Hello All,
I have written a UMAT code to predict the anisotropic elastoplastic behavior of a material (with damage). I have applied this UMAT to a single quad element and it works as predicted. It also works for some multi-elements models with a very coarse mesh (e.g. 5*5*5), but when I apply it to multiple elements model with a little finer mesh (e.g. 10*10*10) it breaks down from the beginning and even cannot start the elastic phase.
Can anyone help me how to solve this problem? I would appreciate all related comments.
As some researchers suggested, I saved my results (Stress and DDSDDE) in a text file for element=1 and npt=1. This loading is a uniaxial loading in which there is only sigma_33. I am not sure how this text file can help me. As you see it cannot progress beyond Kinc=1.
Thanks,
Faezeh
Dear @Faezeh,
When including damage in a standard analysis the calculation of the ddsdde can be tricky and makes it difficult for abaqus to converge. For 1 element the convergence is more easy. Can you disable the damage part of your code and keep only the elastoplastic (or even only the elastic) part and solve again? If your multi element model is solved then something is wrong with the calculation of the material jacobean when damage is included. Also, it is easier to work initially in 3d and afterwards in 2d because in plane stress one extra iterative procedure needs to be included (zero stress constraint).
- Badges
- Science topic
- Similar topics
- Comment