Robustness And Reliability in Simulation Based Design Process

Robustness And Reliability in Simulation Based Design

Process

Gaetan Van den BerghSander de Bruijn

Outline

• Introduction

• Research Projects

• Examples & Applications

Introduction

LMS International & Noesis Solutions

NumericalOptimization

ProcessAutomation

Design ofExperiments

RobustDesign

Simulation Process Management Solutions

NumericalOptimization

Design ofExperiments

RobustDesign

Simulation Process Management Solutions

SaveProcess

Time

No moremanual steps

No moremanual steps

FormalizeSimulationProcess

FormalizeSimulationProcess

Exploit Power ofParallelization

Exploit Power ofParallelization

Build Integrated Processes

Build Integrated Processes

NumericalOptimization

ProcessAutomation

RobustDesign

Simulation Process Management Solutions

MoreDesignInsight

ParameterScreeningParameterScreening

UnderstandParameterInteractions

UnderstandParameterInteractions

VisualizeDesignSpace

VisualizeDesignSpace

IntelligentDesign Space

Sampling

IntelligentDesign Space

Sampling

MoreDesignInsight

ProcessAutomation

RobustDesign

Simulation Process Management Solutions

OptimalProduct

Performance

Reach Pre-DefinedTargets

Reach Pre-DefinedTargets

ParameterIdentification &Test Correlation

ParameterIdentification &Test Correlation

ObjectiveTrade-offAnalysis

ObjectiveTrade-offAnalysis

Hunt for“Extreme”

Performance

Hunt for“Extreme”

Performance

MoreDesignInsight

Less Scrap,Warranty

Risks

Simulation Process Management Solutions

ProcessAutomation

OptimalProduct

Performance

FindReliableOptimum

FindReliableOptimum

AssessRobustnessof Design

AssessRobustnessof Design

MinimizeImpact ofVariability

MinimizeImpact ofVariability

CalculateProbability of

Failure

CalculateProbability of

Failure

Challenges for Product Innovation

• Improve competitiveness through Product and Process Innovation– Improve functional performances– Shorten market introduction and reduce development costs– Design for customer expectation

• Current answer: Systematic use of Virtual Prototyping– Applied more and more upfront– Ever larger models, mature techniques, faster computers

• But: What is the Validity and Relevance of these analysis results?– Production and material tolerances – Environmental condition influences – Structural degradation

Non-deterministic Behavior• Designers must consider this Non-Deterministic Behavior

– Understand potential envelope on performances– Understand main sources of response scatter– Understand which are sensitive parameters to control– Take design measures to minimize response variability (Robust Design)– Take design measures to guarantee specified performance (RBDO)

Some Definitions

• Variability– Variations inherent to modeled physical system or environment under

consideration (scatter, tolerances, … with known distributions)– Probabilistic inputs à Probabilistic responses – Reliability analysis: Assure confidence in response limits

– Design For Six Sigma: Guarantee very high reliability• Robust Design:

– Reduce sensitivity to critical parameters which are subject to U & V

– Reliability analysis, RBDO: Update the design to satisfy reliability targets

Reliability and Robustness

Failure Domain

Initial Design

ISO-Objective Function Contours

Search for Optimal point

Safe Domain

Robust & Reliable

Optimum

DeterministicOptimum

Improving Objective

Constraint Boundary

Variable 1

Varia

ble

2

Obj

ectiv

eAssess and Minimize Influence of Design Parameter Variability

Reliability Analysis Methods

• Limit State Approximations– FOSM (First Order Second Moment)– FORM / SORM (First / Second Order Reliability Method)

• Sampling Methods– Monte Carlo Simulation– Importance sampling, Directional Sampling

Research Projects

Some Research & Development by LMS/Noesis in framework of RTD projects

• EC Marie Curie “MADUSE” (2004-2008) on Uncertainty & Variability– EC Marie Curie Research & Mobility Network, 9 partners (Renault, Fiat, …)

across Europe• IWT “Analysis Leads Design” (2004-2007)

