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.

How to handle the error of "Failed to find a solution"?

Please login with a confirmed email address before reporting spam

Hi, everyone

I am in a trouble when solving several set of PDE equations. I has buit the model carfully. The trouble take places when solve it. It appears "Failed to find a solution,No convergence, even when using the minimum damping factor. Returned solution has not converged." Then how to handle it ?

I has tried to deal with it. After I append the "weak form', the iteration number seems increases a little and the results seems get logical a little. Whether the appending "weak form' is a alternative way?

I has tried to adopt the solver of "PARDISO" instead of "MUMPS", then the error occur as before.

Then how to handle it ? Could anybody guide me a way ?

Thank you very much!

youf.

2 Replies Last Post 10 ott 2010, 22:47 GMT-4
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 ott 2010, 06:14 GMT-4
Hi
there are a quite few threads about solver issues and conversion (try a search), this is not black magic, but something closer to art (just like meshing) based on a solid understanding of how the different solvers works (personally I do not claim I understand the solvers, as the limited error messages does not always tell me precisely enough what is the issue).

Some ideas: once you have re-checked that all materials and BC are coherent, that your inital conditions are really good (default "0" is far from always the best but this is so model dependent) Then you can tweak the solver settings such as force "non-linear" instead of having Comsol decide in automatic, increase the relative error. Perform a stationary study first and store the solution to restart from that, change your parametric sweep parameters, save intermediate steps to try to understand how the solution evolves ...

Hope it helps on the way, but know there is no "one" way, there are to many solvers, to many models and approaches to cover them all like that

--
Good luck
Ivar
Hi there are a quite few threads about solver issues and conversion (try a search), this is not black magic, but something closer to art (just like meshing) based on a solid understanding of how the different solvers works (personally I do not claim I understand the solvers, as the limited error messages does not always tell me precisely enough what is the issue). Some ideas: once you have re-checked that all materials and BC are coherent, that your inital conditions are really good (default "0" is far from always the best but this is so model dependent) Then you can tweak the solver settings such as force "non-linear" instead of having Comsol decide in automatic, increase the relative error. Perform a stationary study first and store the solution to restart from that, change your parametric sweep parameters, save intermediate steps to try to understand how the solution evolves ... Hope it helps on the way, but know there is no "one" way, there are to many solvers, to many models and approaches to cover them all like that -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 10 ott 2010, 22:47 GMT-4
Hi,Ivar

Thanks for your help very much!

I will try the ways which you has showed to me. I believe that I will solve the problem finally even if it is diffucut.

Thank you again!

youf.
Hi,Ivar Thanks for your help very much! I will try the ways which you has showed to me. I believe that I will solve the problem finally even if it is diffucut. Thank you again! youf.

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.