APPLICATION OF AN INTEGRATED HIGH-PERFORMANCE APPLICATION OF AN - - PowerPoint PPT Presentation

APPLICATION OF AN INTEGRATED HIGH APPLICATION OF AN INTEGRATED HIGH-PERFORMANCE PERFORMANCE COMPUTING RELIABILITY PREDICTION FRAMEWORK COMPUTING RELIABILITY PREDICTION FRAMEWORK TO HMMWV TO HMMWV SUSPENSION SYSTEM SUSPENSION SYSTEM

10 October 2010

Dan Dan M

• M. Ghiocel

Ghiocel Dan Dan Negrut Negrut David David Lamb Lamb David David Gorsich Gorsich

GP GP Technologies Technologies Univ

• Univ. of
• f Wisconsin

Wisconsin US US Army Army TARDEC TARDEC US US Army Army TARDEC TARDEC Rochester Rochester NY NY Madison, Madison, WI WI Warren, Warren, MI MI Warren, Warren, MI MI

UNCLASSIFIED: DISTRIBUTION STATEMENT A: Approved for Public Release

An Integrated HPC Reliability Prediction Framework

• Presentation Scope:

– To present on overview of an integrated HPC stochastic physics-based framework that has been developed for vehicle reliability prediction. – Illustrative application to the HMMWV system

DISCLAIMER: The HMMWV dynamic model and the suspension system configuration used in this research are slightly different than the actual HMMWV hardware.

– Focus on:

• The front-left suspension system (FLSS)
• Qualitative aspects and methodology, not on quantitative results

10 October 2010 2

Vehicle Reliability Prediction Process VEHICLE RELIABILITY PREDICTION MULTISTEP PROCESS VEHICLE RELIABILITY PREDICTION MULTISTEP PROCESS

10 October 2010 3

Vehicle Reliability Prediction. Implementation VEHICLE RELIABILITY IMPLEMENTATION CHART VEHICLE RELIABILITY IMPLEMENTATION CHART

10 October 2010 4

Stochastic Road Surfaces

• 10 October 2010

5

• With topography Effects; Smooth and Rough Stochastic Roads
• With topography Effects; Smooth and Rough Stochastic Roads

Vehicle Behavior to Stochastic Road Surfaces

10 October 2010 6

With topography Effects; Smooth and Rough Stochastic Roads Flat road with no topography Effects; Passing a Random Bump

Using Using HMMWV HMMWV Model Model

Vehicle Behavior to Stochastic Road Profiles

!" !"

10 October 2010 7

Straight Road Segment Rolling Hills Segment

ADAMS Model ADAMS Model SPARTACUS Model SPARTACUS Model

# !# # !#

$%&"'( $%&"'( Front-Left Suspension System (FLSS) Models

• 10 October 2010

8

Stochastic FEA for Stress Computation

• )

)

• $ ( *

$ ( *) )+ +

10 October 2010 9

Parallel Multigrid Solver

Block Jacobi Block Jacobi

* *,-../...0./.../... * *,-../...0./.../...# # * *,-../...0./.../... * *,-../...0./.../...# #

Stochastic FEA Using HPC. Scalability Study – Linear, Static

10 October 2010 10

Parallel Parallel Multigrid Multigrid

• .#1
• .#1)

)2 2 Stochastic Response Surface Modeling for Stress Computation

10 October 2010 11

%.3//1 %.3//1

• !

!

