I get very different calculation times for the same calculation. (1min - 20min)
Does anyone have any ideas as to why this could be?
Do you mean resubmitting the same input file ? Does the incrementation/convergence (as reported in log files) differ ?
Yes, I resubmited the same input file!
Are you working on a laptop?. If you are unplugged the battery management could be saving power by suppressing some capabilities.
Other option is that you didn’t kill properly some analysis and it’s running in the background.
Thank you Disla,
I don’t work on a desktop.
And I think there is nothing working in the backround.
It remains mysterious.
What about the differences in incrementation (if it’s nonlinear) ? Which equation solver are you using ?
“What about the differences in incrementation (if it’s nonlinear) ?”
I don’t know. (Sorry my knowledge is limited.)
I use “CONTACT PAIR”. So I think the Model is nonlinear.
The equiation solver is PastiX.
Since I am not such an expert, my model is probably not very good. But the calculation time should always be the same. At least almost the same, right?
How are you submitting the job to calculix?
Are you using a preprocessor for this or do you run the job directly with the calculix executable?
I use a cmd-File. Made by myself.
Looks like it is not related to the pipe reading of the process.
Can you share your .inp file?
I would like to do this but the file contains 30000 lines. Too much for one picture. Is there a way to send the file?
The model includes 6 concrete slabs on a base with 2 wheel loads and temperature. I have done similar calculations many times.
You can upload it to some hosting website and share the link here.
Many thanks to everyone who replied!
I have now found the error. I had positioned my load blocks 10mm above the plate surface and thought that gravity would pull them down. This was the case, but it obviously disturbed the convergence of the method.
I understand that poor modeling leads to poor calculation times, but the fact that the calculation times are always different remains mysterious to me. Incidentally, the result was always correct, only the calculation times varied greatly.
Have a nice weekend!
Greetings
Axel
If calculation becomes unstable and depends on round-off errors instead of algorithm error the solution is no longer systematic and the calculation times may vary.
1 Like