FEA Newsletter August 2006 - Free of Papers from LS
Contents
1. F ainformation com Informational Websites The LSTC LS DYNA Support site www dynasupport com LSTC LS DYNA Support Site www dynasupport com TopCrunch Benchmarks www topcrunch org LS DYNA Examples more than 100 Examples www dynaexamples com LS DYNA Conference Site www ls dynaconferences com LS DYNA Publications to Download On Line LS DYNA Publications www feapublications com LS DYNA CADFEM Portal www lisdyna portal com www dynalook com 38 F information com 39 6 European LS DYNA Conference May 28 29 2007 www erab se conference2007 Engineering Research Nordic will be hosting the 6th European LS DYNA Users Conference held at Quality Hotel 11 in Gothenburg Sweden May 28 29 2007 The conference brings together LS DYNA users researchers and developers to dis cuss LS DYNA simulations of complex mechanical problems The conference also attracts a wide range of hardware and software companies wishing to Showcase their special solutions when running LS DYNA simulations The conference will cover LS DYNA appli cations Among them will be Aerospace Automotive Biomechanics Civil Engineering Impact and Drop Testing Material Modeling Metal Forming Nuclear Occupant Pedestrian Safety Seismic Engineering and more A software and hardware exhibition will be organized during the conference In terested companies should contact Engi neering Research Nordic We are2. Response by Amlanjyoti Barman 1 Select all the beams 2 Go to Beam Panel in HM button is on the left side 3 Check the UPDATE radio button 4 Instead of direction nodes Please do this gt Define x comp 5 7e 4 y 5 7e 4 Z 5 7e 4 5 Click on UPDATE button to update the direction with above values button is on the right hand side 6 Now run the model Response by Bob Browning According to the User s Manual N3 is required for all beam sections except Types 3 6 7 8 and 9 and it is optional for these Therefore you have to define this node so that LS DYNA knows the orientation of the cross section If your reinforcement is circular as as sume it is then the orientation is not critical but still needed for the problem specification You can create a single node to use as the reference node for each element but believe that this only works if you are not using the coordinate update option Check out Figure 12 1 under the Element_Beam description in the User s Manual for the description of this Apparently you can also use the ORIENTATION option but have never dealt with this be fore LS DYNA Yahoo Groups There are over 1870 subscribers from all over the world and this list seems to grow by a hundred new subscribers ever few months no small testament to the rapidly growing popularity of LS DYNA The group currently averages about 250 message per month Le about 10 message per day You can subscribe to t
3. The California office training center ac commodates 23 students The Michigan office has capacity for 12 students Classes are scheduled throughout the year at both locations For the most current schedule visit www lstc com Onsite training is also available Analysis Advanced Options in LS DYNA 750 Sept 05 06 ALE Eulerian amp Fluid Structure 750 Interaction in LS DYNA Concrete and Geomaterial Model Contact in LS DYNA 750 Sept 12 13 Aug 15 16 2 Oct 23 26 Introduction to LS DYNA 750 WE ee Dec 11 14 Introduction to LS OPT 750 Nov 07 10 LS DYNA Composite Materials 750 Sept 14 15 Wa LS DYNA for Heat Transfer 500 amp Thermal Stress Problems Material Modeling Using LS DYNA 750 User Defined Options MESH Free Methods in LS DYNA 750 SPH and EFG 23 F information com EVENTS 2006 If you want your event listed please send the information to mv feainformation com Call For Papers Deadline September 18 2006 NAFEMS World Congress Vancouver Canada May 22 25 2007 www nafems org NAFEMS is an independent not for profit body with the sole aim of pro moting the effective use of engineering Simulation methods such as finite ele ment analysis multibody system dy namics and computational fluid dy namics Simulation is now established in many engineering companies as part of their product development process Whilst the conference will cover many aspects of the use of
4. ANSYS LS DYNA ANSYS LS DYNA Built upon the suc cessful ANSYS interface ANSYS LS DYNA is an integrated pre and postpro cessor for the worlds most respected ex plicit dynamics solver LS DYNA The combination makes it possible to solve combined explicit implicit simulations in a very efficient manner as well as per form extensive coupled simulations in Robust Design by using mature struc tural thermal electromagnetic and CFD technologies Al Environment A high end pre and post processor for LS DYNA Al Environment is a powerful tool for advanced modeling of complex struc tures found in automotive aerospace electronic and medical fields Solid Shell Beam Fluid and Electromagnetic meshing and mesh editing tools are in cluded under a single interface making Al Environement highly capable yet easy to usefor advanced modeling needs ETA DYNAFORM Includes a complete CAD interface capa ble of importing modeling and analyz ing any die design Available for PC LINUX and UNIX DYNAFORM couples af fordable software with today s high end low cost hardware for a complete and affordable metal forming solution ETA VPG Streamlined CAE software package pro vides an event based simulation solution of nonlinear dynamic problems eta VPG s single software package over comes the limitations of existing CAE analysis methods It is designed to ana lyze the behavior of mechanical and structural systems as si
5. SGI Altix 4700 Series Results Number of Dual Core CPUs Wall Time CPU Time 1 4144 63 4088 99 4 CPU bandwidth compute blade performance 1229 84 1215 65 4 CPU density compute blade performance 1329 24 1310 17 8 CPU density compute blade performance 770 58 753 48 Bioengineering Bone Segment Simulation Benchmark The model is a bone joint segment consists of 68 401 grid points and 61 038 eight node isoperimetric brick elements A 10 increment non linear static analysis with contact was run with the MSC Marc parallel sparse direct solver The parallel sparse direct solver is an openmp application which is sensitive to the remote memory latency on a NUMA computer system Therefore the density compute blade of the SGI Altix4700 with its larger node and lower latency is better suited for the task Application MSC Marc implicit nonlinear Hardware SGI Altix 4700 Server BE 1 6GHz dual core Intel Itanium 2 processors with 24MB L3 caches 24MB cache is shared by two cores Operating System Propack 4 SP3 Model Sizing and Requested Options lement type requested umber of elements in mesh 61 038 umber of nodes in mesh 68 401 F iinformation com 11 Values stored at all integration points Number of lists of distributed loads Assumed strain formulation is used Extended precision input is used Number of processors used Y UY Marc input version Hl SGI Altix 4700 Series Results Number of Dual Core CPUs
