We study numerically an inclined two_dimensionnal (2D) wall jet. Experimentally, we observe that the jet reattaches to the wall for a Reynolds around 1000, but with Star CD and Fluent we obtain a critical Reynolds around 10. Have you encounter this phenomen? Why CFD don't give good results? Thank you. C. ALLERY

(1). There are several possibilities: (2). One, you don't understand CFD. (meaning that you don't know how to write equations and code to solve the problem.) (3). Two, the vendor does not know CFD. (they have never solve your problem before.) (4). Three, you never ask the vendor "whether you can use their codes to solve your problem". (5). If you can verify that your results can be repeated, then that is what you are getting. (6). I don't know how to solve your problem, because I have never tried to solve such problem before. It is hard to know what's wrong. (since you are able to obtain some solutions, I am assuming that you know how to use the code.)

Just from curiosity. What's the characteristic length? The width of jet? The flow must be turbulent.

I did the sort of thing years ago. Here, I mean, not the flow you are interested in but the comparison between those two codes. I got totally different. One gives transient and the other one steady. Headache!!

Separately, I did the numerical analysis for 3D jet with inhouse code. Not so great.

Any way, you can either compare the results from both codes or check your input to the codes if you are sure you did sussesfully perform your EXPERIMENTS keeping the flow 2D. However, for NUMERICAL analysis, we can easily expect the difficulties in applying (1) boundary conditions and (2) choosing the very location of boundary which may be different from your experimental setup. We can use boundary conditions either which satisfying Euler equation (at far upstream) or use just neuman. Of importance might be turbulence model and input intensity level etc. Another headache.

But if you are very confident of you experimental results, things are going to be a lot easier.

How were the results from STAR and Fluent. Did they give you much different answers even though you used the same b.c. and domain.

Hello, Thanks for your response.

The wall length is 500mm, the width of jet is 5mm (rtae L/h=100). The reynolds number (Re=h*U/15e-6) is 100 so the velocity is U=0.3m/s. The flow is laminar.

Star CD and fluent give the same results with the same b.c. and domain: the jet is attached to the wall. A 3D simulation gives the same thing.

C.Allery

Several areas to check:

1) Do you have adequate grid resolution in the vicinity of the jet and its reattachment areas? i.e.is the solution grid independent? 2) Are the solutions fully converged? Some times jet interaction solutions can take quite a while to converge. Check convergence not only for residual reduction, but by monitoring the change in properties of interest (wall pressure or shear, separation location, etc.) 3) Is it possible that the jet is tripping your laminar upstream boundary layer to turbulent? 4) Are you including the viscous terms in the streamwise direction? (i.e. full NS vs. thin-layer). Inclusion of these terms is critical for jet interaction flows.

You may have checked all these out, they are just some areas where I've seen difficulties with jet flow solutions in the past. Best wishes for your simulations.