[Zappstar]

Hi!

I am relatively new user in OpenFOAM and in order to tryout the software I decided to simulate flow around a circular parabolic disc. I am using CAE Linux 2013 and thereby
Salome v6 as preprocessor and Code-Saturne wizard to setup the solver. The results are then studied in both Paraview and the included tool in Salome (Post-pro).

The same model has been simulated in ANSYS FLUENT as well, and the result are then compared.

The results are as you can see in the attached photos are rather different.
Fluent seems to give result that are rather intuitive. An by that i meen that the contour plot varies with radial length, and also slightly influenced in the lower parts by the floor.

The OpenFOAM solution however is very different as seen in the attached photos. I have tried to make a similar pressure plot as the one made in ANSYS.

In comparison to the FLUENT result the couture plot in OpenFOAM are a tad more erratic in nature. My rather general question is;

Why are the results not more equal?


Some info of the solution,

The disk is 1 meter in diameter an 5 cm thick with an arbitrary curvature.
The computational domain is (x,y,z)=(10 15 5) disk is located at (5 6 1). The concave face of the parabolic disk is normal to the flow inlet.

Mesh is made in Salome.meca and contains approx. 500' nodes. The main fluid domain is made up by tetrahedral elements. The disc surface is made up by viscous layer.

Se the attached file "listing" for information about convergence and other neat stuff.

Settings:
Mostly default settings in Code Saturne wizard, steady flow, k-omega SST, Least squares method over neighboring cells, upwind scheme on all equations. Reference time step=0.01s , Number of iterations = 100, Max CFL = 0.8

Note:
Inlet velocity is set to 13.412 m/s

[Zappstar]

I have now tested the Salome-meca mesh in Ansys CFX and the results are similar to the former Ansys calculation. So my conclusion is that my my main problem is probably not mesh dependent.

[Zappstar]

I have now managed to extract the surface data from Paraview and it looks almost the same as in salome-meca post-pro.There is a consistent erratic behavior that is apparent in both of the post process applications. My best guess is that I am doing something wrong I code-saturn. Does anyone have an idea what I am doing wrong.

Regards

[Zappstar]

I think that I have solved the problem on my own. It appears to me that the problems I have described are related to how you visualize the results in both "Post-pro" and "Paraview". I have added below a picture from a solution that I recently made in Code-Saturn. This time I used a flat disc instead of a parabolic disc. As before the erratic behavior occurs when plotting the results at the surface. But if results are plotted 0.00001 m away from the surface itself, as seen in attached image, the the behavior then disappears.


I believe that this is due to geometrical tolerances that mismatch with the mesh cell sizing. If any one has another idea please let me know.

Regards,

Peter