6. WWW EE AINEORMATIOhN COM COVER STORY New IBM System x Servers with the AMD Opteron Processor TurboCharge a Cooler Design Center PRODUCT SPOTLIGHT MSC Software Corporation 1 E MD Nastran and Marc Running q A k H H e on SGI Altix Achieves High Performance and Scalability PRODUCT SPOTLIGHT ARUP Announces Image Based Meshing Software FEA INFORMATION RESOURCE MAGAZINE F information com l Contents 01 02 FEA Information Worldwide Participants sgi FUJITSU NEC croso QLOGIC intel est crow Y deet tA x Qas r EST MSC SOFTWARE ARUP NW Index FEA Announcements IBM AMD New IBM System x Servers with the AMD Opteron Processor 03 Turbo Charge a Cooler Design Center Speeds Race Car Design and Production by Eighty Percent 05 MSC Software SGI MD Nastran and Marc Running on SGI Altix Achieves High Performance and Scalability Benchmarks 12 CADFEM GmbH Publication Drop Testing Simulation of a Cooker Including Foam Packaging and Pre stressed Plastic Foil Wrapping 13 Yahoo Yammerings 19 ARUP Arup announces Image Based Meshing Software 21 FEA Weekly News Page July Highlights 22 Consulting and or Distribution August Review 23 LSTC Training Classes 24 EVENTS 25 LS DYNA Resource Page 31 Hardware amp Computing and Communication Products 32 Software Distributors 34 Consulting and Engineering Services 35 Educational amp Contributing Partici
7. USA Austin TX USA Windsor CA www schwer net SECS USA Corvallis OR www predictiveengineering com USA Neenah WI www structuretechnology com Leading Engineering Analysis Providers Greg Horner info leapaust com au 02 8966 7888 Metal Forming Analysis Corporation Chris Galbraith galb mfac com 613 547 5395 Altair Engineering India geet Dias info in altair com 0 80 2658 8540 Altair Engineering Italy Sales altairtorino it Numerica SRL info numerica srl it 39 055 432010 Arup er Walker brian walker arup com 121 213 3317 SE L C Khanh Bui kdbui sbcglobal net 512 363 2739 SE amp CS Len Schwer len schwer net 707 837 0559 Predictive Engineering George Laird 1 800 345 4671 george laird predictiveengineering com Structure Incorporated Todd L Peters 920 722 7060 info structuretechnology com F information com 35 Educational amp Contributing Participants Alphabetical Order By Country Dr Quing Zhou Tsinghua University Dr Anindya Deb Indian Institute of Science Prode Elasis amp Univ of Napoll Frederico II Dr Alexey Borovkov St Petersburg State Tech University Professor Gennaro Monacelli Ohio University Dr Ala Tabiei University of Cincinnati F information com 36 FEA Information China Participants Software Hardware Training Consulting Services Arup China LS DYNA sales support training Ansys China Inc Arup Beijing Yuntong Fore
8. be high and larger size particles where the stress value is low Response by Jim Kennedy From my limited use of the SPH feature do not think you want to do this without some extensive testing validation of your meshing Statements have read from an early paper links below appear to not recommend such A proper SPH mesh must satisfy the following conditions it must be regular as possible and must not contain large discrepancies of the mass of particles F Ginformation com 16 http www dynalook com documents Feainformation nov_2001 pdf Because of the lack of a numerical grid the SPH processor requires some conditions tn setting the initial particle mass and coordinates The particle mesh needs to be enough regular AS a consequence the particles of a same material which have the same initial density need to have the same volume To preserve this they need to be distributed on a uniform mesh http www dynalook com documents 6th_International_Is dyna session7 3 pdf This is my present knowledge however there always is the possibility that something new has been recently implemented that am not aware of to allow more flexibility in the SPH use Response by Len Schwer SPH particles do not have a size rather they have a smoothing length or sphere of influ ence over which the field variables displacement strain stress are interpolated Increasing the smoothing length with improve the interp
9. expecting delegates and presen tations from all over the world to discuss problems facing the industry today and in the future Dr John Hallguist will be a keynote Speaker presenting new LS DYNA fea tures Specifications for Paper Submissions e Paper length 3 000 words maxi mum e Abstract deadline January 26th 2007 e Notification of acceptance Febru ary 9th 2007 e Final paper deadline April 1st 2007 For information on Paper Submissions Exhibit Booth and or Available Sponsor Ships Contact e Dr Marcus Redhe Engineering Research Nordic Brigadgatan 16 SE 581 31 Linkoping Sweden F GAnformation com Whatever your environment there s an SI server that can keep your data m oving Whatever your environment there s an SGI server that can keep your data moving As workflows become more diverse it s critical that your server architecture is flexible enough to efficiently power a wide range of applications That s why we re expanding our innovative line of Linux OS based solutions to de liver rapid insights and maximize your investment even in the most diverse envi ronments Introducing our new SGI Altix XE line of servers and clusters Cluster your power and maximize value A fully integrated cluster solution based on the Intel Dual Core Xeon 5100 proc essor architecture the value priced SGI Altix XE line makes expanding your capa bilities easy and affordable e Industry standard Linux support
10. g stress strain curves LS DYNA will extrapolate from the last two points This is why it is important that the last two points do not have a slope that will produce an unrealistic behavior or introduce numerical instability However also know that this does not apply to data input as thermal parameters e g yield stress vs temperature In these cases LS DYNA will not extrapolate and an error will occur if the temperature goes outside the range supplied Therefore the temperature range defined must cover the entire spectrum of temperatures seen by that material during the analysis For curves relating to loading nodal loads element pressures gravity etc in general LS DYNA will reduce the value of the function to zero once the end point on the curve has been passed So for loading curves it is important that the curve covers the full period of the analysis Follow up Response by Len Schwer Conrad provided a detailed and helpful answer But the bottom line is the USER needs to determine what happens when the limits of a Define_Curve are exceeded this can often be accomplished via a single element simula tion What LS DYNA does after the limits of a Define_ Curve are exceeded may be either correct incorrect depending on what the user is try to accomplish with the model How can the SPH particle size be varied How can we vary the particle size in SPH want to use smaller size particles in the corners where the stress might
