Professional Documents
Culture Documents
Presented By:
Rich Bothmann
Analysis Services Division Manager
IMPACT Engineering Solutions
www.impactengsol.com 2
©2005 – IMPACT Engineering Solutions, Inc.
Implicit vs Explicit
Implicit
Iterates until equilibrium is obtained
If equilibrium is not obtained, time step decreased and matrices
are recalculated
Explicit
Diagonal Mass Matrix allows uncoupling of equations
www.impactengsol.com 3
©2005 – IMPACT Engineering Solutions, Inc.
Range of Structural Problems
Typical 2 meter
drop test velocity
www.impactengsol.com 4
©2005 – IMPACT Engineering Solutions, Inc.
Selecting Elements For Explicit Solvers
Linear elements with reduced integration
points significantly more efficient than full
integration or 2nd order elements.
Reduced integration elements
susceptible to hourglassing (zero energy
modes). Although methods exist to deal
with it, these same methods can
negatively affect results.
Reduced integration elements also more
sensitive to variations in element shape.
To use these elements you must have
nice shaped quad or brick mesh
www.impactengsol.com 5
©2005 – IMPACT Engineering Solutions, Inc.
Meshing Requirements
Many parts that require drop testing do not easily
accommodate shell or brick meshing.
www.impactengsol.com 6
©2005 – IMPACT Engineering Solutions, Inc.
What about Tets?
ANSYS SOLID92 or SOLID187
LS-DYNA Solid168
Accuracy Impact?
Performance Impact?
www.impactengsol.com 7
©2005 – IMPACT Engineering Solutions, Inc.
SOLID168 History
ANSYS & LSTC state SOLID168 not good for high
velocity impacts (i.e. ballistics)
IMPACT also ran studies on SOLID168 that were
not favorable.
New studies show it in a better light.
www.impactengsol.com 8
©2005 – IMPACT Engineering Solutions, Inc.
Old Studies on
SOLID168
www.impactengsol.com 9
©2005 – IMPACT Engineering Solutions, Inc.
New Comparison
of SOLID168 Tet to
SOLID164 Brick
Solid164 Brick
Solid168 Tet
2 meter drop
Lexan 141
Bilinear Isotropic
0.050 wall
Elements Evaluated
ANSYS (185, 187)
DYNA (164, 168)
www.impactengsol.com 11
11
©2005 – IMPACT Engineering Solutions, Inc.
How About for a Real Drop Test?
Post26/Powergraphics Off
SEQV EPTOEQV EPPLEQV
ANSYS LSDYNA % Diff ANSYS LSDYNA % Diff ANSYS LSDYNA % Diff
Shells 39494 30207 23.51 1.1280 0.7833 30.56 1.0270 0.6938 32.44
Bricks 23825 24725 -3.78 0.4129 0.4418 -7.00 0.3520 0.3854 -9.49
Tets 26087 23859 8.54 0.5107 0.4342 14.98 0.4445 0.3528 20.63
% Diff Brick to Tet -9.49 3.50 -23.69 1.72 -26.28 8.46
www.impactengsol.com 13
13
©2005 – IMPACT Engineering Solutions, Inc.
Conclusions of Results
Part too thick to model with shells
Plastic strain (EPPLEQV) artificially high in POST1 for
DYNA. This is due to combined effect of Powergraphics
and EF NU=0
Percent difference between tets and bricks less than 5% for
DYNA. SOLID168 acceptable for use in drop tests.
Tets in ANSYS reported 10-30% higher result values.
Run times were an order of magnitude longer in ANSYS.
Not surprising considering full integration elements being
used
www.impactengsol.com 14
14
©2005 – IMPACT Engineering Solutions, Inc.
Suggestions for Evaluating Results in LS-
DYNA
Determine Max Plastic Strain Location (Node) in Post1.
Set LAYER,FCMAX
In Post26 plot EPPLEQV and EPELEQV for node. Add
these two together to determine true EPTOEQV
Compare this value to your failure criteria.
Look at hourglass energy from GLSTAT file. Rule of thumb
is less than 5%.
Reaction values best found using LSPost
System mass can be calculated from KE and Global
velocity found in GLSTAT.
www.impactengsol.com 15
15
©2005 – IMPACT Engineering Solutions, Inc.
Post1 Results vs. Post26
www.impactengsol.com 16
16
©2005 – IMPACT Engineering Solutions, Inc.
Drop Test Examples