Ansys WB Solving Large Complex Problems

7/30/2019 Ansys WB Solving Large Complex Problems

1/56

2011 ANSYS, Inc. May 10, 20121

Solving Large Complex

Problems

Efficient and Smart Solutions for

Large Models


2/56

2011 ANSYS, Inc. May 10, 20122

ANSYS Structural Mechanics Solutions offersseveral techniques


3/56

2011 ANSYS, Inc. May 10, 20123

Current trends insimulation show an

increased need forthe computation of

large models


4/56

2011 ANSYS, Inc. May 10, 20124

Larger model size for higher accuracy


5/56

2011 ANSYS, Inc. May 10, 20125

Full assemblies for higher realism


6/56

2011 ANSYS, Inc. May 10, 20126

Better design knowledge from designvariations


7/56 2011 ANSYS, Inc. May 10, 20127

Continuous variations of the parametersprovides in-depth information


8/56 2011 ANSYS, Inc. May 10, 20128

Even more performance is required!

( )+ +

=

x n


9/56 2011 ANSYS, Inc. May 10, 20129

High Performance

Computing is a keycomponent of the

ANSYS offering.


10/56 2011 ANSYS, Inc. May 10, 201210

High Performance Computing as an EnablerInsight you cant get any other way

Its all about getting betterinsightinto product behavior quicker!

HPC enables high-fidelity

Include details - for reliable results

Getting it right the first time CONFIDENCE by DESIGN!

HPC enables design exploration &

optimization

Consider multiple design ideas Optimize the design

Ensure performance across range ofconditions


11/56

2011 ANSYS, Inc. May 10, 201211

High Performance ComputingA Software Development Imperative

Clock Speed Leveling off

Core Counts Growing Exploding (GPUs)

Future performance

depends on highly scalable

parallel software

Source: http://www.lanl.gov/news/index.php/fuseaction/1663.article/d/20085/id/13277


12/56

2011 ANSYS, Inc. May 10, 201212

Size of the model how large is large?

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3 4

Number of CPUs

Speed

Up

1000 elements

8000 elements

64000 elements

512000 elements
http://localhost/var/www/apps/conversion/tmp/Documents%20and%20Settings/mscowler/Local%20Settings/Temp/block.gif


13/56

2011 ANSYS, Inc. May 10, 201213

A simple and productive licensing scheme

ANSYS HPC Pack

ANSYS HPC Workgroup


14/56

2011 ANSYS, Inc. May 10, 201214

Which part of the simulation is faster?


15/56

2011 ANSYS, Inc. May 10, 201215

Not all steps of the simulation are parallel

Model solution


16/56

2011 ANSYS, Inc. May 10, 201216

How should I read speed-up curves?This is the solver partexcellent

scaling!

This is YOUR time (elapsed)


17/56

2011 ANSYS, Inc. May 10, 201217

The rightcombination ofalgorithms and

hardware leads tomaximum efficiency


18/56

2011 ANSYS, Inc. May 10, 201218

Shared Memory Parallel vs DistributedMemory Parallel


19/56

2011 ANSYS, Inc. May 10, 201219

Challenges and solutions for thedistributed method

Challenges Solutions

Efficient and relevant

decomposition

Partitioning methods,

Solver

Load Balancing Partitioning methods,

Solver

Speed Hardware(Processors,

Interconnects), Solver

Maximum Problem Size Hardware (RAM), Solver

I/O to communicate

between cores.

Hardware (Interconnects),

MPI, Solver

I/O to write results and

overflow files during

solution.

Hardware (Disks,

Interconnects), MPI,

Solver


20/56

2011 ANSYS, Inc. May 10, 201220

Sparse or iterative solvers?

Solver type Distributed/Shared

Memory

SPARSE (direct) DMP/SMP

PCG (Iterative) DMP/SMP

LANB (direct, modal) SMP

LANPCG (iterative, modal) DMP/SMP

SNODE SMP


21/56

2011 ANSYS, Inc. May 10, 201221

Get it in-core!

Incore - 24GB

Optimal - 24GBMinimum - 24GB

Optimal -4GBMinimum - 4GB

0

500

1000

1500

2000

2500

Time (s)


22/56

2011 ANSYS, Inc. May 10, 201222

Check the PCG level!


