Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

PCM

Please login with a confirmed email address before reporting spam

I want to model the heat transfer in a PCM. But the problem is that every time I run the model (with the exact same settings) I get different results..
How is this possible and how do I know what the right solution is?


3 Replies Last Post 25.09.2015, 10:47 GMT-4
Frank van Gool COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 25.09.2015, 09:21 GMT-4
Dear Astrid,

In your model, you can expect a large change in gradient of the solution in the x direction, and none in the y direction. If there is a gradient anyway, that is typically caused by a suboptimal mesh.

Attached your model where the mesh is changed to be more accurate, and also the relative tolerance is decreased in the time dependent step. In your case, make sure that the mesh is fine enough to capture the phase transition.

To check whether your model is correct, always do a mesh convergence study:
www.comsol.com/blogs/how-to-implement-mesh-refinement-study/
www.comsol.com/blogs/meshing-considerations-linear-static-problems/
And ideally also a tolerance convergence, where you decrease the tolerances in your model, and check the effect.

Best regards,
Frank

ps. I notice that the phase change interval is only 0.1K, try out setting this to a larger value. You might also want to limit the timesteps manually, and see if this has a large effect.
Dear Astrid, In your model, you can expect a large change in gradient of the solution in the x direction, and none in the y direction. If there is a gradient anyway, that is typically caused by a suboptimal mesh. Attached your model where the mesh is changed to be more accurate, and also the relative tolerance is decreased in the time dependent step. In your case, make sure that the mesh is fine enough to capture the phase transition. To check whether your model is correct, always do a mesh convergence study: https://www.comsol.com/blogs/how-to-implement-mesh-refinement-study/ https://www.comsol.com/blogs/meshing-considerations-linear-static-problems/ And ideally also a tolerance convergence, where you decrease the tolerances in your model, and check the effect. Best regards, Frank ps. I notice that the phase change interval is only 0.1K, try out setting this to a larger value. You might also want to limit the timesteps manually, and see if this has a large effect.


Please login with a confirmed email address before reporting spam

Posted: 9 years ago 25.09.2015, 09:53 GMT-4
Thanks a lot for your response.

I already tried to change the relative tolerance and to make the mesh smaller but still I didn't have the same results for different runs...

Unfortunately I can not open the model you send my because I only have the 5.0 version and PCM_fix.mph is made in a newer version.

Astrid
Thanks a lot for your response. I already tried to change the relative tolerance and to make the mesh smaller but still I didn't have the same results for different runs... Unfortunately I can not open the model you send my because I only have the 5.0 version and PCM_fix.mph is made in a newer version. Astrid

Frank van Gool COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 9 years ago 25.09.2015, 10:47 GMT-4
Dear Astrid,

Here the version in 5.0, where the following is changed:
- transition interval 1K
- fine mesh in x direction
- intermediate stepsize, which limits the timestep to 10s

Best regards,
Frank
Dear Astrid, Here the version in 5.0, where the following is changed: - transition interval 1K - fine mesh in x direction - intermediate stepsize, which limits the timestep to 10s Best regards, Frank

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.