Page 1 of 1

Problems with basal sliding when applying Porous Solver

Posted: 10 Nov 2020, 19:21
by glebcher
Dear Elmer users,

I am trying to introduce basal sliding into a thermo-mechanically coupled model for a polythermal mountain glacier. I use snow/firn rheology (Porous Solver). If I apply no-slip conditions at the bedrock, the case works. But when I set the basal boundary condition as described in Elmer/Ice Wiki (http://elmerfem.org/elmerice/wiki/doku. ... ers:porous), no convergence is seen even after 10 000 linear system solver iterations. The error message is:

Code: Select all

      9990 0.9374E+00
     10000 0.9391E+00
     10001 0.9391E+00
NUMERICAL ERROR:: IterSolve: Too many iterations was needed.
Note: The following floating-point exceptions are signalling: IEEE_DENORMAL
The sif and mesh files are attached.
porous_thermomech_slide.sif
(5.43 KiB) Downloaded 104 times
Data.rar
(346.86 KiB) Downloaded 95 times
Also, it may worth mentioning that the solver is restarted from the previous solution (unfortunately, the restart file is too large to be attached). I will also provide additional information, if needed. I do not know what is incorrect and how to obtain convergence. So, I would be very grateful for any help or advice.

Regards,
Gleb

Re: Problems with basal sliding when applying Porous Solver

Posted: 10 Nov 2020, 19:41
by fgillet
Hello;
if the iterative solver is not converging you can try to use a direct solver (umfpack in serial or Mumps in parallel) this should always work (at least if you have enough memory).

If the direct method is too slow you may still try to use another iterative solver and/or preconditionner.

Re: Problems with basal sliding when applying Porous Solver

Posted: 12 Nov 2020, 18:19
by glebcher
Hello, fgillet!

Thank you very much for your reply and recommendations. I will try to do everything that you suggested. Unfortunately, I have not managed to attach any files, although not all of them were too large. I will try to fix it somehow.

Regards,
Gleb

Re: Problems with basal sliding when applying Porous Solver

Posted: 23 Nov 2020, 18:30
by glebcher
An update to the post: the sif and mesh files are now attached and available.

Regards,
Gleb

Re: Problems with basal sliding when applying Porous Solver

Posted: 10 Apr 2021, 17:38
by glebcher
Dear Elmer users,

the problem with convergence has been finally solved by applying the GCR method with the restart parameter equal to 300 and the ILU0 preconditioner in the Porous Solver section. Thanks fgillet for the useful recommendations!

Regards,
Gleb