23/56

2011 ANSYS, Inc. May 10, 201223

Balancing the load: a key to efficiency


24/56

2011 ANSYS, Inc. May 10, 201224

A consequence for contact users


25/56

2011 ANSYS, Inc. May 10, 201225

What could it look like on your model?

6 Mio Degrees of Freedom

Plasticity, Contact

Bolt pretension

4 load steps

1 HPCPack


26/56

2011 ANSYS, Inc. May 10, 201226

The Right Software Architecture for HPC

Workstation-Based

or

Remote Access

Job SubmissionHost

ANSYS WorkbenchUser Environment/Graphics

Compute Cluster

Mobile User

Thin Client

Off-the-shelf High-Performance ComputingANSYS Remote Solve Manager: Workbench-based job

submission with full portfolio support for Platform

LSF, PBS Pro, and Microsoft Job Scheduler

Bundled third-party Message-Passing software with

optimized performance (Intel MPI, Platform MPI) on

gigE, 10gigE, or Infiniband cluster fabric.

Multi-core / Multi-node

Solver Execution

Graphics Server


27/56

2011 ANSYS, Inc. May 10, 201227

Our HPC Partnerships

ANSYS maintains close technical collaboration with the leaders in HPC

This mutual commitment ensures that you get the most possible value from

your overall HPC investment

Some current examples:

Optimized performance on multicore processors from Intel, with R&D focusedon Intels Many Integrated Core (MIC)

Over 60% performance boost for the latest Intel Xeon E5-2600

processor (Sandy Bridge) family compared to previous Intel (Westmere)

generation

GPU computing accelerates ANSYS Mechanical today, with very active R&Dengagement with NVIDIA across full portfolio

ANSYS and IBM Optimized cluster and storage architectures for ANSYS


28/56

2011 ANSYS, Inc. May 10, 201228

What do your peers say?

By optimizing our solver selection and workstation

configuration, and including GPU acceleration, weve beenable to dramatically reduce turnaround time from over

two days to just an hour. This enables the use of simulation

to examine multiple design ideas and gain more value out

of our investment in simulation.

- Berhanu Zerayohannes, Senior Mechanical Engineer, NVIDIA

Application: Deflection and bending of 3-D glasses

Software: ANSYS Mechanical

HPC Solution: From 60 hours per simulation to 47 minutes(77x speedup)

Business Ability to ensure robust performance of the 3-D

Solution: glasses via examining multiple design ideas

Copyright20

11NVIDIACorporation.

Allrightsreserved.


29/56

2011 ANSYS, Inc. May 10, 201229

Taking advantage of new hardwaresolutions: GPU

1.52

1.16

1.70

1.20

2.24

1.44

0

1

2

3

V13cg-1(JCG,

1100k)

V13sp-1(sparse,

430k)

V13sp-2(sparse,

500k)

V13sp-3(sparse,

2400k)

V13sp-4(sparse,

1000k)

V13sp-5(sparse,

2100k)

Distributed ANSYS 14.0 Total Simulation Speedups

4 CPU cores

4 CPU cores + 1 GPU


30/56

2011 ANSYS, Inc. May 10, 201230

Speed-up from GPU technology

Cores GPU Speedup

2 no 2.25

4 no 4.29

2 yes 11.36

4 yes 11.51

Vibroacoustic harmonic analysis of an audio speaker

Distributed ANSYS Results (baseline is 1 core):

With GPU, ~11x speedup on 2 cores!

15-25% faster than SMP with same number of cores

Windows workstation: Two Intel Xeon 5530

processors (2.4 GHz, 8 cores total), 48 GB RAM,

NVIDIA Quadro 6000

Speedup

SMP

DANSYSSMP+GPU

DANSYS+GPU

0.00

2.00

4.00

6.00

8.00

10.00

12.00

2

4


31/56

2011 ANSYS, Inc. May 10, 201231

Speed-up from Multi-Node GPU technology

Solder Joint Benchmark - 4M

DOF, Creep Strain Analysis

Linux cluster : Each node

contains 12 Intel Xeon 5600-

series cores, 96 GB RAM,

NVIDIA Tesla M2070,

InfiniBand

Mold

PCB

Solder

balls

ResultsCourtesyofMicroC

onsultEngineering,

GmbH


32/56

2011 ANSYS, Inc. May 10, 201232

ANSYS has a strong commitment to HighPerformance Computing.