11. information com 17 For your problem consider increasing the number of integration points in the solids via ELFORM 2 fully integrated selected reduced Also consider using the thick shell solid ele ment SECTION_TSHELL which has worked well for me in the past Finally you answered your own question when you stated the results are about 30 dif ferent when two layers of solids are used Beam Element Orientation Node Question Hello I m currently modeling a reinforced concrete column subjected to blast load and de cided to use beam elements to model the reinforcement and stirrups However when run the solver it returns an error message that need to define the orientation node even when use Type 2 beam elements Because my model is quite large it will be tedious to define an unique orientation node for each beam element Can someone please give advice on how to overcome this problem i e how can use beam elements but not defining or just defining orientation node vector once as reference for all beam elements Response by Conrad zatt The Type 2 beam Hughes Liu requires a third orientation node since the beam cross section does not have to be circular The section could be rectangular in which case LS DYNA would need to know which directions were the major and minor axes do not be lieve it is possible to use the Type 2 beam without an orientation node Beam elements are allowed to share a third node Since presume
12. more details Len Schwer will be conducting an LS DYNA Discussion Group Forum during the 77th Shock amp Vibration Symposium October 29 November 3 2006 at the Hyatt Regency Monterey in Monterey CA Visit www saviac org for more details This installment of Yahoo Yammerings features five questions with responses from the past month of postings to the LS DYNA Yahoo Group CONSTRAINED LAGRANGE IN SOLID question For the Load Curves in MAT_ 24 what does LS DYNA do How to vary the SPH particle size Question about modeling thin structures with solid elements Beam Element Orientation Node Question a nb Question on CONSTRAINED LAGRANGE _IN_ SOLID have come across a technical paper where someone has used CONSTRAINED LAGRANGE IN SOLID to couple Lagrangian re bar inside a Lagrangian concrete structure It appears that this approach has been taken in order to overcome meshing of the concrete interior to accommodate the internal reinforcements It was my understanding that the CONSTRAINED_LAGRANGE_ functionality was for coupling La grangian meshes to Eulerian or ALE type structures as a means of achieving load transfer did not think that it was applicable for Lagrange to Lagrange meshes Can someone shed any light on this Response by Conrad I zatt CONSTRAINED LAGRANGE IN SOLID can be used to anchor the rebar generally beam elements inside the concrete solid elements Both concrete and rebar are Lag
13. on a single computer 6 may be solved using parallel computers In addition to the DDM technology MD Nastran has the most advanced Distrib uted Memory Parallel DMP technology for explicit nonlinear NVH and multi disciplinary applications SGI Altix 4700 Platform l Wi S s L is AL TIX 4700 OPEN The distinctive modular blade design of the SGI Altix 4700 platform is made of interchangeable compute memory I O and special purpose blades for plug and solve configuration flexibility SGI s novel blade to NUMAlink architecture enables mixing and matching eight standardized blade choices allowing right sizing the system for a given application The com pact Altix 4700 rack packages the blades for industry leading performance density Additionally the Altix 4700 server with dual core Intel Itanium 2 processors offers leading floating point performance per watt of power consumption compared to other high end platforms Designed for future upgrade expansion and integration of next generation HPC technologies the Altix 4700 platform is F information com designed to be socket compatible with up coming single and dual core Intel Ita nium 2 processors offering an easy up grade or expansion of memory I O or visualization capabilities With this flexible growth path customers can tune system configurations for current and changing requirements easily and cost effectively Through pe
14. simulation a particular focus will be how the appropriate de ployment of simulation can lead to a further competitive advantage through helping to stimulate innovation Held at the Westin Bayshore Hotel Vancouver on May 22nd 25th 2007 x I JAPAN LS DYNA Users Conference 2006 x p Tokyo Japan Hosted by JRI x Lith Korea LS DYNA Users Conference 2006 x Seoul Korea Hosted by Theme Engineering Inc LS DYNA Users Meeting in Ulm x Hosted by DYNAmore x 2006 CADFEM Users Meeting x Oct 25 27 International Congress on FEM Technology Stuttgart area Germany x _ Aerospace Design Expo 06 now Anaheim CA US May 28 29 in l F information com LS DYNA Resource Page Interface Hardware OS And General Information Participant Hardware OS that run LS DYNA alphabetical order LS DYNA has been fully QA d by Livermore Software Technology Corporation for All Hardware and OS listed below TABLE 1 SMP TABLE 2 MPP Interconnect and MPI TABLE 1 SMP Fully QA d by LSTC AMD Opteron FUJITSU Prime Power FUJITSU VPP HP PA 8x00 HP Opteron IBM Power 4 5 IBM Power 5 INTEL IA32 INTEL IA64 INTEL Xeon EMT64 SGI Mips SGI IA64 SUSE 9 with ProPack 4 Red Hat 3 with ProPack 3 25 F information com LS DYNA Resource Page MPP Interconnect and MPI FEA Information Inc Participant s alphabetical order Fully QA d by Livermore Software Technology Corporation AMD Opteron FUJITSU Prime Po
15. that all of your beam elements are circular and therefore the orientations are irrelevant you could use the same orientation node for all of the beam elements as long as it doesn t line up with the 2 nodes of any of the beam elements However this is not what would generally recommend If all of the beam elements have different orientations then looking at the shear forces and bending moments about the lo cal s or t axis will give results in all different directions This could be countered by using an orientation node that is far away from the model This way the beam orientations will only be slightly different LS DYNA only uses the third node to set the beam section orientation at the beginning of the analysis Once the analysis is going the beam orientation is determined by the nodal rotations Many post processors are unable to track the current beam orientation and con tinue to use the position of the third node which is not correct although it may be ap proximately correct depending on how much the beam orientation has changed and how much the third node has moved More generally would recommend that all beam elements should have unique third nodes and that the NREFUP parameter on CONTROL_ OUTPUT is set to 1 This means that the current orientation of the beam element can be tracked in the post processor because the co ordinates of the third node are updated according to the current beam orientation F information com 18
16. 1 4 CPU bandwidth compute blade performance 4 CPU density compute blade performance 8 CPU density compute blade performance MD Nastran Linear Scalability with HPC SGI Altix 4700 and Dual Core In tel Itanium2 Processors Running MD Nastran with such advanced Grid Computing technologies as DDM and DDP on the SGI Altix 4700 platform with dual core Intel Itanium 2 processor provides linear scalability in many applica tions and super linear scalability in others The resulting reduction of FEA solution times with parallel processing is very cost efficient Because the law of diminishing returns does not impact processing speed even up to 128 processors used in the Wall Time CPU Time 9224 66 9200 24 2800 69 N A 2694 10 N A 1708 97 N A benchmarks customers can reduce capital investment costs by replacing slower more expensive proprietary RISC based mainframe platforms with the SGI Altix 4700 Server The end result is to reduce solution time hardware costs and allow the time saved to be invested in additional Simulations sensitivity studies and what if studies to further improving products For Information Visit MSC Software F information com 12 Publication Showcase Paper is located online FEA Publications sidebar link Featured Drop Testing Simulation of a Cooker Including Foam Packaging and Pre stressed Plastic Foil Wrapping Dan Neumayer Bosch Siemens Hausgerate Traunreut Germany Madhukar Ch
