Hi, I am simulating the process of the initiation of the fluid flow in a container. The fluid is mixed by a rotor with eight blades. Since the flow is not fully developed, I wonder if it is appropriate to use Transient Rotor-Stator interface.

Anyone can tell me how to model the process of the initiation and development of the flow in the container?

Thanks,

Fing

Hi Fing,

It sounds appropriate to use TRS. Start off with the rotating domain with a rotating velocity of zero and increase according to how fast you want to accelerate the impeller. This should work.

Glenn

Hi Fing,

The TRS sometimes need very good initialization otherwise, you may sometimes see "ERROR #xxxxx has occurred in subroutine FINMES".

The way i do this is that i start with the Frozen rotor to provide me with initial conditions, sometimes even to give me a developed flow field. Then swith to TRS.

Else, I start with what ever impeller speed I need (but it is a good idea to start with a low speed) and set "Transient initialization override" to T.

Regards, anne

Hi Anne,

In this case, as it seems the impeller is starting from zero velocity, an initial condition of zero velocity everywhere sounds sensible and should not give the solver any problems. Hence here I don't think there will be a need to do a frozen rotor simulation for an initial condition or to use the Transient initialization override expert parameter.

Regards, Glenn

Thank you Glenn and Anne for your response. I did use Transient R/S and unfortunately it seems to me that the start-up flow can not be captured by this method as I got almost uniform velocity (in STN frame) for the points at constant radius. It was expected to obtain a non-uniform velocity and shear strain rate and the shear strain rate and velocity around the blades should be greater than that far from the blades.

I also tried Moving Wall method and I got a extremely non-uniform flow field. The velocity and shear strain rate close to the blades are much greater than somewhere else. I just wonder anybody can tell me which method is best for simulating the transient start-up problem. Since there is application of moving wall boundary condition in Tutorials, I really can not decide which one should be chosen and what is the criterion/reasons for people to choose the moving wall boundary method over trans. rotor-stator interface method.

Thanks,

Fing

Hi Fing,

I am sure TRS can model what you are trying to do. Are you sure you specified the wall velocities relative to the correct frame of reference? Walls which are stationary in the absolute frame of reference, but are in the rotating domain need to be specified as "counter-rotating" walls.

I am not quite sure what you mean by the "moving wall method", but if you mean using a stationary domain and getting the walls to rotate then you will not be including rotational effects on the flow, such as centripetal and coriolis effects.

I suspect TRS will be the way to go, and the problems you are having are due to insufficient mesh resolution, timestep resolution, convergence or some other problem.

Glenn