Please! I really need your help!
Error: Nonlinear solver did not converge.
Maximum number of Newton iterations reached.
Last time step is not converged.
The model is in attachment. Two physics (bioheat physics and moving mesh physics) are included in model to simulate a laser ablation over a part of the bone. It worked well when there is only bioheat physics, but when i use bioheat+moving mesh physics i reach this error.
At first, the laser pulses have a radius of 3.8mm to create a 1mm hole. The radius of the laser pulses is then readjusted to 3mm and the ablation processes is then continued for an additional depth of 3.5mm in the z-direction.
Anyone can help me to see what happened? your help is really appreciated. I have struggled with this problems for long time. Thx!
Hi,
If you are trying to run your model (older version 3.5a) in newer versions of Comsol, some features may be 'obsolete' and need to be reconstructed. In general, if you are certain that your boundary conditions are correct with good initial conditions, you can tune and improve the nonlinear solver convergence following these steps:
1- Use 'Fully coupled' solver
2-Under 'Fully coupled' node, change the nonlinear method to 'Automatic(Newton)', change 'Jacobian update' to 'on every iteration', increase the 'maximum number of iterations' to 25 and reduce the 'tolerance factor' to 1e-3.
Hope that helps
Hello,
in principle there are many reasons why a computation does not converge.
One possibility is, that you solver just needs some more steps. In this case your convergence plot in comsol should show a decreasing curve at the end of your calculation. You can just increase the number of steps and see if convergence improves.
If this doesn't help and your convergence plot plateaus at a non-satisfactory level, you may try other algorithms. For this, the following blog article might help:
https://www.comsol.com/blogs/improving-convergence-multiphysics-problems/
Furthermore, convergence can be bad if your initial guess for the solution is bad. So playing with the initial conditions can also help to converge the results.
I hope this helps you a bit.
Best regards,
Kai
Hi,
If you are trying to run your model (older version 3.5a) in newer versions of Comsol, some features may be 'obsolete' and need to be reconstructed. In general, if you are certain that your boundary conditions are correct with good initial conditions, you can tune and improve the nonlinear solver convergence following these steps:
1- Use 'Fully coupled' solver
2-Under 'Fully coupled' node, change the nonlinear method to 'Automatic(Newton)', change 'Jacobian update' to 'on every iteration', increase the 'maximum number of iterations' to 25 and reduce the 'tolerance factor' to 1e-3.
Hope that helps
- Badges
- Science topic