17. AE process from model building and dataset preparation to result evaluation and design comparisons EASi CRASH DYNA can be used for con cept crash FE crash and coupled rigid body FE crash simulations in conjunction with MADYMO EASI CRASH DYNA s main features in clude e Support of all keywords of LS DYNA 970 971 e Powerful mesh editing features such as automesh and remesh e LS DYNA MADYMO coupling capa bilities for pre and post processing support of MADYMO format till ver sion 6 2 2 e Model Assembler for organizing the model through sub assembly sub models and included files Enhanced Weld tools for manipula tion of connections and Weld com parison Simple dummy positing and seat belt routing Pre and Post processing in same environment Superpose and merge multiple models Animation and plotting Process compatible Full capability to handle IGES CATIA V4 CATIA V5 UG and DU D F is D E m LC Dr F information com 31 Hardware Computing Communication Products AMD www amd com K invent www hp com www intel com SOL www sgi com Microsoft www microsoft com cO FUJITSU www lujitsu com www ibm com servers deepcomputing NEC www nec com XX QLOGIC www qlogic com F ainformation com Software Distributors Alphabetical order by Country Australia Canada China China China Germany German
18. F ainformation com 3 New IBM System x Servers with the AMD Opteron Processor Turbo Charge a Cooler Design Center Speeds Race Car Design and Production by Eighty Percent Reprint from website AMD Copyright 2006 Advanced Micro Devices Inc ef x Red Bull racing F Bi b _ Red Bull s Christian Klien in qualifying at the 2005 United States Grand Prix Red Bull Racing Revs Formula One Design Engine with IBM and AMD Armonk NY August 1 2006 Press Release Conference NYC IBM announced that Red Bull Racing a division of Red Bull Energy Drink has accelerated its Formula One design stu dio s computing power with new IBM Sys tem x servers based on the AMD Opteron processor The high performance comput ing cluster which will contain more than 3 5 million in new equipment will in crease Red Bull Racing s computing mus cle and help the company further improve its race car design and speed car produc tion by a factor of eighty percent Red Bull Racing intends to use the new systems specifically for aerodynamic modeling Aerodynamics is a critical factor In successfully designing a Formula 1 car as they must be designed to maximize the ideal amount of downward force while minimizing wind resistance The IBM solu tion will help Red Bull Racing quadruple its computing power to design more accurate models for virtual testing before starting costly physical product development The competiti
19. al or affordable using physical tests However the use of simulation to reduce the num ber of physical tests before running a physical test for final validation allows a substantial reduction of time and costs Furthermore the time saved can be used for more what if simulations providing a better understanding of product perform ance issues leading to greater innovation and even better products Benchmarking HPC Performance and Scalability Today HPC systems commonly consist of large SMP and blade systems and com puter clusters with each computer con sidered a node and connected by LAN or a more specialized high speed system inter connect Ideally the more nodes proces sors the shorter the solution time In ex ceptional situations super linear scalabil ity occurs that is for double the number of processors the rate of solve time reduc tion is greater than double In uncommon Situations scalability is linear for double the number of processors the rate of solve time reduction is double However the most common scaling follows the law of diminishing returns for double the num ber of processors the rate of solve time reduction is less than double and eventu ally additional processors provide little if any reduction in the solve time reduction rate Various components of the simulation en vironment influence solution time includ ing the operating system simulation soft ware processor and hardware ar
20. atiri CADFEM GmbH Grafing b Munich Germany Mathias Hoermann CADFEM GmbH Grafing b Munich Germany Abstract The present work deals with the numerical simulation of a drop test of a cooker in cluding packaging foam and plastic foil wrapping Additionally the pre stressing of cooker and packaging due to thermal shrinkage of the plastic foil has been taken into account in the numerical investigation For this a thermal pre stressing simula tion of the plastic wrapping has been included before the actual drop test of the whole assembly has been conducted The permanent deformations of the cooker nearby the impacted edge as well as the deformation of package foam in the vicinity of the impacted edge were the primary areas of interest and compared with experi mental data was used to perform the drop test simulation of the cooker as well as the thermal pre stress simulation of the plastic wrapping F information com 13 Yahoo Group Yammerings LS DYNA Yahoo Group Please note the LS DYNA Yahoo Group is neither owned or operated by LSTC and LSTC has no control over content Jim Kennedy Len Schwer KBS2 Inc Schwer Engineering amp Consulting jmk kbs2 com Services Len Schwer net Jim Kennedy amp Len Schwer plan to attend the 5th German LS DYNA Forum spon sored by DYNAmore in Ulm Germany on 12 amp 13 October If you see us please in troduce yourself as a Yahoo Group participant or Yammerings fan Visit www DYNAmore de for
21. chitec ture and run time resource usage man agement tools The result is a need for performance and scalability benchmarks providing a better understanding of vari ous combinations of software and hard ware platforms In conjunction with F ainformation com MSC Software SGI has run a series of benchmarks achieved by using the power ful combination of MD Nastran and a SGI Altix 4700 Server built using dual core In tel Itanium 2 processors MD Nastran and Marc MD Nastran simulation software commer cially available today combines such best in class technology platforms as Nastran Marc Dytran and LS Dyna into one fully integrated coupled multidiscipline simu lation solution for the enterprise Through such industry specific disciplines as im plicit long duration small deformation and explicit crash or drop short dura tion large deformation nonlinear MD Nastran is able to support all the disci plines necessary to meet the needs of all manufacturers i e results that very closely represent real world behavior The parallel processing algorithm of MD Nastran implicit nonlinear technology util izes the Domain Decomposition Method DDM delivering dramatic performance scalability DDM is a methodology for di viding large simulations into smaller simu lations These usually artificial subdivi sions of a large simulation allow the intro duction of parallelism For example simu lations intractable
