I'm applying RSM to simulate a problem of external aerodynamics, but the solution doesn't converge. I'm doubtfull about the following Reynolds stress options: 1)Wall reflection effects 2)Wall B.C. from k equation 3) Quadratic pressure strain model. Could one of you suggest some references I could look at in order to deepen my understanding of these subjects? In addition, one of you kindly suggested that the default constants used by Fluent are good also for bluff bodies problem subject to wind. Could one of you also suggest a reference for this? I will really appreciate any suggestions! Thanks in advance! Anto

If you are calculating a bluff body, then solution is highly unsteady, so RSM will never "converge" in a steady-state simulation. There is nothing wrong with model's settings, just have to start calculating unsteady, using as intial solution the steady-state one.

Sorry, but this is a case where you cannot use RSM with steady-state assumption! If you really don't have the computer power or patience to go unsteady, you must use another model, say Realizable k-e, which will be able to give you a converged solution.

Best whishes, Razvan

Sorry about not being clear in my explanation. The bluff body has an elongated shape, that lets me assume a steady case in the longitudinal direction. This assumption is not possible in the orthogonal direction, which I am studying as an unsteady case.

In the longitudinal direction the solution converges with Spalart Allmaras , and k-epsilon; .

I want to apply also RSM.

Are you now able to reconsider my problem?

Many thanks for all your help. Kind regards, Anto

Doesn't matter. If this body does not have a aerodynamic end shape, there would still exist vortex shedding behind it and RSM will detect it, so you won't be able to get a properly converged solution in steady-state.

I have already discussed this "problem" in several other cases and tried to make people understand that only flows with very low degree of unsteadyness can be treated using RSM in steady-state!

Best whishes, Razvan

The body has an aerodynamic shape in the longitudinal direction, I would have sent you an image if it were possibile. Infact the steady RSM oscillates in a very small range, which is probably due to the fact that RSM is more sensitive. However, I followed your suggestion and I switched it on unsteady and the solution reaches stability . The solution is very close to the steady Spalart Allmaras simulation and to the experimental data. However, the steady k-epsilon option with non-equilibrium function gives a solution which is one order smaller than the others. I expected a smaller coefficient with k-epsilon, but this solution appears much too small. Do you agree?

Thanks for all your advice and for sharing your experience with me!

Anto

Of course it is too small. But this is quite surprising for me (if you used Realizable k-e), because Rke results should have been only max.10% out of range! But if you used standard k-e, then I can't tell you else than: don't use it anymore for this type of application!

Results from Spallart-Almaras are not surprising either if you say your body has aerodynamic shape (I have encountered this myself many times in aerospace domain). This could be a lesson for everybody to know how to choose turbulence model in the future!

Best wishes, Razvan

I used the k-epsilon realizable, and I was also very surprised with the results. I have another request... I have already asked on this forum if the defalt constants in Fluent can also be used for incompressible fluid, and the answer was yes. Now I would like references about it possibly for each different turbulent model. I would particularly like a reference that examines the dependency of the results on the constants. Thank you so much for your help!