Dear collegues

I would need your opinion about this issue. The restriction of Reynolds averaging in transient flows is that the period of the average ( = the time step of the simulation??) must be much longer than the maximum period of the velocity fluctuations and much shorter than the mean flow variation with time.

In turbomachinery,however, the time step must be very short, often shorter than the expected maximum period of the velocity fluctuations. Then, is it inappropiate to use RANS (Reynolds Averaged Navier-Stokes) equations in turbomachinery transient problems?

and, what about the phase averaging in transient periodic flows?

thank you

I don't know what you mean here the "maximum periode of the velocity fluctuation".

It should be that the time step that you choose must be much more greater that the periode of the turbulence when you want to use RANS equations. When this can not be fulfilled, you should find other ways to solve it.

Thank you Wang,

what I meant with "velocity fluctuations" was "velocity fluctuations due to the turbulence" or "turbulent velocity fluctuations", as you prefer. Anyway, the fact is that using RANS in transient calculation of a turbomachine, i.e. a pump, quite accurate values of it's head gain and performance are often obtained, even with a time step as small as 1e-3 sec. Isn't it surprising?

One point is the time scale of turbulence. RANS assumes that the frequency of turbulence fluctuation can be averaged. The most important, i think, would be a characteristic time step, e.g. the time that the flow passes the airfoil. It would be practical when there are several hundreds of steps in such a time,

The other point is, you got "accurate" values. It depends on your case, which turbulence model you have used, with or without separation, and many other issues. In fact, turbulence model is such a delicate thing that we must pay more attention to.