COMSOL Multiphysics: Nonlinear solver did not converge. Maximum number of Newton iterations reached. Last time step is not converged.
As you can see in the model attached, (COMSOL 4.4 , 2D axial symmetry). A plate over a 10 turns coil. The coil is charged by an electric circuit (cir). All to simulate the electromagnetic force kicks off the plate ( mf ). the speed and displacement of the plate is calculated by Global Equation (ge).
Everything works fine until adding the Moving mesh ( ALE) ! gives the error shown before !! i tried everything i know but nothing changed !
Anyone ? your help is really appreciated.
Hi Mohmmad.
Interesting model you have there! I've opened your file and ran the solver. It seems that the problem of loss of convergence is caused by a too large displacement of the mesh. If you plot the mesh (right-click a surface plot on the model tree and then choose mesh) you'll see that you have mesh elements in the central part of your domain that are streched to the point of being inverted (notice the log when you run the simulation) and that their quality is quite low. In my experience, moving mesh doesn't work very well with large displacements. There are two things maybe worth trying:
1) refine the mesh below your subdomain 3. If you have more elements there, they will be less stretched;
2) break up the simulation into small steps: add a stop condition to the solver to make sure that the calculations are halted whenever the quality of the mesh deteriorates. You can then remesh your domain (using the results) and restart the simulations from the previous displacement.
Hope this helps. Good luck!
Pedro
Edit: adding a boundary layer mesh to the bottommost part of domain 3 might also help controlling the mesh.
Hi Mohmmad.
Interesting model you have there! I've opened your file and ran the solver. It seems that the problem of loss of convergence is caused by a too large displacement of the mesh. If you plot the mesh (right-click a surface plot on the model tree and then choose mesh) you'll see that you have mesh elements in the central part of your domain that are streched to the point of being inverted (notice the log when you run the simulation) and that their quality is quite low. In my experience, moving mesh doesn't work very well with large displacements. There are two things maybe worth trying:
1) refine the mesh below your subdomain 3. If you have more elements there, they will be less stretched;
2) break up the simulation into small steps: add a stop condition to the solver to make sure that the calculations are halted whenever the quality of the mesh deteriorates. You can then remesh your domain (using the results) and restart the simulations from the previous displacement.
Hope this helps. Good luck!
Pedro
Edit: adding a boundary layer mesh to the bottommost part of domain 3 might also help controlling the mesh.
Thank you Pedro , that was helpful. I will try and let you know.
Really Thanks
Mohammad
I have been trying to simulate a microcantilever where it is fixed at one end and the other one is free.it is a frequency domain study.a surface preasure is applied on the top surface.
I get the following error message:
Failed to find a solution.
Maximum number of Newton iterations reached.
There was an error message from the linear solver.
Hi Pedro G C Almeida
I am interested in partition study into steps. Can you give me links or models which I can read more about ?
Thank you very much
Respected,
Can you tell me how to give pulstile velocity profile @inlet in comsol?
Hi Zahra Elmi
1) Do you have all the required modules?
2) I solved your model and only had Warnings:
There was an error message from the linear solver.
The relative error (0.0073) is greater than the relative tolerance.
The relative error (0.015) is greater than the relative tolerance.
The relative error (0.024) is greater than the relative tolerance.
The relative error (0.0029) is greater than the relative tolerance.
The relative error (0.0011) is greater than the relative tolerance.
3) Attach a solution to your model.
4) And sample results:
% Model: soal.mph
% Version: COMSOL 5.0.0.243
% Date: Sep 10 2017, 07:21
% Table: Table 1 - Point Evaluation 1 (solid.uAmpZ)
% freq Displacement amplitude, Z component (µm), Point: 34
0 0.07774166739327962
500 0.07867250852024096
1000 0.08160326269133765
1500 0.08700174700801187
2000 0.09587324671733834
2500 0.11031894132941536
3000 0.13517473396634697
3500 0.18410834950658
4000 0.3156862667263846
4500 1.6439470679871646
5000 0.4492030393767804
5500 0.1856263870918627
6000 0.1134110503242486
6500 0.07979680239366714
7000 0.0601713918234586
7500 0.04808271599256202
8000 0.03946258493185837
8500 0.033166791284881564
9000 0.028390632110765247
9500 0.024659820874630884
10000 0.02167724223311475
Hope this helps. Good luck!
Time dependent analysis works better than stationary analysis because inertia and damping help stabilize the model.
- Badges
- Science topic