CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Transition SST model - convergence problem with 2nd order discretization (https://www.cfd-online.com/Forums/fluent/118115-transition-sst-model-convergence-problem-2nd-order-discretization.html)

xara May 21, 2013 08:40

Transition SST model - convergence problem with 2nd order discretization
 
3 Attachment(s)
Hi,
I'm currently running a 2D case - airfoil in the tunnel flow. My task is to analyze the transition onset on the airfoil in the flow velocity of 0.7-0.8 Ma. I use Transition SST model (4eqn) to be able to properly predict flow in the boundary layer. My problem is when I switch form 1st order to 2nd order discretization the whole flow begins to diverge. At first the small oscillations of velocity start to appear on the airfoil lower surface, upstream. They look like small separation bubbles. They spread on the whole lower surface and cause divergence in all results: residuals, Cp, Cf, velocity. While the 1st order discretization works fine, after switching to 2nd order I can't get any converge solution. The same happens if only the 'Flow' is switched on 2nd order.

I attached pictures of the case boundary conditions, velocity contours and Cf results for 1st order and 2nd order solution.

Solver parameter are:
Solver: Density-Based
Time: Steady
Gradient: Green-Gauss Node based / Green-Gauss Cell based

Basic Boundary Conditions for Farfield are:
- Ma = 0.7
- AoA = 0 degrees
- Intermittency: 0
- Turbulence Intensity: 0.01%
- Turbulent Viscosity Ratio: 0.01

I have already tried many solutions, but the problem still exist. I have tried:
- very low under-relaxation factors
- different Mach numbers: 0.1, 0.3, 0.6, 0.7, 0.8 Ma
- different angles of attack: 0, 0.5, 1, 2, 3, 4 degrees
- two times denser mesh
- different values of 'turbulence intensity' and 'turbulent viscosity rato' - up to 10

It seems like it was some trouble in turbulence model.
I'm fighting this problem for a long time now and I really appreciate any help.

Thanks!

RodriguezFatz May 22, 2013 09:15

Sounds like checker-boarding oszillations, doesn't it? If you switch to second order discretiuation you kind of decouple adjacent values and such oszillations show.
Edit: Can you try pressure-based solver?

MAC88 May 27, 2013 07:43

Re-check your boundary conditions again. In first order, it add dissipation to the main flow that's why you get results but not in the case of second order.
I think you should look on couple of things:
1. Reduced the under-relaxation parameters
2. Run your case in unsteady mode

Best of luck!!!

xara May 27, 2013 08:20

Hi,
thanks for your help. I've already tried lower URFs and transient flows - it didn't help. However, today I've tried another approach, recommended by my colleague, and it seems like problem is finally solved. I switched solution method to "coupled scheme" and set "pressure-based solver" and I did calculations for Standard SST model (second-order discretization). After the convergence I switched the model to Transition SST model, leaving all the other settings untouched. The residuals and other parameters are converging and there aren't any oscillations on the airfoil surface. The results seem to be correct. It looks like it all depended on setting the right scheme of "pressure-velocity coupling".

I hope this tip will be useful for other Fluent users who use Transitional models.

Thanks!

MAC88 May 27, 2013 14:16

Good best of luck!!!
:)


All times are GMT -4. The time now is 00:53.