I will make CFD calculations of a ongoing wind tunnel test. The Machnumber is 0.7 - 1.4 and the wind tunnel nozzle is perforated (transonic nozzle). Normally one have to build a very large grid around the free flying object, but how do I do in a wind tunnel. Must I have the nozzle walls in the model?

Anyone got experience?

I've done few simulation for models in a supersonic and subsonic tunnel. Depending on the exact conditions and the location of the model in the tunnel, the presence of the tunnel walls may cause a slight change in the free stream direction. We noticed this both during the wind tunnel tests and in CFD.

What helped me though was that I was only modeling the centerline of the model and tunnel (ie one z-plane). Fully 3-D computation with reasonable accuracy would have taken way too long...

-- Jarmo

I would recommend doing the problem both ways and comparing the results with experiment. I have done holographic interferograms in both a supersonic tunnel and in a transonic tunnel and found wall effects in both( ei, in the Mach region around one). Our CFD results agreed with the wind tunnel when the wall was included but did not when the wall was not included. A lot depends upon the size of the model relative to the tunnel cross section.

Thanks

Yes I think I have to model the walls, perhaps it is easier, as I can have a smaller mesh.

I'm not sure if about the savings in gridding.. the domain will be smaller, but you need a good boundary layer grid on the walls. If I remember correctly the total number of cells in the computation with walls ended up being larger than in the no-walls case.

-- Jarmo

Yes your right. The problem may be what I have to do with the holes as the transonic nozzle is perorated. Do I have to model the holes on the walls?

Good question... at first I would say yes. But I probably would try to run a case with holes and without holes to see whether that changes anything. It will naturally change something, but will it change the temperature distribution etc. on the model?

The holes will be difficult to include not only because of the grid generation issues, but you may have trouble implementing boundary conditions for those. So your life would be much easier if you did not need to model those. Question remains, how to convince the audience that you are really modeling the experiment if you do not model the holes...

-- Jarmo

I will test without holes. In the end of June I have the results from the Wind tunnel test, and the calculations will be done during this summer, then I know.

Good luck. Keep us posted!

-- Jarmo

You need to model the tunnel walls and make provision for the holes or slots via boundary conditions. The effect of the holes or slots is two-fold:

1) Reduces the growth of the boundary layer on the test section walls thereby keeping the test section area roughly constant as a function of model station.

2) Allows shocks to pass through into the plenum thereby avoiding reflected shocks onto the model.

Both effects have an effect on the results so to accurately model the wind tunnel you need to include these effects.

I don't have the references, but work was done in the 70s and 80s to numerically quantify and model these effects. You might try looking for papers by Jacocks or Erickson. Their work specifically related to perforated tunnel walls.

An alternate suggestion is to compute increments via CFD and compare these to increments measured in the tunnel. The idea is that biases introduced by the experimental setup or by the CFD modeling technique cancel when differencing "similar" solutions giving you an accurate increment. This is exactly how most people use experimental wind tunnel data anyway. Rarely are absolute values used, but rather, absolutes are built up out of a series of increments. This approach would make your far-field approach less important and I would lean toward the free-air technique if you go this way.

Let us know how this turns out.