• Sensitivity Studies:

Effects of Road Topography on FLLS Joint Forces and Stresses

10 October 2010 12

4#5 4#5 Progressive Damage Modeling for Interactive Mechanisms; HFC, LCF, Corrosion

10 October 2010 13

1 1) )4 4 # # # # $6 7( $6 7(

"## #!$ %$

Crack Initiation Crack Initiation Crack Propagation Crack Propagation

q C p th n m

K)] K ) R 1 [( ) K K ( K ) R 1 ( C dN da ∆ − − ∆ − ∆ ∆ − =

Forman FCG Model (NASA JPL, 1996)

F CF

K ) t ( K ∆ ψ = ∆

) n ( a ) t ( p 1 ) t ( + = ψ

Uncertain Uncertain Parameters Parameters

γ β α

• +

+ = ] ) t ( p ) n ( a [ IDS ) t ( a

t , n

14 10 October 2010 &'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

Uncertain Uncertain Parameters Parameters

Predicted FLSS Life Usind LDR and DCA Progressive Damage Models

## # ## #

Sensitivity Studies: Effects of Damage Mechanism Model on Life Predictions

89-& 89-&

LDR LDR DCA DCA

10 October 2010 15

8:9; 8:9; 89-& 89-&

LDR LDR DCA DCA

<< << <4 <4 $;.=,( $;.=,( Sensitivity Studies: Effects of Vehicle Weight Increase on Predicted FLSS Life

10 October 2010 16

8-.0: 8-.0: 8>&9 8>&9

)0102$ )0102$ ) ) +33)" +33)" 01+02$!)% 01+02$!)%

&4'% &4'% &4'% &4'% + + + +

10 October 2010 17 &'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

)0$+0 )0$+0

5)4 5)4 6" 6" % % 7 7 + + + +

10 October 2010 18

5)4 5)4 6" 6" % % 7 7 + + + +

&'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

)0$00 )0$00

('6" ('6" % % 7 7 + + + +

10 October 2010 19

4'6" 4'6" % % 7 7 + + + +

&'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

$ $ / / /*/( /*/(

Low Nonlinear Low Nonlinear Correlation Correlation

#$ 89 +"1",8,3,2

10 October 2010 20

High Nonlinear High Nonlinear Correlation Correlation

&'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

,16 ,16

Low Nonlinear Low Nonlinear Correlation Correlation High Nonlinear High Nonlinear Correlation Correlation

#$ 89 +"1",8,3,2

10 October 2010 21 &'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

Lack of Data Effects (250 simulations) on Life PDF, and 99% & 95% Reliability Life 1 1

• 10 October 2010

22

8"8"# +$ $! 66# $0;.(7* $0;.(7*

10 October 2010 23

)#+$ 86) !6#6# 7 7 2 2

10 October 2010 24

)#+$ 8% 6#6# 1 1

• 10 October 2010

25

?)!

Critical Length Before Inspection After Inspection Before Inspection After Inspection

Crack Size

22426) 47 47

63)+-# 63)+-# !0.#, !0.#, :%;,6 :%;,6

26 Detection Level Inspection 1 Inspection 2

Time

Time Time

ti Pf

SME UME SME 10 October 2010 &'(')*+,-$,. &'(')*+,-$,./ // /$ $ $" $"

1 1.2 1.4 x 10

• 5

e Failure Risk Evolution including Maintenance Given Probabulity of Failure (1.0e-5) 0.8 1 1.2 x 10

• 3

r e Failure Risk Evolution including Maintenance Given Probabulity of Failure (1.0e-3)

63) 63) !) !) )#. )#. *+5 *+5

10 October 2010 27

500 1000 1500 2000 2500 3000 3500 0.2 0.4 0.6 0.8 Time (days) P r o b a b ilit y o f f a ilu r e 500 1000 1500 2000 2500 3000 3500 0.2 0.4 0.6 Time (days) P r o b a b ilit y o f f a ilu r

$4(72 $4(72

63) 63) !) !) )#. )#. *+5 *+5

10 October 2010 28

7#0.@ 7#0.@ 63) 63) !) !) )#. )#. *!# *!#

10 October 2010 29

+"#":3"""$ ":#. 2"#", 2"# 3))"), 2#3)"", 2"), 2"##)""

Summary/Concluding Remarks

10 October 2010 30

2"##)"" 23)")"$

• ##)")

!!7#$ *# "#$ 3"", "#"38"#")3)$ )##,)#"3"$

• #"3:)$