Hi.

I have made a simulation upon a rectangular box. Inlet in one end, and outlet in the other end. Inlet vel. is approx. 20 m/s. There are no temperature differences. The problem is that I get a too high turbulent viscosity (the turbulent/laminar viscosity ratio is limited to approx. 10^5). I first considered that it could be due to a too fine grid combined with a "coarse" turbulent model (k-e model), so I tried with a coarser grid. No effect. I have also tried other models (RS, LES) but it did not do the trick (well, the LES did with a good initial guess).

Q: Is it true that there are problems with combining a fine grid with a "coarse" turbulent model? Q: Do anyone have suggestions as to solve the high turbulent viscosity problem?

I would appreciate comments to the two questions.

Cheers Christian

In fact turbulent/laminar viscosity ratio=10^5 is typical value. Don't worry. For instance, in my numerical simulations of supersonic turbulent jets flowing into atmosphere I got maximum dynamic turbulent viscosity approx 0.5 Pa*sec (molecular vis is about 1e-5 Pa*sec). I used Menter's k-w model and got good agreements with experiments.

Alexey.

I would think that the more turbulent the flow is, the higher the ratio. And I guess that supersonic can be considered very turbulent, and a high ratio could be expected. My case on the other hand is not that turbulent, and I am just surprised that the ratio is that high, and I am afraid that it might contaminate the solution. I can see that the convergence is more stable when I lower the inlet vel. Do you have any suggestions what to do about a potentially too high viscosity ratio ?

Christian

(1). Take the flow over a flat plate as a test case. (2). Run some cases of flat plate flow, refine the mesh, and check the results against the test data.

Hi,

Try to check the mesh quality, by means of the distributions and values of the y-plus and y-star.

Good luck,