Clear all

using smaller time step in vumat  

Posts: 1
New Member
Joined: 12 years ago

Hello all,

I have developed a vumat routine which normally works fine. Its an elastoplastic model that at each step checks to see if stress state is located within the yield surface or not. If its not the routine returns the stress state to the yield surface according to normality. Obviously for the routine to be able to return the stress state back to the yield surface the time increment should be small enough so that the change in stress increment wouldnt be large. In abaqus explicit the program decides about the time step and chooses small time increment to satisfy the stability. However I tested my routine and I realized when the deflection is high at some point the stress increment will be large and thereafter the routine cannot return the stress state back to the yield surface. I thought if I was able to use a time increment less than what abaqus decides then I can solve this problem. Is there any way in abaqus explicit to decrease the time increment of steps? Has anyone ever faced such problem in their routine? I really appreciate any comment or suggestion. And by the way I use both ALE adaprive meshing and nonlinear geometry.


Topic Tags
2 Replies
Posts: 3881
Joined: 12 months ago

I dont have all the details of your material model. But in my experience, most plasticity model implementations are rather stable with respect to large increments. Perhaps you have a new an different model that you are working with.

About increment size. Yes, you can specify the max allowed time increment size in Abaqus/Explicit. In the inp-file, the line after *Dynamic, explict specifies both the total time and the max allowed increments size. See the manuals for details of how to specify these values.


Topic Tags
2 Replies
Posts: 59
Trusted Member
Joined: 11 years ago

The problems which are solved by dynamic explicit solvers almost always supply

sufficiently small time increments due to the stability limitations.

If your return mapping involves correct linearisation you would not expect any problems. I suggest you to chect number of iterations and the magnitude of your residuals at load steps. This way you may have an idea of the quality of your code. In ABAQUS Explicit with automatic time stepping with a hardening material model my codes supply the required tolerance in at most 3 iterations at a GP with a tolerance of 1e-6.