Successful meshing, but high residual of the MUMPS solver

Hi, I am new to SimScale, but very much like interacting with it so far.

  • I am trying to simulate the pressure and deformation at the center (sample location) of a pressure cell, where the sample is contained in a hole inside a steel disc, which is squeezed between diamonds.

  • After optimizing the meshing parameter, the meshing seems successful, but the simulation run always fails, with the error that the residual of the MUMPS solver is too high.

  • Please find attached the error screenshot, and the link to the project below:
    Pressure cell project

I will very much appreciate some help in pointing me in the right direction. Thank you!

Hey adwivedi712,

I am not sure if you tried the suggestion from the error and refined the mesh. Could you please share the project URL here as well, so we can take a closer look and investigate if everything else is in order!

Best,
Satvik

Hi @shenoy , thank you for getting back. Yes I did try to refine the mesh which solved the meshing errors, but maybe I am not understanding the correct refinement process. The project URL is included in the last bullet point. Thanks!

Hi @shenoy , I have checked again all parameters of the mesh, and they seem to conform to the system requirements. But the simulation run still fails. Is there something I am unable to optimize? I am tempted to increase the ‘Linear system relative residual’, but from what I read, it’s not the best practice, especially since in my case it is 1e+10 larger than the max. allowable value.

Please let me know your opinion. Thanks!

Hi @adwivedi712,

If I may drop my 2 cents , I believe the issue lies in these contacts being sliding and there being only a fixed support as a constraint on the whole assembly:

Remember that sliding contacts will alow for relative motion between parts. Since you have a static analysis, the presence of rigid body motion will make it unstable and ultimately fail.

Some possible solutions in this case:

  1. Change all of the contacts to be bonded instead
  2. Set some sort of elastic support to stabilize the simulation
  3. Run a dynamic study instead, since it allows for rigid body motion

Cheers
Igor

Hi Igor @igaviano ,

Thank you very much for getting back! Your suggestion to change the contacts from sliding to bonded made the simulation work very nicely :smiley: .

Thanks again and best regards,
Anand