup up to implementation of advanced functionalities such as birth and death. In addition,

Among all the topics of the ANSYS Workbench Community Forum, one of the most popular isadvanced post-processing, which deals with questionssuch as “How can I add energy density plots in ANSYS Workbench?”

The answer is quite simple: Use an APDL command block inserted under the solution item in yoursimulation tree. Generally speaking, command blocksare used to insert any APDL command, whether it’s for pre-processing tasks, solution settings or post-processing. All APDL commands introduced that waywill be sent to ANSYS when you ask for solving.

In the case of additional plots to be inserted in yoursimulation, the trick is to divert ANSYS graphical outputto the png format. So, for an elastic energy density plot,the three lines of APDL you need are:

/post1/sho,pngplesol,send,elastic

Outputting to the png format allows ANSYS Workbench to include the plot in the simulation tree and,of course, in the automated report generation. Morecommands can be included to change the graphics layout, add additional plots or even issue time historyplots using /post26. All subsequent plots included in thecommand block will be listed under this one in the simulation tree, as shown in the accompanying figure.This procedure is useful to plot any result that is notexposed in the ANSYS Workbench interface. You eventually can use the mechanism to plot an axisym-metric or cyclic expansion plot of your model as well.

you have the opportunity to share your knowledgewith other users worldwide.

Postings in the public forum are quite informativeand focus on popular topics of interest, such as meshsizing controls and ways to get a pure hexahedralmesh for parts that are non-sweepable (that is, partsthat are not the result of some kind of extrusion or apartial revolution). Have a look at two comprehensivepostings on the subjects, “How to Use Mesh Controlsin Simulation” and “Swept Meshing with DM and Sim-ulation.” Finally, read the instructive “Mapped Meshingin Simulation” posting for additional information. �

Want to Learn More?Join the ANSYS Workbench Community today. Visithttp://www.ansys.com/products/workbenchportal.aspand learn about accessing the ANSYS WorkbenchCommunity Portal.

Popular Forum Topic: Advanced Post-Processing

32

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

CAE Community

Topics relating to advanced post-processing in ANSYS Workbench are coveredin the forum. This sample tree and embedded plot on the forum show how topost-process cyclic symmetry results in ANSYS Workbench.

Page 35: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

33

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Software Update

Tools from Harvard ThermalNow Part of the ANSYS SuiteSoftware focuses on advancedsimulation of electronics cooling.

Late in 2005, ANSYS, Inc. acquired the assets of Harvard Thermal Inc., a software company specializing in thermal analysis of electronics. This article explores the history of the company, explainsthe evolution of its technology and presents anoverview of its leading-edge products.

A Software Evolution

As a thermal analyst in the early 1980s, Dave Rosatofound that the software then available for thermalmodeling of electronics lacked a graphical user interface, so models would takes weeks of hand calculations and complicated bookkeeping to construct. Rosato decided to develop a quicker wayof building models and started writing his own program. In 1994, he founded Harvard Thermal Inc.(HTI) to sell the software.

Now known as ANSYS TAS, the software is usedat numerous military and defense companies and wasdesigned to retain much of the versatility of NASAsoftware. Models are built like other ANSYS productswith geometric elements, but the thermal solver isbased on the same finite difference techniques foundin the NASA software. This unique strategy finite element–style model generation and finite differencesolving accommodated interfaces to both FEA toolslike ANSYS and finite difference tools like the SINDA products.

Simulating PCBs

Seeing the need for a thermal simulation tool designedspecifically for printed circuit boards (PCBs), HTIbegan development on such a tool using TAS as thefoundation. ANSYS has strong interfaces with MCADtools required to represent complex geometry, butdata needed to simulate a PCB exist in ECAD tools.HTI’s first choice was to try interfacing with these tools

using an interface called intermediate data format(IDF). Unfortunately, the version of this interface supported by the electrical tools does not includeinformation needed to perform an accurate simulation.In addition, it was apparent that the more completeversions of the interface were not going to be supported in the future. Therefore, HTI decided to usea third-party tool that had direct interfaces with theECAD tools. Without this data, it would be up to theuser to define this missing information, which mightcompromise the accuracy of the tool and require additional user expertise. This decision would prove to set the tool apart from all existing and future competitive tools. The HTI tool was released in 2003under the name TASPCB.

