Convergence issues in a problem with time derivative terms in coefficient form PDE module

Please login with a confirmed email address before reporting spam

About the simulation: I am trying to study the time evolution of fractional polarization in my geometry when they satisfy certain boundary conditions up to 500us of evolution.

Simulation setup: I am using the coefficient form PDE module and performing a time dependent study on my geometry. The coefficient form that i am using is a basic diffusion equation with a relaxation term. There are two boundary conditions here, one of them is a normal zero flux condition and the other one is another PDE that contains a time derivative of my dependent variable "u". I applied this constraint using the constraint node. Since in this module standard constraints are not allowed to have time derivatives of "u", I turned on the weak constraints check and applied the reaction term on individual variables under constraint settings.

The issue: In this simulation my current goal is to plot the volume integration of "u"/(some constant) for different "sigma_s" parameters. I call it magnetization and its under the desired values tab. I want to keep the value of the remaining parameters as it is under the parameters tab. Now when I set my sigma_s to 1e20, the simulation seems to work fine and convergence plots look fine and the results make sense as in it outputs what I would expect. But when I change the sigma_s to 1e17 or lower then the simulation, although completes running, has issues in its convergence. Its convergence doesn't go lower than 1e5 or something. The results generated by these set of parameters for sigma_s of 1e17 also doesn't make sense and is unphysical.

What I have tried so far: 1. I tried reducing my mesh size, still convergence plot value too high and unphysical results. 2. Reducing tolerance under time dependent solver from 0.01 to 0.00001, no change. 3. Changing solver types, running into errors. 4. Changed initial time steps, no change (personally, I don't think the 1st step is a problem since when i run the simulation for 50us instead of 500us, it does work fine) 5. Reducing maximum step constraint up until 1e-8 s, no change 6. under fully coupled tab i changed the non linear method to Automatic and also to Automatic highly nonlinear, no change. 7. Increased maximum number of iterations to 25, no change. 8. Reduced tolerance factor to 0.0001 from 1, no change.

I tried some of these things together for example i tried 6., 7., 8. together but no change.

What I am hoping from the simulation: As a check if you change sigma_s under Parameters tab to 1e20, you get a decay profile for "magnetization". I expect very similar behavior when I change the sigma_s to 1e17 or 1e16. The time scale of the decay might go up as in it could decay slower so i might need to increase my output times from 500 us to something higher like 2-3ms. But at least for 1e17, i think a time dependent study till 500us should suffice.

It would be great if I can understand why the simulation doesn't seem to converge and what can be done to fix this. Thank you!



0 Replies Last Post 2023年12月26日 GMT-5 12:32
COMSOL Moderator

Hello Pritam Priyadarsi

Your Discussion has gone 30 days without a reply. If you still need help with COMSOL and have an on-subscription license, please visit our Support Center for help.

If you do not hold an on-subscription license, you may find an answer in another Discussion or in the Knowledge Base.

Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.