Hello cfd-community,

actually I am working on a problem with a - for me - very high Re-numer (about 1E6). Since I am accustomed to lower numbers I am a little uncertain about how to handle some of the problems which appeared:

1.) During the solving process the turblulent vicosity ratio was limited just inside the area of interest. When I estimate the turbulent intensity I by I=0.16*Re^(-0.125) it gave similar values to what Fluent reported. So my question is whether it is reasonable to increase the default limit due to the high Re to avoid hitting the limit or whether s.th else could be wrong (may be it is related with 2.

2.) To achieve 30 < y+ < 60 I would need too much cells. Sung-Eun Kim from Fluent mentioned in the past that it is ok for high Re flows to increase the upper limit. I followed the advice and generated cells to achieve y+ about 500. The question is: Is s.o. able to assess, whether 500 is still ok for such a Re? (the domain is similar to a t-pipe)

Volker

1) I've seen Fluent have difficulty with the turbulent viscosity ratio. If you expect the result to be near the limit, then by all means raise the limit. If you expect the result to be much smaller than the limit, then you need to try to solve it in a different manner. Some possibilities: (A) temporarily change the boundary conditions (less turbulent) and get a solution, then use that as your initial conditions for the real boundary conditions. (B) Take time and give good initial conditions (use the patch tool to patch your turbulent parameters - more turbulent near disturbances...). (C) Solve the flow with laminar fluid, change viscosity to turbulent, turn off momentum equations and then solve just the turbulent equations, turn on momentum equations and solve from there.

2) Try to model a small section of your domain. Do it with an initial y+ of 50 and again with an initial y+ of 500. Do the results change? If so you cannot use the larger grid.

(1) I've had problems with high and limited TVR as well. I'd like to know more on this subject as well. I've spent a lot of time trying to avoid this limit with limited success and almost no difference in final results.

(2) I've had good correlation to test data with max Y+ of 150 to 200. I think 500 is too high. You can try region adapting in just the area of interest to get Y+ levels down and not increase the model size unnecessarily. In the worst cases you'll need to generate a prism boundary layer mesh.

John

Hi Scott,

thank you for your answer.

Due to the high estimated turbulent intensity I mentioned I should expect high turbulent viscosity ratios. Hence:

Meanwhile I raised the limit and got rid of the warning.

Next I should test 2.) of your proposals.

Volker

Hi John,

thank you for your answer.

1.) I think 2 is a good idea which I already carried out. This helps to keep the mesh small and useable.

2.) Your statement concerning y+=500: is it just a feeling or can you deduct it from the correlation at y+=200?

Volker

P.S. Have a look to my answer to scott.

I've actually had areas of flow separation and recirculation that did not occur in a model with Y+ of 500 that were indicated in a flow viz test. When the Y+ levels were brought down into the 100-150 range it did predict the test results.

John