In designing TASPCB, the goal was to producean easy-to-use product that solves quickly andrequires a minimum of heat transfer background yet isversatile enough to satisfy a full-time thermal analyst.Having a direct ECAD interface allowed the softwareto accurately represent the complex thermal conduction in the board.

Temperature of the printed circuit board as well as eachcomponent junction and case are indicated by graphicalresults from ANSYS TASPCB.

Page 36: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

34

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

A PCB is made up of multiple layers of copper traces that carry electrical signals betweencomponents. These layers also contain large copperplanes to carry voltages and ground to the components. Vias through the board electrically connect the traces and planes on different layers. Thecopper layers are electrically isolated with a plastic(dielectric) material, which gives the board the often-seen green color. Copper, having a thermal conductivity more than 1,000 times higher than thedielectric material, provides the primary heat transferpath in the board. Importing every trace, plane and via from the ECAD tool allows ANSYS TASPCB to accurately and three-dimensionally represent the heattransfer in the board. Without this capability, the userwould have to define the board manually (as is done inother tools), requiring an added expertise by the user.

One method used to reduce solve time involvesdifferent meshes on the board- and air-side of theproblem. This way, the two meshes can be optimizedwithout impacting each other, which increases accuracy yet reduces solve time. In most cases, areasin the CFD mesh that require refinement to get accurate results are not the same areas that needrefinement in the board conduction mesh. Unlike TAS,in which models are built manually, all models inTASPCB are generated automatically. Logic built intothe code creates a good mesh every time with minimalinput from the user.

For many years, HTI had been involved in the IPC (formerly the Institute of Interconnecting andPackaging Electronic Circuits) 1-10b Current CarryingCapacity Task Group formed to update IPC guidelinesfor sizing PCB traces to handle required electrical currents. The existing IPC guidelines had been developed in the late 1950s and didn’t account for the

complexities of today’s PCBs. A software tool wasneeded to predict trace temperatures. Since TASPCBimported all trace geometry from the ECAD tool, thecapability for the software to define trace current, predict trace and plane voltage drops and calculatepower dissipation was added. This enabled thermalsimulation to be used as an alternative to testing inpredicting if the trace will fail or not.

Developing Package Thermal Designer

A single electronic component today contains theelectrical functionality of several components from justa few years ago. The complexity of components combined with their power dissipation means thatpackage designers must increasingly address thethermal design. Again, seeing a need in the industryfor a detailed yet easy-to-use tool for thermallydesigning electronic components, HTI began development of such a tool.

Following the approach of TASPCB, this newtool, called Package Thermal Designer (PTD), neededinterfaces with the tools used to design components.This time, HTI internally developed the direct interfaceto the two major tools, APD from Cadence® and UPDfrom Sigrity.® PTD also includes an MCAD interface toimport complex geometry, such as lead frames orcomplex heat spreaders.

PTD and the other HTI products are now part ofthe many ANSYS software products designed for theelectronics industry. �

Software Update

Lead frame and die paddle temperatures on this package aredetermined by ANSYS PTD, which can simulate the thermalbehavior of nearly any type of electronic component.

Velocity vectors generated by the ANSYS TASPCB computational fluid dynamics solver predict how cooling air flows over a circuit board.

Page 37: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

35

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Innovation is not just about agood idea; it is a process of managing what can appear to bean army of people over a setamount of time making multipleinterconnected decisions. Ratherthan micromanaging, let theproduct requirements guide thelegions who make the detaileddaily trade-offs. Yes, these product requirements emergefrom an early research and planning stage that is chaotic.

But that is good, for the chaos enables exploration andlearning. The more you can learn about your market, thebetter the framework for your decisions.

Organizing the Decision-Making ProcessIn every new product, many key decisions must be made if the prospective innovation will reach maximum potential and generate equity in the brandand profit needed to sustain a company. Not only arethere numerous decisions, but each decision is relatedto many others, typically as a trade-off. There is noway to make a product, say an SUV, that has fuel efficiency, lots of cargo room, three rows of seats, premium features, high performance, tight craftsman-ship and individualized feature choices, all at a lowcost. Choices are made, and aspects are sacrificed.

