CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)
-   -   k-omega-SST model (OF 1.6) - turbulent flat plate (https://www.cfd-online.com/Forums/openfoam-solving/72978-k-omega-sst-model-1-6-turbulent-flat-plate.html)

tcarrigan February 5, 2011 02:41

To cboss:

Do you think you could post your setup, 0/nut, 0/k, 0/omega, fvSchemes, fvSolution, etc...?

I've been trying to simulate turbulent flow over a flat plate in OpenFOAM-1.6-ext for both a resolved grid with a y+=1 and an unresolved grid with a y+=60. I'm also using the kOmegaSST model with and without wall functions to handle both grids.

The problem I'm having is that when running the resolved grid using the kOmegaSST model, I'm slightly underpredicting the skin friction by just a few percent. However, when running the unresolved grid with wall functions for the kOmegaSST model, I'm severly underpredicting the skin friction.

Looking at your images, the behavior I'm seeing for the unresolved grid is similar to the behavior of the y+=30 grid you plotted in your third image when you were comparing the results of cases without using wall functions. This leads me to believe that either I'm not implementing the wall functions correctly, or the wall functions aren't being utilized during the simulation at all.

Thinking this was an issue with 1.6-ext, I installed OpenCFDs 1.6 version from mid February 2010, about the time of your original post. I reran the case using this version and observed the same behavior. So it seems it might be my problem.

Any help would be much appreciated.

cboss February 5, 2011 07:46

Hi,

nice, that this post is still recognized.

To your questions: The problem is, that I haven't been working with OpenFOAM for a long time, and that I don't have access to the done simulations. And it's quite complicated to get into it again, if you have neither a version of OF, nor any data. So I'm afraid, that I cannot give you a good advice.
Perhaps someone else can help.

Regards

maruthamuthu_venkatraman February 5, 2011 10:50

Hi,
Your answers are kept in Turbulence modelling for CFD by Wilcox . Just have a look at it how to set omega values at the nearwall cells. In general refer Ercoftac best pracise guidelines recommended BC's for external flows.

Good Luck. I dont think schemes will make big difference in this case.

Regards

Far November 13, 2011 02:28

Quote:

Originally Posted by cboss (Post 257936)
Nick,

(3) Yes, I set k as fixedValue=0.

In the original paper it is mentioned as flux = zero

choux February 24, 2016 04:52

Hello, cboss!
You said you were working on turbomachinery computations. I am wondering your progress because I plan to star my PhD research in this field. However, I am not familiar with this field. Could you recommend some review articles, especially thsoe with dynamic grids? Besides, could you give me some advice to seek some new research points in this field?
Best wishes!

Eloise August 9, 2016 10:53

1 Attachment(s)
Hello Foamer,

I'm digging out an old post with a validation case of a turbulent boundary layer over a flat plate. The details of the validation case are as in the first post of this thread.

I ran the case with two solvers (FineMarine4.2 and simpleFoam from OF2.2.1). I used the same mesh with both solvers, using the wall function (yPlus~60).

I compared the wall shear stress along the plate with Wieghardt's data in the attached figure. The results obtained with FineMarine match well the experimental data while the OpenFoam results are off.

However, if I check the drag force on the plate, I obtain the following values which are relatively close to each other:
OF -> Fx = 0.405871 N
FM -> Fx = 0.414316 N
Therefore, it looks like the OpenFOAM simulations are correct but that I actually don't post-process the wall shear stress properly.

I extract the wall shear stress using the wallShearStress and the sample utilities:
Code:

// Fields to sample.
fields
(
    wallShearStress
    wallGradU
    nut
);
sets
(
    plate
    {
        type        face;
        axis        xyz;
        start      ( 0.0  0.0 0.0);
        end        ( 6.0  0.0 0.0);
    }
);

Then, I compute Cf(x) = wallShearStress_X/(0.5*U_0^2)
omitting the density since simpleFoam uses values normalized with rho.

I also attempted to compute Cf(x) based on wallGradU:
Cf(x) = wallGradU_X*muEff/(0.5*rho*U_0^2) with muEff = rho*(nu+nut)
but it doesn't seem to be the proper way neither.

Both of those expressions for Cf seems to be wrong, is there anyone who can help me?

Regards,
Eloïse


All times are GMT -4. The time now is 11:01.