CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Transitional Turbulence (

Andres Bernal Ortiz January 18, 2007 16:35

Transitional Turbulence
Hi, i'm testing the Transitional Turbulence Model in CFX's a very tricky model to make it coverge..Mr. Horrocks and all the other Gurus can tell me some advices to make these model work?

Re 600e3; Airfoil NACA4415 and NACA4418 AoA from -4 to 12 degrees High Resolution, Physical Timescale, MAX 1e-4

I change the High Resolution for Specified Blend factor of 0.75, 0.8 and 0.9...the solver shows an image like an electrocardiogrm.

After that, I change the Physical Timescale for Agressive, Conservative and Local TimeScale factor of 3,4 y 6...nothing happens...

I have a mesh with expantion ratios < 1.02 and y+ < 1.

Please help me, i been trying for 5 days and my solution don't converge.

Thanks in advance,

Andres - Colombia.

Glenn Horrocks January 18, 2007 17:46

Re: Transitional Turbulence

The transitional turbulence model has many enhancements in the current version (CFX10, soon CFX11) compared to CFX5.7.1. One recommendation is to update to current version of the software.

I assume you are doing a steady state simulation. If the residuals converge a little but do not fully reach convergence, first try a larger physical timestep. Do not use local timescale factor for the final run to convergence.

If that does not work then consider whether you have sufficiently converged anyway. Plot important parameters (maybe lift and drag, maybe transition point whatever you regard as important) and if they are steady then you may well be OK. As a final check here, write the residuals to the results file to check where the high residuals are. If the high residuals are not in an area of importance then you are probably OK.

If the high residuals are in an area of importance (and this will probably be the case if the airfoil is stalling or close to it) then a steady state simulation probably is not suitable. You may be able to do a RANS transient model, but more likely you will need a SAS/DES/LES type solution. These models increase the sophistication of the model considerably and therefore will require significant research and development before you can trust the answer.

Regards, Glenn Horrocks

(and I assume everything else in your simulation is OK) the simulation 2) If you are running steady state and the solution is really transient

Andres Bernal Ortiz January 18, 2007 19:04

Re: Transitional Turbulence
Thanks Mr. Horrocks I really appreciate your comments.

I assume steady state and MAX 1e-4...i use monitor points to check Cd, Cl, Cm results....the NACA summary says that at -4g the Cl must be around zero..the solution dont't converge but i obtain a Cl of 0.01 in these case..i'm gonna check the residuals in the results file...


Andres Bernal.

All times are GMT -4. The time now is 11:22.