Managing product development in this context ofmaking decisions about conflicting trade-offs is achallenge. It requires balancing equal and oppositeforces. The best managers seem to find a way to keepthe big picture and goal in mind and also feel free tovary the program as it develops. Instead of hoping theprogram will go according to the ideal, these managers realize from the beginning that the ideal isthere as a reference for support, not a process carvedin stone. They can shift and interpret the process asneeded and address new issues accordingly. Insteadof feeling threatened and trying to make the idealprocess model fit the unanticipated issues, they enjoymeeting the challenges. They realize that in product

Innovation in product development depends ona process that provides structure with enoughflexibility to adapt to the unpredictable.

Making Trade-Off Decisions in the Chaos of Product Development

development you need a process that provides structure, and you need to adapt to the unpredictable.That is the foundation of innovation.

The Butterfly EffectWithin that structure, if the product developmentprocess at times seems chaotic, it is! If a butterfly flaps its wings in Brazil, this could cause a storm in Belgium the next week. This is the basis of themathematics of chaos, in which a small event canhave enormous and unpredictable consequencesdown the road. The decision-making process in product development has many similarities to the butterfly flapping its wings, the seemingly small event. Decision-making is highly causal: One apparently insignificant decision can significantlyaffect the outcome, just like one flap of a butterfly’swings can have a profound effect on weather.

So it is with product development that every decision affects every other decision, with one connected to the next, and each downstream decisiondependent on previous choices. If slightly differentdecisions were made — say, to create a different visual line in a vehicle, orselect a different springmechanism in a toaster orchoose a different materialfor the bristles in a tooth-brush — the implications ofthose decisions produce avery different vehicle,toaster or toothbrush.

Unlike the butterfly,product developers caninfluence the outcome ofthe process. The research,insight and feel for theproduct and process allowthem to make decisions

By Jonathan Cagan, Ph.D.Professor, Mechanical EngineeringCarnegie Mellon University

Guest CommentaryP

hoto

by

Larr

y R

ippe

l

Page 38: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

36

Guest Commentary

that are likely to lead to more successful outcomes.Each informed or insightful decision affects the next,and so on. The butterfly wing-flapping is very much arandom occurrence in the weather system; the butterfly at best can control its own flight and has noawareness beyond its own activities. The butterflydoes not understand causation. Successful productdevelopers understand the cause–effect relationshipand the overall implications of feature selection, andform choices for the product’s gestalt — how theproduct looks and functions as a whole. Unfortunately,some product developers are more like the butterfly.They make independent decisions without a thoroughunderstanding of the market opportunity, the customeror the rest of the product as a whole, never understanding why their product falters or fails in the marketplace.

Decision-making in product development is not arandom event. That said, many random influencesprovide fodder for decision-making. For example, thebrainstorming process used early in design to stimulate ways of conceptualizing possible productsolutions is wrought with random thoughts and analogies. Also, throughout the product developmentprocess, random external events occur that cannot bepredicted. Political and social events rapidly change theneeds and desires of a customer base, for example.

Chaos within StructureAlthough external influences are unpredictable, astructure to good product development guides theprocess to success and provides methods to improvethe robustness of decision-making. The structure ofthe product development process guides you throughthe unknown, helping you define your goals, con-straints and variables. The challenge in developingtruly innovative products is first to identify a unique set of goals, then to identify a set of variables that can be modified to reach those goals, and then tounderstand the real versus perceived constraints on those variables.

When Palm Computing® came out with its firstPDA, competitors believed the form factor to be a realconstraint, because the computing power (and thelarger chips back then) required to recognize hand-writing took up substantial space. Palm’s innovativesolution came from the recognition that form factorwas a variable after all, at least as long as customerswere willing to learn a new graffiti alphabet. Palm’sinnovation launched the whole PDA category. The butterfly flaps its wings within the constraints ofphysics. The product development process must workwithin the bounds of physics, but it also is influencedby humans, culture, society and thought, all of whichwere key to Palm’s success.

The structure of the process does not define thegoals, constraints or variables — it does not do thework for you. It provides guidance on how to navigatethe space of the unknown. It helps you make robustdecisions based on insights and incomplete or evenincorrect facts. The fodder it provides to make thosedecisions is based on the centrality of the customer.The customer unites all divisions of the innovativecompany — the user is the fulcrum that balancesgoals, constraints and variables.