33/56

2011 ANSYS, Inc. May 10, 201233

High Performance Computing is dynamic,technology dependent.

1980

2010

2020


34/56

2011 ANSYS, Inc. May 10, 201234

Reductiontechniques help

reduce the CPU time

and can helpcompute designvariations more

efficiently


35/56

2011 ANSYS, Inc. May 10, 201235

Submodeling is the solution when only aportion of the model matters.


36/56

2011 ANSYS, Inc. May 10, 201236

The coarse model provides accuratedeformations but inaccurate stresses


37/56

2011 ANSYS, Inc. May 10, 201237

The refined model(s) will provide accuratestresses


38/56

2011 ANSYS, Inc. May 10, 201238

From the coarse model to the submodelthrough results mapping

Displacements are

mapped to the common

boundary


39/56

2011 ANSYS, Inc. May 10, 201239

Solving two models can be faster thansolving a very detailed one


40/56

2011 ANSYS, Inc. May 10, 201240

General Procedure

1. Create and analyze the initial model

2. Create the submodel

3. Perform cut boundary interpolation

4. Analyze the submodel

5. Verify that the distance between the cut boundaries and the

stress concentration is adequate


41/56

2011 ANSYS, Inc. May 10, 201241

Good practice:Verify the cut-boundary distance

Compare path plots at that location between the initial modeland submodel.

Use the query option.

List the results.

Etc.


42/56

2011 ANSYS, Inc. May 10, 201242

Submodeling with ANSYS Workbench


43/56

2011 ANSYS, Inc. May 10, 201243


44/56

2011 ANSYS, Inc. May 10, 201244

Submodeling worksfor a variety oftopologies and

nonlinear models as

well


45/56

2011 ANSYS, Inc. May 10, 201245

From a solid model to a solid model

Stress contourfull model

Stress contourSubmodel


46/56

2011 ANSYS, Inc. May 10, 201246

From a shell model to a solid model

Initial geometryDefeatured shell model - deformations

Solid submodel - stresses


47/56

2011 ANSYS, Inc. May 10, 201247

Nonlinearities can also be included

Plastic strain

Coarse model

Submodel (line) vs

refined full model

(cross)


48/56

2011 ANSYS, Inc. May 10, 201248

Substructuring or CMS allows for collaborativework or long transient simulations.


49/56

2011 ANSYS, Inc. May 10, 201249

Static (Guyan)

Guyan Reduction procedure

inertia forces are negligible compared to elasticforces

Net result: the reduced stiffness matrix is exact,

whereas the reduced mass and damping

matrices are approximate

Note: Choosing master DOF is an important step in a

reduced analysis, impacting accuracy of results


50/56

2011 ANSYS, Inc. May 10, 201250

Component Mode Synthesis

CMS is a type of substructuring whichperforms a modal analysis of a structure

based on independent modal analyses of

its parts

The synthesis involves making the

components work together as a single

structure by satisfying inter-component

compatibility and equilibrium constraints

Master DOF are required only at

interface nodes


51/56

2011 ANSYS, Inc. May 10, 201251

CMS with ANSYS Workbench


52/56

2011 ANSYS, Inc. May 10, 201252

APDL macros embedded in the simulation treefor generation, use and expansion pass


53/56

2011 ANSYS, Inc. May 10, 201253

Results are available through standardoperations


54/56

2011 ANSYS, Inc. May 10, 201254

Reduce solution time for harmonic andtransient analyses


55/56

2011 ANSYS, Inc. May 10, 201255

Comparing the accuracy of a CMS analysis to astandard one


56/56

Learn about the methods available in ANSYSStructural Mechanics Solutions.

Ansys WB Solving Large Complex Problems

Documents

Transcript of Ansys WB Solving Large Complex Problems