Transient RANS Simulation
 

Hello,

to this point I have only worked with steady state RANS Simulation. I have found a thesis where a transient RANS Simulation was described including animations over time.

As far as I understood the RANS Simulation will give us averaged values so the animation for a transient RANS Simulation doesn't make sense to me. Am I correct in assuming that it just shows physical nonsense till the mean values do not vary anymore (respectively the residual target are met) and afterwards the animation will not change anymore?

Thanks for your help!

Think about the compression/expansion flow cycle in an engine chamber. At each time (or crank angle), the URANS provides the solution that has the meaning of representing an "ensemble average" of N solutions at that time.
The animation does not represent the real physics you would observe in an experiment (or in DNS).

Thank you very much for your reply!

So what the URANS does is it gives me an average solution for every crank angle. I get the idea for a changing geometry like in an engine chamber.

But, when I have a geometry that isn't changing - let's say a flow over a cube - there is no reason to use a URANS.

Am I correct in my thinking?

Thanks again for you help :)

the URANS is based on a particular decomposition, the averaged field is time dependent in terms of external forcing frequency (as in the engine) or low frequency as vortex shedding behind a bluff body where the main frequency can be identified. Again, the animation does not represent a point-wise time-evolution of the flow

What is the exact purpose of an URANS then? I was under the impression, that when a steady state RANS struggles to converge or oscillates this indicates transient effects and one switches to transient formulation (=URANS) and get's a physical meaningful, time dependant result. Or asked differently: Does a simulation which contains unsteady, transient phenomena need to be simulated as at least Large Eddy Simulation to get meaningful results? As always in CFD, still a lot that can cause confusion..

Yes, I agree that a RANS that does not drive the residuals to vanish has somehow a URANS-like behaviour. This is because URANS is nothing else that a RANS where the time derivatives are added. Actually. there is no rational theoretical justification for URANS... The fact that the statistically averaging is time-dependent instead of being rigorously steady is justified by the presence of external time-dependent forcing so that the solution at each instant is the results of the averaging of (theoretical) infinite events. Without a clear external force I don't see a clear URANS definition. However, the turbulence modelling is practically the same for RANS and URANS. Somehow, URANS could be also presented as a form of time-filtered LES but this is a complex issue...