The fuzzy front end, or the early stage of productinnovation, will be chaotic. That chaos is a good thing.It enables explorations and learning. The more you canlearn about your market, the better your filter on thechaotic ideation process. Chaos helps with the accuracy in finding and defining a good product direction. Later, the process begins to change. There is enough focus, enough variables and constraints identified, that the system begins to be more predictable, and precision becomes the focus.Although not every random event can be anticipated,many can — and a robust process accounts for scenarios of disruption. Accepting chaos allows you to deal more aptly with random impact as it takesplace, to work with it rather than against it. If yousquash the chaos, you squash the exploration andresearch so critical to the success of developing a product. Rather than fight it, be the choreographer of the chaos. �

This article was excerpted with permission from the book The Design of Things to Come: How Ordinary People CreateExtraordinary Products by Craig M. Vogel, Jonathan Caganand Peter Boatwright. Published by Wharton School Publishing. Copyright 2005 by Pearson Education, Inc. All rights reserved.

Page 39: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

By John CrawfordConsulting Analyst

Use beam elements to satisfy analysis needs when buildingmodels with different generations of elements in ANSYS 10.0.

Practical Considerations in Using Beam Elements

and BEAM189) do not allow for intermediate loads using the SFBEAM command. Some of the earlier elements act only in two dimensions, which issomewhat more efficient and simpler if you have amodel that is truly 2-D.

There are practical limits that should be kept inmind when using any type of finite element, of course.But what are some good suggestions, particularlyregarding the usage of beam elements?

The general recommendation for beam elementsis that they should have a slenderness ratio greaterthan 30. The slenderness ratio is defined as GAL2/(EI),in which G is the modulus of rigidity, A is the cross-sectional area of the beam, L is the length of the beam,E is the elastic modulus and I is the moment of inertia.For a rectangular cross section, a slenderness ratio of30 works out to be an L/h ratio of about 2.5, in which his the height of the beam. (This illustrative exampleuses a rectangular beam in which the height is twicethe base, but other cross-section shapes will yield different L/h ratios depending on their moment of inertia.) Beams with a slenderness ratio less than 30 tend to be too stiff, although the new elements aremuch better for short beams than the earlier elementsdue to using Timoshenko theory rather than Bernoulli.It’s important to note that when we talk about slender-ness ratio, we are not referring to the slenderness ratioof a beam element but to the slenderness ratio of thephysical beam that the beam elements represent.

In the last issue of ANSYS Solutions, we talked with PeterKohnke, Ph.D. of ANSYS, Inc.about the differences betweenthe earlier generation of beam

elements and the later generation that has been introduced and enhanced for the last few years. Peterdiscussed the differences in the shape functions andsome of the theoretical issues that drove the creationof the newer elements.

Now that we have an idea of what the differencesare between the two generations of beam elements,we’ll talk about some practical issues to keep in mind.

Still a Place for Earlier Elements

In the last article, we mentioned that the newer beamelements are a substantial improvement over the earlier elements in a number of ways. For example,they handle transverse shear more accurately, andthey handle nonlinearities more efficiently. Is there still a place for the earlier beam elements in the analyst’s toolbox?

Peter says the older elements (BEAM3, BEAM4,BEAM44 and BEAM54) still have some advantages,especially in linear analysis. Their shape functionsautomatically include a full cubic bending shape function, which may be useful in thin member bending situations. The newer elements (BEAM188

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

37

Tech File

Page 40: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Mesh Descretization

I asked if mesh descretization is important whenusing beam elements and, if so, how do we knowwhen we have enough elements to get a validanswer?

Peter said the general, all-purpose answer tothis question is to try two or more models using a different number of elements. If the results are substantially the same, you have enough elementswith the coarse mesh. If the results are substantiallydifferent, your fine mesh may not be fine enough.

If you have distributed loads along the beam, afiner mesh usually is called for, as the shape functions cannot exactly model this situation. If youhave point loads applied to the interior of a beamstructure, those are usually good places to add anode that ends one element and begins the next. Ifthe load is relatively minor, you can apply that loadusing the offset distance options on the SFBEAMcommand using the older elements. Of course, if youwant to monitor the displacement at a point in anystructure, that is also a good place to put a node.

Since BEAM188, by default, has a constantbending moment throughout, you may need to refinethe mesh to more accurately calculate results. Again,this limitation can be bypassed by using KEYOPT(3)= 2 or moving to BEAM189. Note that BEAM188with KEYOPT(3) = 0 is the element of choice whenadding stiffeners to shell models while doing nonlinear analysis for reasons of compatibility.

