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.

Error message: Nonlinear solver did not converge.

Harikrishnan Jayamohan

Please login with a confirmed email address before reporting spam

I am trying a 2D axisymmetric flow (Turbulent flow) with "Transport of diluted species" module. But it gives me error "Nonlinear solver did not converge.
Time : 23.45118732031299
Last time step is not converged." during Time dependent solver.

It works fine when I try for time steps <20 (secs).
Any idea why this is happening. Is it something with optimizing the mesh.


3 Replies Last Post 2011年4月29日 GMT-4 16:10

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年3月23日 GMT-4 14:16
I am using turbulent model and Nernst-Plank model(diluted mass transfer balance) too. And I cannot converge neither.
Mine is steady state
I am using turbulent model and Nernst-Plank model(diluted mass transfer balance) too. And I cannot converge neither. Mine is steady state

Harikrishnan Jayamohan

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年3月24日 GMT-4 20:30
U'r rite. I tried the steady state condition also.

Here's the suggestion I got from support --> "The convergence problems could either be due to the numerical instability being induced due to a sudden change in flow features beyond run time of 20s or insufficient mesh resolution to support such change in the flow.

It could also occur if the system becomes highly nonlinear and the solver settings are not adequate to achieve convergence.

We would recommend you to activate the probe plot for your results while you are running your model, to check whether the solution looks reasonable before you actually get the error message. The probe plot would help you in identifying the cause of the convergence problems which, once known, can be addressed appropriately."

Have your tried increasing the mesh?
U'r rite. I tried the steady state condition also. Here's the suggestion I got from support --> "The convergence problems could either be due to the numerical instability being induced due to a sudden change in flow features beyond run time of 20s or insufficient mesh resolution to support such change in the flow. It could also occur if the system becomes highly nonlinear and the solver settings are not adequate to achieve convergence. We would recommend you to activate the probe plot for your results while you are running your model, to check whether the solution looks reasonable before you actually get the error message. The probe plot would help you in identifying the cause of the convergence problems which, once known, can be addressed appropriately." Have your tried increasing the mesh?

Harikrishnan Jayamohan

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年4月29日 GMT-4 16:10

U'r rite. I tried the steady state condition also.

Here's the suggestion I got from support --> "The convergence problems could either be due to the numerical instability being induced due to a sudden change in flow features beyond run time of 20s or insufficient mesh resolution to support such change in the flow.

It could also occur if the system becomes highly nonlinear and the solver settings are not adequate to achieve convergence.

We would recommend you to activate the probe plot for your results while you are running your model, to check whether the solution looks reasonable before you actually get the error message. The probe plot would help you in identifying the cause of the convergence problems which, once known, can be addressed appropriately."

Have your tried increasing the mesh?


Finally. I got it to work. One reason mine was goin wrong was that I had put wrong values for dynamic viscosity of the fluid (very high values instead of using ones for water), also the velocity should have been specified in the phi direction (i had it in the r-z plane). If you use the tutorial file in COMSOL (swirl flow) these things shud be sorted out.

Another general solution COMSOL has suggested is (thanks to Ms. Patro from comsol support)-
"
The convergence problem occurs because of the very high value of omega. Convergence can be achieved by proving a better initial guess/solution. One way to do this is to utilize COMSOL's continuation feature, and ramp up the value of omega automatically; each new stationary solver step will use the solution from the previous value as
initial value.

For more details refer to the solver settings used in the rotating_disc example from the model library (CFD Module>Tutorial Models>rotating_disc). As the CFD part of your problem is solved in this model, this can also be
used as a base model for your case. The stationary solution of the flow field can then be used to compute the transient mass transport as follows:

1. Right click on Model 1 and select add physics.
2. Add transport of diluted species.
3. Click next and add Time-dependent study. Click Finish.
4. Set up the Transport of diluted species physics interface by defining
appropriate initial and boundary conditions.
5.Go to
Study 2> Step 1:Time-dependent. Expand the Physics selection
section in the settings window.
6. Click on Turbulent flow and clear the Use in the study check box.
7. Right click on Study1>Solver Configurations and select Show default
solver.
8. Go to the settings window for Study1>Solver Configurations>Solver 2>
Dependent Variables 1. In the initial values of variables solved for and
values of variables not solved for sections use Method: solution and
Solution: Solver 1. Select 12*pi as the Parameter value (omega).
9. Right click Study 2 and compute.
"
This worked for me... instead of taking a few days to solve this gave results in a few minutes.
-Hari
[QUOTE] U'r rite. I tried the steady state condition also. Here's the suggestion I got from support --> "The convergence problems could either be due to the numerical instability being induced due to a sudden change in flow features beyond run time of 20s or insufficient mesh resolution to support such change in the flow. It could also occur if the system becomes highly nonlinear and the solver settings are not adequate to achieve convergence. We would recommend you to activate the probe plot for your results while you are running your model, to check whether the solution looks reasonable before you actually get the error message. The probe plot would help you in identifying the cause of the convergence problems which, once known, can be addressed appropriately." Have your tried increasing the mesh? [/QUOTE] Finally. I got it to work. One reason mine was goin wrong was that I had put wrong values for dynamic viscosity of the fluid (very high values instead of using ones for water), also the velocity should have been specified in the phi direction (i had it in the r-z plane). If you use the tutorial file in COMSOL (swirl flow) these things shud be sorted out. Another general solution COMSOL has suggested is (thanks to Ms. Patro from comsol support)- " The convergence problem occurs because of the very high value of omega. Convergence can be achieved by proving a better initial guess/solution. One way to do this is to utilize COMSOL's continuation feature, and ramp up the value of omega automatically; each new stationary solver step will use the solution from the previous value as initial value. For more details refer to the solver settings used in the rotating_disc example from the model library (CFD Module>Tutorial Models>rotating_disc). As the CFD part of your problem is solved in this model, this can also be used as a base model for your case. The stationary solution of the flow field can then be used to compute the transient mass transport as follows: 1. Right click on Model 1 and select add physics. 2. Add transport of diluted species. 3. Click next and add Time-dependent study. Click Finish. 4. Set up the Transport of diluted species physics interface by defining appropriate initial and boundary conditions. 5.Go to Study 2> Step 1:Time-dependent. Expand the Physics selection section in the settings window. 6. Click on Turbulent flow and clear the Use in the study check box. 7. Right click on Study1>Solver Configurations and select Show default solver. 8. Go to the settings window for Study1>Solver Configurations>Solver 2> Dependent Variables 1. In the initial values of variables solved for and values of variables not solved for sections use Method: solution and Solution: Solver 1. Select 12*pi as the Parameter value (omega). 9. Right click Study 2 and compute. " This worked for me... instead of taking a few days to solve this gave results in a few minutes. -Hari

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.