Abaqus vs LS dyna

[dogbural82]

Hi all,

Im currently comparing the max strain values using different codes: Abaqus and LSdyna.

It was found that while both codes give pretty close values of force and displacement etc to each other, LSdyna derives much much higher than Abaqus for max strain as seen below:
LSdyna: 0.0240226
Abaqus: 0.00861379

So im wondering that there is a big difference in terms of max strain while other varialbes are close enough.

Much appreciated if you can give me some pieces of advice on this.


Also, in Abaqus, is there any way to check the estimated cpu (or clock) time to complete simulation while running simulation as LSdyna does?


Regards,
Minki

 

[dogbural82]

Further to this,

Im simply impacting the rigid ball to a composite panel, both ends (or tips) constrained.

I'm using explicit method for both codes; but they still give me the different results.

In LSDYNA, lower surface was picked; in abaqus, SP1 (section point 1??) was picked.

1) element formulation
For lsdyna: ELFORM=2
For abaqus: S4R

2) number of intergration points
For LSdyna: 16 , i.e. one point per ply
For Abaqus: to be honest, im not sure how many. It is veiwed that when one element is picked, SP1 and SP 48 were shown in ODB field output using Abaqus/Cae for post processing. According to this, i just guessed that it might have 48 integration points, which does not look right. Please kindly correct me if im wrong to check number of integration points.

3) Material model for LSdyna - *MAT_ENHANCED_COMPOSITE_DAMAGE

4)Contact
For Lsdyna - *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE, penalty constraint
For Abaqus - *CONTACT PAIR, mechanical constraint=penalty, based on surface

Im also wondering if Laminate Shell Theory in LSdyna does make its strain result different to that using Abaqus.

Thank you for your help in advance.

Regards,
Minki

 

[Drej]

Interesting. Section pt. 1 doesn't necessarily coincide with a "lower surface" in DYNA. Try a modal analysis of both systems prior to the impact case. That will give you the (linear) dynamics of each model - this is a good starting point.

All other things being equal, it probably comes down to a mixture of the material model, the element formulation and the solver architecture/defaults/etc.

What are the time steps for each code by the way?
Is the same variable of strain being compared at the same location?

I would also try a very simple test model and compare values at known locations.


------------
See faq569-1083 for details on how to make best use of Eng-Tips.com