CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   SST k-w model--the parameter 'omega' does not convergence (http://www.cfd-online.com/Forums/fluent/101510-sst-k-w-model-parameter-omega-does-not-convergence.html)

zyy95900 May 6, 2012 00:22

SST k-w model--the parameter 'omega' does not convergence
 
Hi everyone,

I use SST k-w model simulate air flow in a duct. However, the parameter 'omega' does not convergence (about 5e-2). This is happen in all my case--different mesh (good and bad) and different y+(1-20). Does anyone have any ideal to help me fix the problem?

Thanks.

LuckyTran May 6, 2012 14:21

Quote:

Originally Posted by zyy95900 (Post 359518)
Hi everyone,

I use SST k-w model simulate air flow in a duct. However, the parameter 'omega' does not convergence (about 5e-2). This is happen in all my case--different mesh (good and bad) and different y+(1-20). Does anyone have any ideal to help me fix the problem?

Thanks.

There are many reasons for poor convergence and they can enter at any point. Starting from mesh, to boundary conditions, to the setup of the physical models. How good is your mesh? You said good and bad, but how good is good. The list goes on, try to provide some details.

Check some actual engineering values to see your simulation is converged. Scaled residuals are a poor method for judging convergence in the first place. How many iterations have you tried?

zyy95900 May 6, 2012 16:14

Quote:

Originally Posted by LuckyTran (Post 359578)
There are many reasons for poor convergence and they can enter at any point. Starting from mesh, to boundary conditions, to the setup of the physical models. How good is your mesh? You said good and bad, but how good is good. The list goes on, try to provide some details.

Check some actual engineering values to see your simulation is converged. Scaled residuals are a poor method for judging convergence in the first place. How many iterations have you tried?

Thank you Lucky Tran.
The good mesh is in smooth duct geometry with Quad map type mesh (The mesh uniform everywhere). The bad mesh is in some roughness surface duct (The skew>98).
Can you tell me how can I check the actual engineering values to see the simulation is converged? I have tried 2000 iterations but the parameter 'omege' is not changed after just 25 iterations (keep at 5e-2) and other parameters are not changed after 500 iterations.
Any more idea to solve this problem?

Thanks.

LuckyTran May 6, 2012 16:42

Quote:

Originally Posted by zyy95900 (Post 359586)
Thank you Lucky Tran.
The good mesh is in smooth duct geometry with Quad map type mesh (The mesh uniform everywhere). The bad mesh is in some roughness surface duct (The skew>98).
Can you tell me how can I check the actual engineering values to see the simulation is converged? I have tried 2000 iterations but the parameter 'omege' is not changed after just 25 iterations (keep at 5e-2) and other parameters are not changed after 500 iterations.
Any more idea to solve this problem?

Thanks.

If your y+ is ~ 1, and your mesh is uniform everywhere, then you basically have a DNS quality mesh. Is this the case? For some reason, I doubt you have that type of mesh.

Check solution values by creating point, line, or surface monitors of some relevant engineering quantities (pressure, velocity, temperature, turbulent kinetic energy, etc).

zyy95900 May 6, 2012 23:58

Quote:

Originally Posted by LuckyTran (Post 359590)
If your y+ is ~ 1, and your mesh is uniform everywhere, then you basically have a DNS quality mesh. Is this the case? For some reason, I doubt you have that type of mesh.

Check solution values by creating point, line, or surface monitors of some relevant engineering quantities (pressure, velocity, temperature, turbulent kinetic energy, etc).

You are totally right. I just re-run the case. The convergence is fine for good mesh:). The y+ is 1.1-1.2. I think I confused the cases because I got so many to run:o. I think I should try to improve the mesh quality for no convergence cases.
Thank you for your help.


All times are GMT -4. The time now is 21:33.