KKS+Tensor Mechanics ; Solution not converging #16712
Replies: 5 comments 10 replies
-
Dear MOOSE team, I am trying to simulate the equilibrium shape of a precipitate in a infinite matrix (stress-free boundaries). I systematically built up the input scripts as suggested in the MOOSE workshop. But after I added tensor mechanics in the input script, my solution is not converging on my supercomputer. But it is converging in my laptop where I have conda-based MOOSE installation. Step:1) single precipitate without elasticity Step:2) single precipitate with elasticity added. But the elastic energy was not coupled with Allen-Cahn equation. Step:3) Fully coupled Allen-Cahn + Tensor mechanics. Regards, |
Beta Was this translation helpful? Give feedback.
-
Framework Information on my supercomputer:
|
Beta Was this translation helpful? Give feedback.
-
Some solvers' behavior might depend on the number of MPI processors, so try to modify the solver option. petsc_options_iname = '-pc_type -sub_pc_type -pc_asm_overlap -sub_pc_factor_shift_type' I believe @fdkong will give you better suggestions here. |
Beta Was this translation helpful? Give feedback.
-
Thank you Wen. But this change does not fix the problem. My solution still does not converge.
-Kamal
…________________________________
From: Wen Jiang <[email protected]>
Sent: Sunday, January 17, 2021 12:03 PM
To: idaholab/moose <[email protected]>
Cc: Kadirvel, Kamal <[email protected]>; Author <[email protected]>
Subject: Re: [idaholab/moose] KKS+Tensor Mechanics ; Solution not converging (#16712)
Some solvers' behavior might depend on the number of MPI processors, so try to modify the solver option.
petsc_options_iname = '-pc_type -sub_pc_type -pc_asm_overlap -sub_pc_factor_shift_type'
petsc_options_value = 'asm lu 2 nonzero'
I believe @fdkong<https://urldefense.com/v3/__https://github.com/fdkong__;!!KGKeukY!nsz8JpM-Sw4pRifvKcs0gggbQKt5jDqgYTsozSjrcTCsSQVjtCEb8BEzp06WQjfFigmQl_Pbh9w$> will give you better suggestions here.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/idaholab/moose/discussions/16712*discussioncomment-289059__;Iw!!KGKeukY!nsz8JpM-Sw4pRifvKcs0gggbQKt5jDqgYTsozSjrcTCsSQVjtCEb8BEzp06WQjfFigmQtYGBc5I$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AIUJ4LPE7OAFN4DIYU6O7NLS2MJ63ANCNFSM4WEUDQBA__;!!KGKeukY!nsz8JpM-Sw4pRifvKcs0gggbQKt5jDqgYTsozSjrcTCsSQVjtCEb8BEzp06WQjfFigmQzevlwps$>.
|
Beta Was this translation helpful? Give feedback.
-
Hi Wen,
I am using the same solver ( by which I mean same version of PetSc in both cases). But in my laptop, the solution converges. In HPC, the solution does not converge. This is the case for Step2 and Step3 scripts. My Step1 script converges in both my laptop and in HPC
…-Kamal
________________________________
From: Wen Jiang <[email protected]>
Sent: Tuesday, January 19, 2021 7:05 PM
To: idaholab/moose <[email protected]>
Cc: Kadirvel, Kamal <[email protected]>; Author <[email protected]>
Subject: Re: [idaholab/moose] KKS+Tensor Mechanics ; Solution not converging (#16712)
Sorry I did not get a chance to run your problem. Just to be clear: with the old solver setup, it converges on your laptop but not on hpc, is this correct? And for the new solver, it does not converge on neither your laptop nor the hpc.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/idaholab/moose/discussions/16712*discussioncomment-294763__;Iw!!KGKeukY!k60CjHa2RPwP_oPobpMRG5xKRigFIyg4Np5SGZUwhxIGjQoLfFchk6nj_46F66GCviEOm3bA6uA$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AIUJ4LLJC43ZAAL6HXZ2C7DS2YM3ZANCNFSM4WEUDQBA__;!!KGKeukY!k60CjHa2RPwP_oPobpMRG5xKRigFIyg4Np5SGZUwhxIGjQoLfFchk6nj_46F66GCviEO4HMOxzI$>.
|
Beta Was this translation helpful? Give feedback.
-
Accidently Posted. Let me add my question in the response.
Beta Was this translation helpful? Give feedback.
All reactions