Tying Beam Elements to Solid Elements

All beams have rotational degrees of freedom, andwe sometimes want to tie them to solid elementsthat have only translational degrees of freedom.What methods are available for tying beam elementsto solid elements, and what are their strengths and weaknesses?

There are three methods for doing this, Petersaid. First, you can bury some beam elements inthe mesh of solid elements. This is simple but mayadd unwanted local stiffness. Second, you can use constraint equations. This is more difficult to input,but it is more correct theoretically unless largedeflections are used. Third, the MPC method can be used, which generates constraint equations internally and is valid for large deflections.

The two families of beam elements that are available in ANSYS offer a comprehensive range ofcapabilities that can be used for almost all cases inwhich beam elements are appropriate. The earlier generation of beam elements works well for framestructures and similar applications in which thesophisticated capabilities of the newer elementsmay not be needed. The newer generation of elements has nonlinear and section capabilities thatallow the user to model some very complex andsophisticated structures with a full offering of linearand nonlinear capabilities.

It’s important for the analyst to use beam elements where they satisfy the needs of the analysis. If your problem is suitable for beam elements, you’ll get an accurate answer more quickly than you would from a shell or solid model.Additionally, they allow the user to change cross-sectional properties without remeshing the entiremodel. This is useful for parametric modeling anddoing optimization studies. �

Special thanks to Peter Kohnke, Ph.D. and GramaBhashyam, Ph.D. of ANSYS, Inc. for their contributions tothis article.

38

Tech File

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 41: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

Tips and Techniques

39

A variety of options allows users to solve multipleproblems locally or remotely.

ANSYS Workbench (WB) simulation allows the user to solve multiple jobs locally or remotely. These capabilities are defined within the solution branch of Workbench simulation as well as in the ANSYS Workbench remote solution manager. Thedefault solution behavior in Workbench simulation is tosolve models synchronously. This means that whenthe user initiates a solution by clicking on the solve icon, any environments on and below the currentlyselected branch will be solved and results will beretrieved automatically.

The specification of synchronous solution can beverified under the solution branch details view, asshown in Figure 1. When “run process on” is set to“local machine,” this reflects a synchronous solution.The number of processors used for solving also can be specified. (Default is to use two processors, if available.)

This method is useful when the user has only asingle ANSYS license. Also, by having multiple modelbranches and/or several environment branches in the same Workbench simulation database, a user can select the parent branch and solve multiple environments sequentially.

An asynchronous solution allows a user to solve the model on remote machines. Generallyspeaking, there are three configurations supported inWorkbench simulation:

■ Solving directly on a UNIX or Linux machine■ Solving on an ANSYS Workbench cluster■ Solving using Platform LSF® software

As in Figure 2, when solving directly to a UNIX orLinux machine, “run process on” should be set to “WBcluster.” “RSM Web server” can be left as “localhost,”and “assignment” should be changed to “server.”Through this method, the UNIX/Linux hostname andlogin information can be specified, and the solution willbe submitted to that UNIX/Linux machine.

The WB cluster method requires the use of Windows machines appropriately configured. (Detailson configuring a WB cluster can be found in the“ANSYS Workbench Products Remote Solution

Using the ANSYS WorkbenchRemote Solution Manager

Manager Configuration Guide” in the Installation andConfiguration Guides section.) Through the use of IISand .NET, a Web service machine running WindowsServer 2003 or later receives requests to solve jobs oncompute server machines. Queues and compute serverdesignations are set on the Web service machine. A user then can run Workbench simulation on a separatePC and monitor/submit jobs to the Web servicemachine, which then sends the jobs to be solved oncompute servers, based on the queue selected.

The user also can utilize LSF software under Platform Computing (www.platform.com/) to submit jobsto queues of compute servers. The asynchronous solution capability provides many benefits:

■ The computing is done on a remote computeserver, so the user’s PC can be used for other purposes and not be bogged down by the computation.

■ The current Workbench simulation database canbe saved and closed, and a different model can be opened for the user to work on.

■ Depending on the number of licenses availableand the configuration of the Workbench cluster,more than one analysis may be initiated at agiven time, leading to faster turnaround times for multiple jobs.

