Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
Convergence problem in a simple geometry with Joule heating module
Posted May 8, 2014, 1:40 a.m. EDT Studies & Solvers 2 Replies
Please login with a confirmed email address before reporting spam
Hi guys,
I am modeling a simple geometry (some stacked layers) with um and nm dimensions in comsol 4.3b.
The convergence plot goes down rapidly at first and after some iterations it seems that the software changes the value of parameters. Then the convergence plot goes up and the cycle continues. I have many upward and downward trends in my convergence plot and can not get fully converged solution. The plot is attached as well. Could anyone suggest me what should I do? Should we play with different solution methods, which I did and couldn't get satisfactory results? I appreciate for any help.
Thank you.
I am modeling a simple geometry (some stacked layers) with um and nm dimensions in comsol 4.3b.
The convergence plot goes down rapidly at first and after some iterations it seems that the software changes the value of parameters. Then the convergence plot goes up and the cycle continues. I have many upward and downward trends in my convergence plot and can not get fully converged solution. The plot is attached as well. Could anyone suggest me what should I do? Should we play with different solution methods, which I did and couldn't get satisfactory results? I appreciate for any help.
Thank you.
Attachments:
2 Replies Last Post May 8, 2014, 9:19 a.m. EDT
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hello Mohsen,
As your memory usage is already on the high side I assume that a direct solver is out of the question.
But you can try to solve the model on a much coarser mesh, and use that as the initial guess for your fine mesh.
This can be done manually in separate steps, or you can set up a multigrid iterative solver with your own multigrid levels. This is often a very efficient way of dealing with slowly converging models.
As you did not include the model, it is hard to see what the best solution would be here.
Best regards,
Frank
As your memory usage is already on the high side I assume that a direct solver is out of the question.
But you can try to solve the model on a much coarser mesh, and use that as the initial guess for your fine mesh.
This can be done manually in separate steps, or you can set up a multigrid iterative solver with your own multigrid levels. This is often a very efficient way of dealing with slowly converging models.
As you did not include the model, it is hard to see what the best solution would be here.
Best regards,
Frank
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Dear Frank,
Thank you a lot. I will do the procedures suggested by you and see what happens. I will get back to you ASAP. Thanks again.
Mohsen,
Thank you a lot. I will do the procedures suggested by you and see what happens. I will get back to you ASAP. Thanks again.
Mohsen,