22. e available With unique expertise in the fields of product design manufacturing simulation and testing LEAP brings optimum solu tions using the world s leading CAD and CAE tools Expertise in advanced analysis software packages such as ANSYS and LS DYNA Engineering services to Government and commercial clients Services include the application and development of compu tational mechanics techniques with spe clalizations in nonlinear transient phe nomena and constitutive modeling With a specialization in the application of the LSTC code LS DYNA Predictive Engineering Predictive Engineering is a mechanical en gineering consulting company specializing in finite element analysis FEA Under this banner a broad range of capabilities are brought to bear in developing predic tive engineering models via expertise in thermal fluids CFdesign drop testing and impact analysis LS DYNA mw w Structure Incorporated Structure is a company targeted expressly towards Advanced Analysis Engineering and Design They support the aerospace and industries with expert consulting and contract engineering in the fields of fluid dynamics structures and advanced pro pulsion Structure meets the definition of a small business as required for Federal Contracting purposes and particularly specializes in the expert application of NASTRAN and LS DYNA F information com LSTC Training Classes 2006
23. er I O the SGI Altix 4700 is the first SGI platform designed to support new computing paradigms such as recon figurable computing with field program mable gate arrays FPGAs which take over where Moore s Law ends SGI s shared memory NUMAflex archi tecture provides scalable system size for simplified programming administration workload management and very high sus tained performance With NUMAlink in terconnect up to 512 processors 1 024 cores under one instance of Linux and as much as 60TB of globally addressable memory can be supported NUMAlink leads the industry in interconnect band width as well as latency for superior per formance on cluster applications Dual Core Intel Itanium 2 Proces sors The SGI Altix 4700 platform used for the benchmarks that follow feature 1 6GHz dual core Intel Itanium 2 processors with 24MB L3 caches 24MB cache is shared by two cores These processors were designed for such data intensive applications as HPC computing The processors provide excellent proper ties for parallelism and scalability which it so happens is exactly what high fidelity FEA models need when being solved for NVH crash nonlinear and multidiscipline Simulations Three Car Crash Benchmark When a van crashes into the rear end of a compact car which in turn crashes into a midsize car the simulation is very com plex A high fidelity model is necessary for real life results and because
24. es 38 seconds 8 8 hours 10 minutes 29 seconds 16 4 hours 3 minutes 47 seconds 32 2 hours 11 minutes 42 seconds Solid Cube Benchmark The model is a square cube consisting of 300 763 eight node isoperimetric brick elements and 314 432 grid points A 10 increment non linear analysis with the MARC solver2 or MD Nastran Sol 600 which employs an iterative method with an incomplete Cholesky pre conditioner is done The solver is sensitive to the amount of aggregated memory band width the computer system can provide This is where the bandwidth compute blade of the A4700 really shines Application MSC Marc or MD Nastran Sol 600 Hardware SGI Altix 4700 Server Beer 1 6GHz dual core Intel Itanium 2 processors with 24MB L3 cache 24MB cache is shared by two cores Operating System Propack 4 SP3 Program Sizing and Requested Options Max number of elements in any dist load list JO Maximum number of point loads O J Load correction flagged or set L V Tape no for input of coordinates connectivity 5 o of different materials 1 max no of slopes umber of points on shell section 3 5 N Number of lists of distributed loads Option for debug print out alues stored at all integration points ee No of di N F information com 10 New style input format willbeused gt Extended precision inputisused Marc input version uppress echo of list items uppress echo of bc summary Pd uppress echo of NURBS data
25. esident IBM System x The power and precision of our systems coupled with our deep industry expertise makes us a perfect choice for Red Bull Racing as it positions itself as a world leader in Formula 1 racing Red Bull Racing built out its existing high performance computing cluster from IBM with Dual Core AMD Opteron processor based System x servers The entire plat form is comprised of a mix of System e325m e326 x3455 x3655 x3755 and 4 Intellistation workstations IBM System Storage DS4800 with 100 terabytes of ca pacity and a General Parallel File System GPFS were also added to improve data storage and high speed file access to ap plications executed on multiple nodes of its computing cluster The high performance calculations and optimized software performance enabled by the AMD Opteron processor will help Red Bull Racing meet the incredibly com petitive demands of Formula 1 racing said Kevin Knox vice president World wide Commercial Business AMD IBM System x servers successfully marry AMD s innovation at the processor level with IBM s system level expertise provid ing customers who demand the most ad vanced technology with truly pioneering solutions For more information about Red Bull Racing go to Red Bull For more information about I BM go to BM F GAinformation com MSC Software Corporation MD Nastran and Marc Running on SGI Altix Achieves High Performance and Scalabili
26. ete Product Service Information can be found on company websites Information is copyright to the respective listed companies NE Rack mountable servers operated as Linux Clusters have gained considerable recognition as working tools for scientific and engineering calculations Whenever the software allows to distribute the cal culation over several nodes these Linux Clusters offer tremendous compute power at a very competitive price performance point ESI GROUP EASI CRASH DYNA is an integrated pack age for crash simulation which covers the CAE process from start to finish It achieves this by integrating all aspects of model building dataset preparation re sult evaluation and design comparisons IBM IBM Deep Computing Visualization pro vides a scalable collaborative middle ware infrastructure to help support and enhance the graphics functions of OpenGL software applications A Deep Computing Visualization implementation is economi cal and highly adaptable solution designed with a commitment to interoperability based on open standards QLogic InfiniPath InfiniBand HCAs smoke the competition with the industry s highest message rate Combined with the lowest MPI latency and highest effective band width this enables MPI and TCP applica tions to scale to thousands of nodes with unprecedented price performance Now we bring you InfiniPath InfiniBand HCAs that set new performance and scalability records on nearl
