CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   flow over a smoothly contoured ramp (

schludom July 6, 2013 11:17

flow over a smoothly contoured ramp
5 Attachment(s)
Hello everybody,

currently I am investigating on the flow over a contoured ramp using different turbulence models and 2 meshes (one with y+=30 and one with y+=1)

At first I used the k-omega-sst model for both meshes. Everything seems to be ok. There are just small differences between that cases regarding boundary layer thickness and the velocities, but that should be because of the wallfunctions i used for the y+=30-meshes.

Now I have set up my cases with the k-epsilon-model. Again the y+=30 meshes look ok, even in comparison to the k-w-sst cases (y+=30 and y+=1). My problem now are the low-Reynolds-cases using the k-epsilon-model. The velocity-profiles differ greatly from the others.
I think my problem are the BC's at the wall. for k i use fixedValue 1e-12 and for epsilon i tried now 3 ways. At first i used epsilon =fixedValue 1e-12, then zeroGradient and at least i tried the epsilonLowReWallfunction.

I attached some pictures to make my situation clear. One picture shows the geometry. I am solving only 2 dimensional. The point where i extracted the velocityprofiles from is upstream, near to the ramp, but far away enough to be not influenced by the ramp. the Freestream velocity should be around 20.4 m/s and the Bl thickness around 25.3mm(experimental data). The y+=30 picture shows how my other solutions have been so far, even when i used kwsst for the y+=1-mesh.

Sooo. I would really appreciate some hints, how to set up my case using the k-epsilon model and my y+=1-Mesh, to get nearly the same solutions as before :)

fumiya July 7, 2013 02:33


I understand the k-epsilon model implemented in OpenFOAM is standard high-Re type.
Please correct me if I'm wrong.

So, you might want to try out the low-Re turbulence models such as LaunderSharmaKE:

Hope this helps,

schludom July 10, 2013 06:25

Hi Fumiya,

thanks for your reply. Unfortunately the LaunderSharmaKE-Model didnt work for me. I also tried the SpalartAllmaras.model and it worked fine. Using LaunderSharma caused an unphysical thickening of the BL and i still have no clue why. My BC's for k and epsilon on the walls are fixedValue 1e-12 and i set nut to calculated. Regarding the way nut ist calculated in LaunderSharma this should work, or am i wrong?

All times are GMT -4. The time now is 05:33.