I am a PhD student working with Abaqus to perform some numerical studies of Mass Timber connections. I am using a UMAT subroutine and was wondering if there is someone who could help me to identify the origin of one issue that I am having.

I am running the same input file on a Windows-based desktop PC and on a Linux-based server. The problem is that, for some reason, on Windows, the analysis converges to 100% of the expected displacement, while on Linux, it reaches about 10% to 12% of the total displacement. The key point is that the difference between both cases starts when a different sub-step is given in Linux, and then the analysis becomes different in both environments.

The subroutine has been verified several times in previous research, and I tested it with other models, and it works without any problem. The main question would be if there are any specific requirements for the intel (Fortran compiler) version on Windows and Linux for Abaqus 2022. In addition, is there any additional configuration that could be modified to avoid those differences in the sub-steps from Abaqus in Linux?

The point is that if there is a problem with the UMAT subroutine, at least the result should be the same in both Windows and Linux but they differ.

Thanks in advance for your help with any advice or recommendations on what to do.

​​​​​​​