27. gy Scannina 2 Image Processing 4 gt FE CFD RP amp CAD 3D imaging data ScanlIP is used to import 3D imaging data from MRI CT Micro CT and Ultrasound scans It provides a series of image proc essing and segmentation tools which allow the user to define areas of interest in the image based on grey scale values Smoothing Algorithms The smoothing algorithms used by ScanlIP are volume topology and geometry pre serving This ensures the accuracy of both the generated surface reconstructions and mesh models is based on image accuracy alone The segmented areas can then be exported as a 3D stereo lithography file or exported into ScanFE for meshing The stereo lithography files can either be used directly for producing rapid prototype parts or imported into CAD software Model F ainformation com One Step Operation 20 contacts boundary conditions and loads can be applied In a simple one step operation ScanFE generates the mesh and material proper ties for the solid parts created in ScanlP The material properties can be derived from the greyscale values in the image An LS DYNA keyword file can be exported from ScanFE into Oasys PRIMER where Information To find out more about Simpleware and Oasys Software please contact Brian Walker of Arup brian walker arup com tel 44 0 121 213 3317 F information com 21 Previous Month J uly Weekly News Page Highlights Review Compl
28. he group by sending an email request to LS DYNA subscribe yahoogroups com or by visiting the Yahoo Groups web site http groups yahoo com Generally the quickest best responses are to those questions posed with the most specif ics General questions such as How do use XXX feature either go unanswered or are answered by Jim Kennedy with links to appropriate references in the growing LS DYNA re lated literature e g see the archive of LS DYNA Conference proceedings at www dynalook com F information com 19 ARUP Announces I mage Based Meshing Software Copyright Arup 2006 Arup are pleased to announce the release of a new Suite of software for image based meshing developed by Simpleware Ltd at the University of Exeter Arup are supporting the development of this software and distributing it Three dimensional replicas Simpleware provides what is effectively a 3D photocopier three dimensional repli cas can be generated automatically based on scans In parallel computer simula tions can be used to assess the suitability or performance of objects in operation Simpleware s technology has opened up FEA and RP manufacturing to a variety of applications and research fields including e Industrial reverse engineering Research in materials and compos ites Non destructive evaluation NDE Biomechanical Research Implant design and manufacturing Surgery simulation and planning Forensics Biomimicry Archaeolo
29. it is a very complex simulation it will be a very com pute expensive simulation This makes it a good example for simulation using HPC The vehicle models were created by the National Crash Analysis Center NCAC and assembled into an input file by Mike Berger consultant Livermore Software Technology Corp LSTC The MD Nastran model was prepared by Casey Heydari Sr Manager MSC Software Corporation The model consists of 827 627 grid points and approximately 795 160 elements F information com 8 Three Car Crash Benchmark Application MD Nastran Sol 700 explicit nonlinear Hardware SGI Altix 4700 Server Processor 1 6GHz dual core Intel Itanium 2 processors with 24MB L3 cache 24MB cache is shared by two cores Operating System Propack 4 SP3 Program Sizing and Requested Options umber of Nodes x umber of Solid Elements umber of materials or property sets umber of beam elements umber of shell elements 85 021 umber of rigid body constraint sets umber of rigid body merge cards umber of rigid wall definitions umber of joint constraints umber of extra node blocks umber of spring damper material types umber of discrete springs and dampers umber of lumped masses umber of load curves umber of accelerometers F information com 9 Scalability Linear SGI Altix 4700 Series Results Number of Elapsed Elapsed Time Hours Dual Core Time Sec CPUs onds 4 16 hours 1 minut
30. mple as linkages and as complex as full vehicles MSC Software MSC Dytran LS DYNA Tightly integrated solution that combines MSC Dytran s advanced fluid structure interaction capabilities with LS DYNA s high performance structural DMP within a common simulation environment In novative explicit nonlinear technology enables extreme short duration dynamic events to be simulated for a variety of industrial and commercial applications on UNIX Linux and Windows platforms Joint solution can also be used in con junction with a full suite of Virtual Prod uct Development tools via a flexible cost effective MSC MasterKey License System Filled Fuel Tank Side Impact With Fuel Oil Inside F information com MSC Software MSC Nastran SOL 700 The MSC Nastran Explicit Nonlinear product module SOL 700 provides MSC Nastran users the ability access the explicit nonlinear structural simulation capabilities of the MSC Dytran LS DYNA solver using the MSC Nastran Bulk Data input format This product module offers unprecedented capabilities to analyze a variety of problems involving short dura tion highly dynamic events with severe geometric and material nonlinearities MSC Nastran Explicit Nonlinear will allow users to work within one common mod eling environment using the same Bulk Data interface NVH linear and nonlin ear models can be used for explicit appli cations such as crash crush and drop test simulation
31. olation by including more particle to particle interactions and increase the CPU cost You can also increase the number of particles in a region of the mesh and improve the in terpolation Question about modeling thin structures with solid elements am modeling a thin structure with both shell elements and solid elements The thin struc ture is subject to a static bending load The geometry of the structure is quite complicated The solid element model has two layers of 8 node hex elements The results between the Shell model and solid element model are quite different about 30 cannot increase the layer of solid elements because of long run times My question are two layers of solid ele ments enough to capture bending behavior of a thin structure The mesh is quite fine in both shell and solid element Response by Jim Kennedy believe the usual guideline for most constant stress solid element models to capture bending properly is 5 elements in the thickness direction At least this is my guideline Thin structures are usually best modeled with shell elements Follow up Response by Len Schwer think Jim Kennedy s guideline is appropriate for models that include plasticity which it the typical case for LS DYNA models For elastic bending two elements integration points through the thickness is the mini mum number For plastic bending three elements integration points through the thick ness are the minimum F
