Numerical artefacts?
 

Dear all, it's now a problem of a cylindrical tank, with a pump around. We know that in real life, when we are operating in the turbulent zone, the plume that is formed from the jet injected near the bottom of the tank, is some what stable and it rotates around a central mean location. We can explain it in terms of slight instabilities which may at any time disturb the plume from its normal position, and after that, the balancing forces makes the plume revolve around. However in the laminar zone, the plume is well behaved, and will not revolve round like this. Now the question is: CFD codes in the turbulent zone generally takes the time-averaged NavierStokes Equations in the turbulent zones, so that these disturbances are also time averaged, and their effects do not show up as disturbances. But if we get such behavior from simulation in FLUENT also, can we believe them as actually solved by the solver, or are they just numerical artefacts, I mean th rotation of the plume is done by the numerical disturbances? Please note that I am getting the usual well behaved results in the laminar zone, but a rotating plume in the turbulent zone using a k-e model.

Re: Numerical artefacts?
 

I don't know an answer to this question, but to my opinion the problem is the seperation of time scales in the flow field. The averaging time of reynolds equation must be long enough to smooth out the turbulent fluctuations (life time of turbulent eddies) but short enough not to average out the unsteadiness of the mean flow field. But how to takle Your situation, where the time scales of the mean flow field and the turbulent fluctuations in the plume sem to be of similar size?