NewMaterial - Numerical issues

For developers writing C++, Fortran, Java, code who have questions or comments to make.

Moderators: silvia, selimgunay, Moderators

Post Reply
stav
Posts: 1
Joined: Mon Sep 10, 2012 12:40 am
Location: Politechnical School, Department of Civil Engineering

NewMaterial - Numerical issues

Post by stav »

Hello everyone!!!

I'm very new in programming, so I would appreciate some help...

I'm trying to build a new Uniaxial Material, and I'm currently testing its accuracy. I'm using the new material to define the force - deformation behavior of a ZeroLength that connects an entirely fixed node to another that has only one free translational DOF. I impose different displacement histories on the free DOF, using DisplacementControl integrator and either KrylovNewton or Newton algorithm.

In general, I get the expected results. However in some cases numerical instabilities appear, which seem to be sensitive on the value of the adopted displacement increment... Yet, I cannot really understand the source of the problem and how to resolve it.

Some specifications:

1. For a specific displacement history, the material might work fine for "smaller" (eg. Dstep = 0.00001) or "larger" (eg. Dstep=0.0001) displacement increment and yet become unstable for a "medium" value (eg. 0.00002)

2. The analysis does not fail, but the results are inconsistent. I suddenly get very large values of displacement at the free DOF, totally irrelevant of the displ. increment imposed. (By the way, in this case I would expect to get an error message, no?)

3. The problem basically appears when the material, unloading from a compressive cycle in relatively large strain values, enters a zero-stress state (where both stress and tangent are set to 1e-20)...

Any ideas???

Thank you in advance!
Post Reply