32. pants 36 China Participants 38 Informational Websites 39 e European LS DYNA Conference 40 SGI Whatever Your Environment There s and SGI server that can keep your data moving Editor Technical Writers Trent Eggleston Dr David Benson Managing Editor Uli Franz Marsha Victory Dr Ala Tabiei Technical Editor Suri Bala Art Shapiro Technical Consultants Graphic Designer Steve Pilz Wayne L Mindle Reza Sadeghi F information com FEA Information Announcements In the August issue we have included many hyperlinks and with the September issue all links will be converted to hyperlinks Special Announcement and article Arup announces Image Based Meshing Software Courtesy Notice Call For Papers Deadline September 18 2006 NAFEMS World Congress Vancouver Canada May 22 25 2007 Sincerely Trent Eggleston amp Marsha Victory The content of this publication is deemed to be accurate and complete However FEA Information Inc doesn t guarantee or warranty accuracy or completeness of the material contained herein All trade marks are the property of their respective owners This publication is published for FEA Information Inc copyright 2003 All rights reserved Not to be reproduced in hardcopy or electronic copy Note All reprinted full articles excerpts notations and other matter are reprinted with permission and full copyright remains with the original author or company designated in the copyright notice
33. rangian elements The nodes of the beam elements are constrained to remain in the same parametric posi tions within the solid elements F information com 14 The main advantage of this is that the meshes of the rebar and concrete do not have to co incide However this method does not account for bond slip between the rebar and concrete i e it is a fixed connection Response by Len Schwer too read Akram Abu Odeh s paper Application of New Concrete Model to Roadside Safety Barriers from the recent 9th International LS DYNA User Conference talked with Akram briefly about this as too was amazed at this inventive use of CONSTRAINED LAGRANGE IN SOLID so Suggest you try this technique and report back Follow up Response by Akram Abu Odeh may add that the user should be careful about the total mass when there are extensive re bars in a structure If that is the case the user might want to reduce the mass density of the steel and or concrete to get the total mass right if the inertial effects of the structure have significance for a specific solution Follow up Response by Jim Kennedy The following paragraph was taken from an example also appears to be available http www dynasupport com Support howto concrete models Materials 16 72 96 and 84 include the option of considering reinforcement rebar in a smeared fashion Alternately reinforcement can be modeled in a discrete manner using beam element
34. s These reinforcing beams may be merged to the solid concrete elements shared nodes or may be tied to the concrete elements using 1 D contact which can ac count for bond slip or may be coupled to the concrete elements via CONSTRAINED_LAGRANGE_IN_ SOLID CTYPE 2 The last method removes the burden of having to align the beam nodes to the solid element nodes how ever beware the constraint of the rebar to the concrete may conflict with and disrupt other constraints such as symmetry boundary conditions An example of rebar coupled to con crete Is in available or request as newhalf k There is some discussion in the following report http batchelor uc3m es expro D24 pdf Remark 2 on page 5 52 of the LS DYNA Version 970 User s Manual states that the option may also be used to model rebar in concrete or tire cords in rubber It would appear guessing that the nodal representation of the rebar interfacing with the concrete segments lends itself to using the algorithms of the CONSTRAINED LAGRANGE IN SOLID feature F information com 15 For the Load Curves in MAT_ 24 what does LS DYNA do For values of strain beyond the defined stress strain values in the DEFINE CURVE for MAT 24 what does LS DYNA do Does it extrapolate based on the last 2 points or does it as say my colleague the strain increases without any increase in stress Response by Conrad zatt believe that in general for curves relating to material data e
35. s This reduces the time required to build additional models for another analysis programs lowers risk due to information transfer or translation issues and eliminates the need for addi tional software training MSC Software Gateway for LS DYNA Gateway for LS DYNA provides you with the ability to access basic LS DYNA simulation capabilities in a fully inte grated and generative way Accessed via a specific Crash workbench on the GPS workspace the application enhances CATIA V5 to allow finite element analysis models to be output to LS DYNA and then results to be displayed back in CATIA Gateway for LS DYNA supports explicit nonlinear analysis such as crash drop test and rigid wall analysis 29 Gateway products provide CATIA V5 us ers with the ability to directly interface with their existing corporate simulation resources and exchange and archive as sociated simulation data F information com Oasys software for LS DYNA Oasys software is custom written for 100 compatibility with LS DYNA Oasys PRIMER offers model creation editing and error removal together with many Specialist functions for rapid generation of error free models Oasys also offers post processing software for in depth analysis of results and automatic report generation EASi CRASH DYNA EASiI CRASH DYNA is the first fully inte grated environment for crashworthiness and occupant safety simulations with LS DYNA and covers the complete C
36. s a wide range of 32 and 64 bit applications e High density clusters for unmatched energy and space efficiency e A choice of density optimized or O rich server options Custom configured factory integrated clusters for flexible easy deployment Introducing our latest SGI Altix 450 servers Manage every workload and scale with ease Built on the advanced Intel Itanium 2 processor architecture SGI s Altix server line speeds time to insight with powerful performance for even the most complex workloads e Uncompromised Linux performance and scalability e Leading energy and space efficiency for high demand applications e Proven reliability availability and serviceability e Low cost of development administration and management Get our new white paper to learn more A Multi Workflow Architecture for High Performance Computing 40
37. ty Benchmarks MD Nastran powerful simulation software commercially available today com bines such best in class technology platforms as Nastran Marc Dytran and LS DYNA To drive down costs customers are reduc ing the number of expensive and time consuming physical tests by Increasing the use of simulation This is increasing the level of detail in finite element FE mod els making them more and more com plex In turn this is driven by customer demand for higher accuracy results and an expanding number of regulatory test re quirements However the use of high fi delity FE models in the growing number of more complex simulations leads to longer simulation times To reduce simulation time organizations are accelerating the use of High Performance Computing HPC Benchmarks of such platforms as MSC Software s MD Nastran and MSC Marc with advanced Grid Computing technologies and the new SGI Altix 4000 HPC platform with dual core Intel Itanium 2 processors demonstrate a crucial and beneficial impact on cost cut ting Innovation and speed to market leading to reduced warranty and field fail ure costs Less Cost More Innovation Computer simulations cost far less and require much less time than physical tests Additionally simulations provide data not always available or practical with physical tests For example design of ex periments DOE what if studies and sen sitivity studies are not always practic