– RBDO framework, vehicle crashworthiness, sensitivity, durability, assembly, …

• EC NoE “InMAR” (2004-2008) – Intelligent Materials for Active Noise Reduction (ANC)– Optimization of ANC in presence of variability

• …

Examples & Applications

• AIM: Safer structural design by modifying material properties

• Use of Tsai-Hill criterion

Reliability Application: Composite Wing

+

+−

−=

2

12

12

2

2

22

1

21

2

1

11FFFF

G σσσσσ {{ G > 0 G > 0 SAFESAFE

G G << 0 0 FAILUREFAILURE

Reliability Application: Vehicle Knuckle

-10

-5

0

5

10

-10 -8 -6 -4-2 0 2

4 6 8 10

-2.5

-2

-1.5

-1

-0.5

0

0.5

E

Tensile Strength

Max

imum

Dam

age

-10 -8 -6 -4 -2 0 2 4 6 8 10-10

-8

-6

-4

-2

0

2

4

6

8

10

TS,

Ten

sile

Stre

ngth

E , E last ic Modulus

STARTING POINT

OPTIMIZED POINT

… and much faster Reliability-Based Design Optimization, with comparable accuracy as on full FE model RS RBDO Time

(seconds) FE refinement RBDO Time

(seconds) 26 80920

• Vehicle Knuckle with variability in material propertiesAim: Guarantee that fatigue life is sufficiently long

• DOE+RSM approach has been used, which enabled fast Reliability Assessment …

Robustness and Reliability Application: Slat Track Variability in Geometrical Properties

• Aim: Guarantee Robust and Reliable Fatigue Life Assessment• Measured test data to quantify Variability in geometrical properties• Mesh morphing to include geometrical tolerances

Robustness and Reliability Application: Slat Track Variability in Geometrical Properties

• DOE+RSM approach: fast Robustness & Reliability Assessment• Much faster Reliability-Based Design Optimization, with comparable

accuracy as on full FE model

Robustness and Reliability Application: A-Pillar Trim Impact

• Aim: Guarantee trim panel prevents head injury• Variability in Ribs Thickness, Impact velocity, Stress-Strain curve• Minimize Head Injury Criterion (HIC)• Analysis Procedure:

– Design Space Exploration– Refined Design of Experiments – Response Model using all DOE points

Rib Thickness most important parameter

800

1000

1200

1400

1600

1800

0,5 0,7 0,9 1,1 1,3 1,5

Ribs_thickness (mm)

HIC

(d)

Robustness and Reliability Application: A-Pillar Trim Impact

• Reliability analysis of initial configuration: pf = p(HIC > 1000)

• Deterministic Optimization on the HIC

• RBDO: Reliable optimum practically coincident with deterministic optimum

0

50

100

150

200

250

300

350

400

0,000 0,005 0,010 0,015

Time (s)

Acc

eler

atio

n (G

's)

0,5 mm0,85 mmRib

ThicknessOptimized

Conclusion

Conclusion• Reliability Analysis are key to robust design

• Assess the response scatter & failure probability• Understand root causes • Reliability Based Design Optimization

• Industrial awareness is increasing• Technology is getting to stage where application

to industrial problems becomes feasible• Automotive and Aerospace applications

• Tools (even first product releases in OPTIMUS) are available• LMS and Noesis are active in research & technology development

• On analysis methods• On enabling technologies for fast re-analysis

More information• Visit us at our booth

• OPTIMUS & LMS Virtual.Lab Product Information• Uncertainty, Robustness and Reliability Papers• OPTIMUS & LMS Virtual.Lab Industrial Application Cases

• Register for OPTIMUS Webseminar• Tuesday 2 Oct 2007, 10 am• Send an email to [email protected]

Thank you!

Gaetan Van den Bergh

[email protected]