■ If ANSYS Prep-Post licenses are available, these can be used to perform the pre- and post-processing tasks in Workbench simulation whileregular ANSYS licenses (such as ANSYSMechanical or ANSYS Multiphysics) can be usedto solve the model. This can be a cost-effectiveway of managing multiple ANSYS licenses formany users.

By Sheldon ImaokaTechnical Support EngineerANSYS, Inc.

+

Figure 1. The specification of synchronous solution can beverified under the solution branch details view.

Figure 2. With proper settings, models can be solved directly on aUNIX or Linux machine.

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Page 42: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community

+

40

However, there are some points to keep in mindwhen using asynchronous solution:

■ An additional ANSYS license is required to initiate the solution on the remote compute server.

■ Some features, such as convergence objects or thermal-stress simulations, are not supported. (Refer to “Simulation Help |Using Simulation Features | Synchronous and Asynchronous Solutions” in the Release10.0 Documentation for ANSYS Workbenchfor a listing of items unsupported in asynchronous solution.)

■ It is important to note that while the model is being solved, the user cannot change any input parameters. This prevents inconsistencies between the model and theretrieved results once the solution is completed.

■ Some configuration needs to be done in orderto enable this type of computing environment.

Another option is the ability to combine both ofthese features. The remote solution manager (RSM)can be used locally to allow a user to run multiple jobsfrom different Workbench simulation databases.

The RSM can be controlled via the ANSYS icon on the bottom-right section of the Windowstaskbar. By right-clicking on this icon and selecting“open job status,” the ANSYS solution status monitorwill appear, which lists running, pending and completed jobs.

The first thing to note is that there are two itemsthe user can configure — servers and queues — similar to the previously mentioned asynchronousremote solution methods.

Servers are compute servers, and this can be either the local machine or a remote UNIX/Linuxworkstation. If the user has dual processors and wantsto solve two jobs simultaneously, an additional server can be created with “name = localhost2” (anarbitrary name), “machinename = localhost” and“enabled = true,” as in Figure 3.

Queues can be thought of as the grouping ofcompute servers. If a user just wants to solve locally,the default queue of “local” is sufficient. Per the above

scenario, if the user wants to solve two jobs simultaneously on a dual-processor PC, the “local” queue should be opened and both “localhost” and “localhost2” should be activeunder “assigned servers.”

Once this is complete, in Workbench simula-tion under the solution branch, the user may usethe local RSM by selecting “run Process on = WBcluster” with “RSM Web server = localhost.” Thequeue name would be the same as above. (Defaultis “local.”) The “license to use” also should be specified since, as noted above, asynchronoussolutions require an additional license even ifsolved locally because this allows the user to solveone model via RSM while working on a differentWorkbench simulation database.

Once jobs are submitted, they will be queuedto the local RSM. The user then can save themodel and exit Workbench simulation or work on a different model. If desktop alert is activated on theRSM icon on the taskbar, a listing of pending andrunning jobs will appear in a small, separate window, as shown in Figure 4. To remove this window, right-click on the RSM icon and unselect “desktop alert.”

The jobs will be solving in a directory called“ce_” (in which “_” is a number) in the Windowstemporary directory (designated by the systemenvironment variable TEMP). When the solution is complete, the user can open the Workbenchsimulation database and right-click on the solutionbranch to “get results.”

In addition to the benefits already noted, thismethod allows the Windows user to run two jobssimultaneously on the dual-processor machine.This doesn’t require running two instances ofANSYS Workbench simulation. Therefore, with onlythe ANSYS solver running, additional memory isfreed up for larger jobs.

Some additional points to note:■ The user should not completely log out or

shutdown the computer, as that will haltany running jobs.

■ There are two RSM applications that arerequired — WBProcStat.exe (WB processstatus) and JMService.exe (job managerservice) — and these applications shouldnot be terminated. �

Contact the author at [email protected] for theentire paper from which this article was excerpted.

Tips and Techniques

www.ansys.com ANSYS Solutions | Volume 7, Issue 3 2006

Figure 3. An additional “server” specification can be added tosolve two local jobs simultaneously on dual-processor PCs.

Figure 4. Multiple jobs are queued to the local RSM withpending and running status listed in the desktop alert window.

Page 43: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community
Page 44: 62438, page 1-44 @ PDFReady - MAPDataweapons, armor, vehicles, ships and other products that respond to new threats in an ever-changing world. 11 4 15 18 Departments CAE Community