Using different solvers

[ACinteriorstress]

Hi group,

Question:

Is it possible that when running a FEMAP model with different solvers, say MSC NASTRAN vs NX NASTRAN, one can obtain slightly different results?

Thank you,

A

 

[ThomasH]

Hi

I would say that it depends on what you are simulating more exactly. And what you mean by "slightly different" .

Since you mention MSC vs NX, NX is at least partially based on MSC. Or more exact, NX is based on MSC.Nastran v 2001 (I think it was). And since that date they have been developed by different teams. But certain solutions in NX are based on Adina (SOL 601 and 701) while MSC uses Marc and LS-Dyna for implicit nonlinear and explicit nonlinear.

So, if you example is linear static or modes or linear buckling I would not expect any difference at all. But for an nonlinear solution there might be a larger difference since there are different solvers inside.

But since there is only one correct solution, for the same mesh I would not expect any significant difference provided all the parameters are the same. The solution time may differ but not the solution.

Best regards

Thomas

 

[ACinteriorstress]

Thank you Thomas,

More specifically, the differences I'm seeing are between Autodesk NASTRAN and MSC NASTRAN, and the analyses I'm comparing are Linear/Static.

I'm hearing different opinions on this. Personnaly I wouldn't expect a differences, but I'm still learning.

Thanks for your input.

A

 

[ThomasH]

Hi

Autodesk Nastran and MSC Nastran have different backgrounds. So there may be some differences in the element formulations.

I have used them both (currently I use Autodesk) but I have never seen any significant differences. All depending on what you mean by "significant" of course.

How much does the solutions differ and in what respect (stress/force/displacement/etc)?

Thomas

 

[ACinteriorstress]

Hi Thomas,

I'm extracting interface loads (Forces) from this model and I'm getting than 10% difference on some of my main loads.

A

 

[ThomasH]

Hi

Ok, 10% sounds like a lot to me. I don't know what i can be but I would have expected < 1%.

I would contact Autodesk support and see if they can help you regarding what differs. Therew could be something with the treatment of interface loads that differes between the solvers.

A first test might be to cvompare displacements and stresses/forces.

Thomas