Convergence issue on Linux version of OpenSEES

[ymdarestani]

Hello,

I have been trying to run a pushover analysis of a transmission tower on a Linux based supercomputer. So, I used the source codes of OpenSEES versions 2.5.0, 3.0.0 and 3.0.2 provided in Github and I compiled an executable for OpenSEES in the super computer. However, no matter what version of OpenSEES I install, I get convergence issue in Linux while my code works perfectly fine in a Windows machine with OpenSEES 2.5.0. I was wondering if there is any fundamental difference between the Windows version and the source code? If not, do you have any idea why I get the convergence issue in Linux?

Thanks,

[EricsonEncinaZ]

I have had similar issues with different version of OpenSEES running in both Windows and Linux machines and HPC facilities. So, I may doubt whether it is completely because of the OS or it is because of the different versions and revisions of OpenSEES. For example, the current version for Windows is v2.5.0 rev6536, a while ago it was v2.5.0 rev6477. I have noticed that the same model sometimes fails to converge in one and actually gets to the end in the other. The good news is that after tweaking the models (No of elements, No of IPs, Mesh, integration schemes) when they converge in both versions they do give the same results .