38. ve nature of the racing in dustry requires us to be on the cutting edge of technology so that we can speed past our competition in car design said Nathan Sykes Computational Fluid Dy namics Manager Red Bull Racing Every second counts whether it s in our design labs or on the racetrack This new state of the art computing platform built on IBM systems now enables us to stream line research and processes associated with simulation testing and shorten the time it takes to bring our designs to life IBM was the only vendor that could pro vide us with a complete solution to ad vance our business Red Bull Racing s design models have seen an increase in size by twenty percent in the past year alone Along with the need for a high performance computing cluster to crunch a growing number of calculations accurately Red Bull Racing needs to continuously increase the speed in which the calculations are completed to maintain its lead against the competition With IBM System x powered by the AMD Opteron processor Red Bull Racing was able to leverage IBM Xcelerated Memory F information com Technology to access data stored in memory more quickly With IBM s eX tended I O Red Bull Racing will also have the ability to add additional accelerator or graphics cards to the system We are very excited to work with Red Bull Racing and do our part to help ad vance the world of Formula 1 said James Gargan vice pr
39. ver CPC Co Ltd Engineering Technology Asso clates China Inc Hewlett Packard Asia Pacific Ltd IBM China MSC Software Corp Kimbal Virdi Tel 86 21 5396 6633 Contact Kimbal virdi arup com Website www arup com dyna Tel 86 10 84085558 Website www ansys com cn Contact China ansys com cn Kimbal Virdi Tel 86 21 5396 6633 Contact Kimbal virdi arup com Website www arup com dyna Tel 86 10 82561200 01 03 Website http cpc ytforever com Sole Distributor of LINUX NETWORX INC USA in China Contact service ytforever com Martin Ma Tel 86 21 64385725 Contact support eta com cn Jerry Huang Tel 86 10 65645261 Contact Huang hp com Ms Ling WANG Tel 86 10 6539 1188 x4463 T L 901 4463 Website http www iom com cn Contact wangling cn ibm com Tel 86 10 6849 2777 Website www mscsoftware com cn Contact mscprc contact mscsoftware com F information com 37 FEA Information China Participants Software Hardware Training Consulting Services Carl Zhang SGI Tel 86 10 65228868 Ext 3362 Contact carl sgi com Qing Zhou PhD Professor Tsinghua University Department of Automotive Engineering Beijing 100084 China Larry Liang Zhongfang Information Tel 86 21 54973162 Technology Ltd Website http www cntech com cn Contact info cntech com cn Yang Xiaojum Zhong Guo ESI Co Ltd Phone 86 020 8235 6272 Contact Yang Xiaojun
40. wer SUN OS 5 8 FUJITSU VPP HP PA 8x00 HP Opteron HP Alpha IBM Power 4 5 IBM Power 5 INTEL IA32 INTEL IA64 INTEL Xeon EMT64 SGI Mips SGI IA64 SUSE 9 with ProPack 4 Red Hat 3 with ProPack 3 F ainformation com 27 TABLE 2 MPP Interconnect and MPI Vendor OS HPC Intereconnect MPI Software InfiniBand SilverStorm LAM MPI MPICH HP lt D Q O AMD Opteron Linux MyriCom QLogic InfiniPath MPI SCALI FUJITSU Prime SUN OS 5 8 FUJITSU VPP HP PA8000 sss SUSE 9 0 HP Alpha IBM Power 4 5 IBM Power 5 LAM MPI InfiniBand Voltaire LAM MPI MPICH HP MyriCom MPI SCALI Linux LAM MPI MPICH HP MPI InfiniBand Topspin Voltaire LAM MPI MPICH HP MyriCom MPI INTEL MPI INTEL IA32 Linux Windows INTEL IA64 ES O D INTEL Xeon EMT64 QLogic InfiniPath SCALI SUSE 9 wProPack 4 NUMAIink InfiniBand Vol MPT Intel MPI taire MPICH NEC SX6 SGI Mips 0 E al e D gt gt RedHat 3 w ProPack 3 SGI IA64 F ainformation com 28 LS DYNA Resource Page Participant Software Interfacing or Embedding LS DYNA Each software program can interface to all or a very specific and limited seg ment of the other software program The following list are software programs interfacing to or having the LS DYNA solver embedded within their product For complete information on the software products visit the corporate website ANSYS
41. y India Italy Italy Japan Japan Japan Korea Korea Leading Engineering Analysis Providers www leapaust com au Metal Forming Analysis Corporation www mfac com ANSYS China www ansys cn Arup www arup com eastasia MSC Software China www mscsoftware com cn CAD FEM www cadfem de DynaMore www dynamore de Altair Engineering India www altair india com Altair Engineering Italy www altairtorino it Numerica SRL www numerica srl it Fujitsu Limited www fujitsu com The Japan Research Institute www jri co p CRC Solutions Corp www engineering eye com Korean Simulation Technologies www kostech co kr Theme Engineering www sdyna co kr 32 F ainformation com 33 Software Distributors cont Alphabetical order by Country Infinite Simulation Systems B V www infinite nl Strela LLC www ls dynarussia com Engineering Research AB www erab se Flotrend www flotrend com tw Engineering Technology Associates www eta com Dynamax www dynamax inc com Livermore Software Technology Corp www lstc com Arup www arup com dyna F ainformation com 34 Consulting and Engineering Services Alphabetical Order By Country Australia Manly NSW www leapaust com au Canada Kingston Ontario www mfac com India Bangalore www altair india com Italy Torino www altairtorino it Italy Firenze www numerica srl it UK Solihull West Midlands www arup com
42. y every HPC application SGI Altix XE offers superior performance and energy efficiency at a breakthrough price point The combination of the advanced Intel Dual Core Xeon processor archi tecture delivered in a fully factory inte grated cluster solution backed by SGI s industry leading service and support pro vides customers with exceptional value in the SGI Altix XE platform ANSYS ANSYS Workbench is the integration tool for our advanced physics capabilities that enable you to model all of your applica tions from very simple to very complex The integration of these and your own solver technologies can take weeks out of your CAE process by eliminating manual file transfer result translation and re analysis time HP Powered by next generation Intel Ita nium 2 processors the HP Integrity Su perdome delivers the flexible capacity re quired to get the work done The Integ rity Superdome with the new sx2000 chipset also offers scalability to 64 Ita nium 2 9M processors in a single server and you can scale out in a cost effective clustered environment to thousands of processors F information com Ze August Consulting and or Distributing Companies Websites Contain Complete Company Information EAC A full service supplier to the metal form ing industry offering consulting services software sales training and support Regularly scheduled classes on the use of LS DYNA for sheet metal forming ar
Download Pdf Manuals
Related Search