The subroutine UEXPAN is called twice. Can I update SDVs only in one of the two calls?

[Shufen]

Dear researchers,
I am using the user subroutine UEXPAN in ABAQUS. I want to update a specific set of state variables in the subroutine. The manual of UEXPAN says that ABAQUS calls this subroutine twice-the first based on the values of previous increments and the second based on the values returned from the first increment. In the subroutine, I am calculating some SVDs and updating to after the increment. However, the SDVs are getting calculated twice and getting updated twice. My question is this: Can I update the SDVs to the values for the second increment only? Abaqus mentions that it can be done. Please provide me your valuable suggestion.

Regards
Shufen

 

[Michael Koopman]

Unless the code has changed dramatically, which is possible, the Eulerian strains in the UMAT routines do not update the Langrangian vectors in the time step forward. They do address the virtual work functions and the first pass is where you need to introduce any virtual work energy change. You may not directly enter a Hill's anisotropic plasticity with this routine and the best choice may be to use a percentage of update work in each pass with the first pass likely less than fourty percent of the instantaneous virtual work represented by the Eulerian strains of the element. The residual energy change should be included in the second pass. The default steel defined by Sorenson is not a Greene's Almansi strain domain and the 200% work kinetic may work with the death and remapping as an intensified value limit but a gradient mechanic seems a more precise approach.
Good luck, hope you work out your problem and this weighted two pass strain idea is helpful. You may want to consider a variable weighting between passes that even goes to 0.999 on the first pass to match the 200% energetics of the isotropic steel basis model. There appears to be a predefined UMAT variable named to imply application in this weighting concept. Heads up here, however, as negative weighting on the second pass throws an error and you will lose your output.
It has been several years since working the Abaqus source code for this issue but I expect less has changed then has remained constant in the interim.