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 The information provided may be out of date.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.

High pressure around rotating domain boundary

Please login with a confirmed email address before reporting spam

Hi, and thanks for reading.

I am modelling an impeller in a turbulent vessel (2d for now). I have an ALE moving mesh rotating with the impeller. When I compute the model the areas of highest pressure and velocity are at the boundary between the moving and the fixed mesh - regardless how large I set the rotating domain, this pattern always follows it.

Has anyone got any advice on what I am doing wrong? The file is 180MB, but if it would help I can attach it.

Many thanks,
Andrew

8 Replies Last Post Apr 30, 2012, 3:52 a.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 13, 2012, 8:13 a.m. EDT
Hi,

As always it'd better to have a model to explore. If you feel like uploading it you may delete the solvers to reduce the size or try some uploading web-based software.

Cheers
Hi, As always it'd better to have a model to explore. If you feel like uploading it you may delete the solvers to reduce the size or try some uploading web-based software. Cheers

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 23, 2012, 10:21 a.m. EDT
Hi,

Sorry for the delay, I have attached the model to this post now.

Thanks for the help,
Andrew
Hi, Sorry for the delay, I have attached the model to this post now. Thanks for the help, Andrew


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 27, 2012, 4:22 p.m. EDT
Hi,

I am not sure what you are trying to do; you do not need to use the ale physics in this simple case, You have all the elements you need: you only need to take care in defining the time step, it must be much less than 0,1s because the rotating speed is more than 8 revoluton per second, otherwise you won't see anything moving. Furthermore, you'd better to use a high relative tolerance.

Please find attached an example dervied from your model as well as a movie of the velocity, please note that the movie is ten time slower than the results of the actual simulations. I have deleted the solver of the model, it was far too big, you may re-run it.

Cheers
Hi, I am not sure what you are trying to do; you do not need to use the ale physics in this simple case, You have all the elements you need: you only need to take care in defining the time step, it must be much less than 0,1s because the rotating speed is more than 8 revoluton per second, otherwise you won't see anything moving. Furthermore, you'd better to use a high relative tolerance. Please find attached an example dervied from your model as well as a movie of the velocity, please note that the movie is ten time slower than the results of the actual simulations. I have deleted the solver of the model, it was far too big, you may re-run it. Cheers


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 27, 2012, 6:12 p.m. EDT
Thank you very much for the reply and time you put in, however comsol crashes when I try to open your model. Extracting the mph archive it appears you are using 4.2a, I only have access to 4.2.

Let me get it clear what you did:

Remove the ALE moving mesh
Reduce the time step to much less than 0.1
Raise the relative tolerence

Run the model?

Thanks
Andrew
Thank you very much for the reply and time you put in, however comsol crashes when I try to open your model. Extracting the mph archive it appears you are using 4.2a, I only have access to 4.2. Let me get it clear what you did: Remove the ALE moving mesh Reduce the time step to much less than 0.1 Raise the relative tolerence Run the model? Thanks Andrew

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 28, 2012, 3:25 a.m. EDT
Hi,

Yes, I did use 4.2a; you may try to remove the ALE, set a time step at 0.01 and use a relative tolerance of 0.2 or 0.1, version 4.2 was a bit more troublesome than 4.2a, I am not sure what relative tolerance may be required.

Cheers
Hi, Yes, I did use 4.2a; you may try to remove the ALE, set a time step at 0.01 and use a relative tolerance of 0.2 or 0.1, version 4.2 was a bit more troublesome than 4.2a, I am not sure what relative tolerance may be required. Cheers

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 28, 2012, 6:01 a.m. EDT
May I ask, what convergence were you getting when you solved this problem?

I left the model running overnight, but it is still computing, with convergence in the order of 10^5 or 10^6.

Thanks again,
Andrew
May I ask, what convergence were you getting when you solved this problem? I left the model running overnight, but it is still computing, with convergence in the order of 10^5 or 10^6. Thanks again, Andrew

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 28, 2012, 4:00 p.m. EDT
Hi,

Since you are using 4.2 I don't think it is possible to compare straightforwardly the two convergence graphs. At any rate, the convergence you get seems to be all right: comsol uses an explicit method in this case, which is extremely time consuming and depends strongly on the characteristics of your pc and the overall time you set as a target.

Cheers

Hi, Since you are using 4.2 I don't think it is possible to compare straightforwardly the two convergence graphs. At any rate, the convergence you get seems to be all right: comsol uses an explicit method in this case, which is extremely time consuming and depends strongly on the characteristics of your pc and the overall time you set as a target. Cheers

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago Apr 30, 2012, 3:52 a.m. EDT
True, I didnt think the solvers would be so different from 4.2 to 4.2a. I managed to decrease the convergence by simply tweaking the solver some. Its now acceptable for my machine - 10^1 to 10^2.

Thank you ever so much for your guidance.

Andrew
True, I didnt think the solvers would be so different from 4.2 to 4.2a. I managed to decrease the convergence by simply tweaking the solver some. Its now acceptable for my machine - 10^1 to 10^2. Thank you ever so much for your guidance. Andrew

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.