Sth new for Multi-element test VUMAT

[yaning]

Hello Dr. Bergstrom,

I found why the multi-element test quits. The codes terminates when the F matrix (defgradNEW) of one element becomes [1 0 0, 0 1 0, 0 0 1] which means no deformation at all. Then stress of this element becomes zero and singularity occurs.

How could that happen: during simple tension, the deformation of one of the element is identity? When I introduce a artificial perturbation (1d-8) to F matrix to avoid the sigularity, then it can be completed, but the yield point was jumped. BTW, I am modeling Argon model for PC. However the data for one-eleemt and four-elemet test are very good, since no elements is assigned to be zero deformation during the calculation by ABAQUS.

Any suggestions for this kind of problem?

Thanks.
Best,
Yaning

[Jorgen]

That sounds very strange! The stress should be zero if F=1, you should not get a "singularity".

- Jorgen

[yaning]

Dr. Bergstrom,

Yes, you are right. But if the stress is zero, the driving stress (T*) is zero too (T*=T-TB). And then sigularity occurs when calculating N= T*'/(Norm T*) (N is the normalized tensor aligned with the deviatoric driving stress (T*) (this is for Argon Model)

I introduced a small numerical perturbation to F or TB to avoid this singularity. Sometimes it works, some times doesn't. The result is very sensitive to the perturbation introduced.

Acctually, what I can't understand is why there is such an element with exactly zero deformation during the calculation?


Thanks.
Best,
Yaning

[Jorgen]

I see, well, mathematically as you take the limit as the stress goes to zero, the velocity gradient should also go to zero. Hence, when the stress is zero, it does not matter what the tensor N is. I therefore recommend that you check to see if the stress is equal to zero, and if it is, then you can simply set N=I.

- Jorgen

[yaning]

Dr. Bergstrom

It is a good idea. I get it.

Thanks a lot!


Best,
Yaning

[Ramesh kadam]

sir i m doing geometrically nonlinear static analysis of piezolaminated composite beam with 6 layers out of which one layer is piezomaterial layer in ABAQUS.i have modeled beam as 3-D planer shell.Beam is simply supported at both ends. i used linear SHELL element, but simulation existed with an error ....error msg is solution appears to be diverging...So many attempts made for this incriment. warning is....one eign value negative.

so please sir give me some tips.

thank you very much

Ramesh
IIT delhi
(India)

[Jorgen]

That error message typically occurs when you don't have enough boundary conditions specified. That is, at least one degree of freedom for the beam is not constrained which causes a numerical problem.

- Jorgen