Comparison of EFG and Standard Elements for the Rubber Membrane in a Biomedical Valve in LS-DYNA...
-
Upload
mikaela-addis -
Category
Documents
-
view
218 -
download
3
Transcript of Comparison of EFG and Standard Elements for the Rubber Membrane in a Biomedical Valve in LS-DYNA...
Comparison of EFG and Standard Elementsfor the Rubber Membrane in a Biomedical Valve
in LS-DYNA 970.5434
Rudolf Bötticher
www.rudolf-boetticher.de
Dresden, 11.11.2004www.rudolf-boetticher.de
Motivation• Assess whether EFG with FSI and MMALE
is possible.• Compare the results with standard elements.• Assess whether EFG is more robust.
Dresden, 11.11.2004www.rudolf-boetticher.de
EFG is easy!*CONTROL_EFG$ in 970 EFG and (dormant) IMPLICIT cards are not tolerated $ in the same deck$ implicit and axisymmetric EFG not implemented!
*SECTION_SOLID_EFG5,41$ the non-default bigger support helps to have consistent $ EFG simulations for *MAT_MOONEY-RIVLIN_RUBBER
1.5,1.5,1.5
Dresden, 11.11.2004www.rudolf-boetticher.de
MM-ALE is easy,if you got a working deck to refine!
*ALE_MULTI-MATERIAL_GROUP
Proceedings
Dresden, 11.11.2004www.rudolf-boetticher.de
FSI: Tweaking of *CONSTRAINED_LAGRANGE_IN_SOLID
TSSFACNADV
METHCTYPE
DIRECPFACILEAK
3854, 5434 and newer beta versions deliver different results for identical decks!
Dresden, 11.11.2004www.rudolf-boetticher.de
*MAT_MOONEY-RIVLIN_RUBBER
*EOS_GRUENEISEN
ELFORM=11
*MAT_NULL
*EOS_IDEAL_GAS
AET=4
Dresden, 11.11.2004www.rudolf-boetticher.de
Why LS-DYNA for this problem?
• Curiosity
• Code is at your disposal
• Expect the same efficiency as for parallel crash simulation
• Rely on the advanced material modeling
CFX may be better
Dresden, 11.11.2004www.rudolf-boetticher.de
Membrane covered with null shells
Filling with *MAT_VACUUM
Dresden, 11.11.2004www.rudolf-boetticher.de
CTYPE=4
CTYPE=5
DIREC=3
PFAC=0.1
Dresden, 11.11.2004www.rudolf-boetticher.de
A=500MPa
A=100MPa
PR=0.49
near incompressibility and mm dimensions require tiny time step
no implicit, no time step split between rubber/ALE domain, no mass scaling!
Dresden, 11.11.2004www.rudolf-boetticher.de
5831 beta version delivers different results.
However, problem not solved.
FSI not robust against artificial shortening of time step.
No difference between EFG and standard elements.
Dresden, 11.11.2004www.rudolf-boetticher.de
Robustness
• It proves difficult generating an extreme situation where EFG works but standard elements do not!
• EFG may be superior preventing hourglass modes.
Dresden, 11.11.2004www.rudolf-boetticher.de
EFGELFORM=1
*HOURGLASS,6
EFG performance lack: 10% elements switched, CPU time +20%
Dresden, 11.11.2004www.rudolf-boetticher.de
Conclusions
• FSI simulations with MMALE and structural EFG solids are possible in LS-DYNA.
• Here no real advantage of EFG over standard elements.
• EFG may be better in hourglass prevention.• Time step dependence of FSI needs further
investigation!