CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   Duct structure in Wind Tunnel: Continuity problem? (

JR22 March 31, 2013 16:20

Duct structure in Wind Tunnel: Continuity problem?
2 Attachment(s)
Hi Foamers,

I am trying to model a duct on a pole within a wind tunnel. I am using helyxOS initially for meshing and problem setup as I develop the skills to correctly use OpenFoam. I am using simpleFoam, SST-kappa-omega turbulence model, and about 8000 iterations. I am trying to measure the flow coming into and out of the duct, and at the center of it.

The duct is composed of a nozzle, a center section, and a diffuser (See top of figure). Using Paraview, I placed 3 planes, one at the front, one at the center, and one at the back. I then use paraview to resample to source (project onto the surfaces, called sources in paraview), then I integrate on them (See bottom of figure).

I took the integrated values for the velocity vector and compared them for several mesh qualities. Towards the higher quality meshing (bottom of the table), results seem to not vary too much. However, the velocity vector in the front is NOT the same as the velocity vector at the back. Does this make sense?

Thanks in advance.

JR22 April 1, 2013 08:09

Appropriate mesh size for the SST-kappa-omega
2 Attachment(s)
As I increase the mesh size in the inner portion of the duct, the picture starts to look clearer. Here is an updated image of the 3 surfaces in the duct + the data from the last run (ran overnight for 7 hours). The back surface looks like the image the commissioner uses to call batman, but any resemblance to reality is mere coincidence. The updated table here shows that the integration values are changing and starting to look more like reality. I was expecting more acceleration at the center of the nozzle-diffuser.

Any advice on mesh refinement?
Is there something wrong with my model besides mesh size?
Is there a more appropriate way to integrate (putting a patch that allows flow through it? -- is there something like that?)

JR22 April 1, 2013 09:25

Maybe the outlet is not an outlet?
1 Attachment(s)
While going back to another thread ( ), I tried to put the money where my mouth was, and found that actually the outlet on the duct is not really an outlet in the strict sense. I mapped the glyphs on the outlet surface of the duct, and there is actually some flow going in. So it is possible that all of the flow that goes through the center doesn't necessarily addup to the outlet of the duct, or should it? Does this make sense? See the image below.

immortality April 1, 2013 09:51

an interesting do you resample and integrate on planes as you told?could you explain me a little?

JR22 April 1, 2013 10:39

Integrate on Mapped Surface using Paraview
I posted the steps to get the Glyphs on the mapped surface on the thread in the link (bottom of post). I am replacing the Glyphs with the Integrate here. If you are doing both, they have to both be children of the ResampleWithDataset Object.

1. Create a Plane from the Source menu,
2. Change the X and Y Resolution of the plane to something like 200x100 (x,y). By default, the resolution is 1x1, which is wrong if you are going to do step 3 below.
3. Use the Resample with Dataset filter (the source is the created plane, and the input is your entire dataset).
4. Use Integrate Variables filter on your ResampleWithDataset Object. The integration results will show on a new pane with a spreadsheet kind of look. You can go back to changing the resolution of your source (part of step 2) until your results don't vary much.

Paraphrased from (that one shows how to do the glyphs):

All times are GMT -4. The time now is 14:49.