Slow or even non-converging analysis problem

[Charon]

Dear STKO Team，

In this example, my model starts out fast when performing a dynamics analysis, but becomes particularly slow once it reaches 25% (roughly) progress, and even stops running after it reaches more than 30%. Can you help me to check if this is a problem with the material or what? I would appreciate it.

Best wishes，
Charon


Update：After almost 19 hours，it reaches 54％，but running is very very slow.

[STKO Team]

There were some issues with your model:

• You were adding the structure from the beginning, while it should be added later on
• The initialStage Off should be called after the plastic gravity (before adding the structure)
• The load of the structure is added with a wrong time Series: The Ramp reaches lambda=1 at time t=1.0e6, but your analysis only lasts 1 second

Here is a modified version of your model:

20230404_0.3g_Test.scd.zip

(462.42 KiB) Downloaded 142 times

It works fine if you don't update it to plastic (so you can check if the workflow is correct)
If you activate the plastic behavior you will see abnormal displacements around the holes. I suggest you to double-check your material parameters

[Charon]

Thanks for your quick reply! It really helped me a lot!
One more question. For the number of meshes I have for this model, how many cores do you think would be used for maximum efficiency when computing in parallel? My workstation has 54 cores, and I've been running with 48 cores before, and I'm not sure if that's the fastest I can get.

[STKO Team]

This really depends on your machine and the problem type. There is not unique answer. Few thousands per core should be fine.

You can try to run a simple elastic analysis with just 1 step and measure the elapsed time.
Do it with increasing number of cores and you will see what is your limit