CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   RSM issues (http://www.cfd-online.com/Forums/fluent/37534-rsm-issues.html)

 Jonathan Vogt August 10, 2005 06:28

RSM issues

Hi there, I read Razvan's post in one of the ealier threads where he mentioned that the RSM model sometimes has difficulty converging in steady problems - possibly due to problems modeling the flow shedding off surfaces (didn't quite follow it). Could someone eleborate slightly or direct me to where I can get a better understanding of this limitation? thanks. Jonathan Vogt.

 razvan August 11, 2005 05:33

Re: RSM issues

What I said was:

"RSM will very improbably be able to obtain a stationary solution and you will be forced to do a transient calculation. This is maybe the most common reason for difficult convergence with RSM (but people generally do not understand that RSM is sometimes too much physics for a stationary solution)."

And what I meant is:

- RSM, as we all know, is a much more complex turbulence model than all the other 3 I mentioned, let's say quite a "complete" one, which can account for many effects (like curvature and turbulence anisotropy) that other models cannot.

- this complexity can sometimes lead to difficulties for the steady solvers to "average-out" major and even medium instabilities (like vortex shedding) detected by the model, which translates into impossibility to obtain converged solution (the flow isn't settling down in a steady manner). Ironically, this is not a deffective calculation, just the assumption of steady flow is wrong!

- the solution is very simple (if you still want to use RSM):after detecting the instability in the solution (the residuals seem haotic but if you'd have just a little patience you'll see they will remain approximately at the same level!!!), switch to unsteady and continue just as in a say, cylinder calculation, using data sampling to do a later averaging of the results.

Not surprisingly, this behavior is quite often when using RSM with coupled solver (in which time dependency is always included, no matter what formulation is used, steady or unsteady), but I've encountered this problem using two-equation models in specific problems, like axisymmetric supersonic inlets with normal wave "buzzing".

I know that this is no good news, but try to think about LES and you'll see that all this makes sense (LES is even more physics than RSM and a steady calculation using LES is a blasfemy!).

I hope you're satisfied with my answer now.

Best whishes, Razvan.

 All times are GMT -4. The time now is 21:34.