Discussion Forum

Dear Community

I am trying to build a 3D model where a coil wound around a soft magentic iron core with a copper cylinder inside the core. I am trying to inject a current in the cylinder and sense the induced voltage in the wounded coil. When we define the material of the soft magentic core with only permability "mur", the model works well. However when we try to define the core material with Jiles Atherton model for Hystersis , the model did not work at all. When using either the iterative solver or the direct one ( PARDISO , MUMPS), the time dependent solver has trouble converging, and the following messages appeared in the trails : Error Message: Failed to find a solution. The relative residual (12) is greater than the relative tolerance. Error message: Singular Matrix Error message: Matrix has zero on diagonal

We tried to adjust the model by using the suggestion to improve the convergence in the link below: " adjust are the Jacobian update (from Minimal to On every iteration), the Maximum number of iterations (from 4 to 25), and the Tolerance factor (from 1 to 1e-3). " https://www.comsol.com/support/knowledgebase/1127/

but the same error still there. After that we try to use the following suggestion : "Try the Vanka preconditioner/smoother instead of SOR and Jacobi (diagonal scaling). " https://www.comsol.com/support/knowledgebase/1179/

The returned solution is not converged. it seems that failure to converge is due to strong non-linearities.

Do you have any suggestions to solve this problem ?