Timescale for rotating harware with long domains
 

Hello,

I am modeling an cavitating axial pump in water at 6000 rpm. The pump itself has very low blade angles and so I have included many diameters in length upstream and downstream to fully capture recirculation. The documentation suggests using a timescale approximately equal to the residence time of the fluid in the domain.

For my application the residence time in the domain is ~1.2 seconds, but the residence time along the blade (which is the area of interest), is only ~0.05 seconds.

I am getting an oscillation in the residuals when I try to converge the problem with cavitation.

What is a good timescale to use in this case? Shouls I use a timescale equal to the total residence time or a timescale closer to the residence time in teh bladed area? Since there is recirculation at the leading edge, the residence time for some fluid is even higher since it recirculates over and over again, should I capture this in my timescale faxctor somehow?

Thanks for any help.

There are some tips here: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But it is always very difficult to get cavitation models to converge to a steady state, because cavitation often is not steady state but pulses. In this case increasing the physical timestep may help but you will probably be forced to look at transient simulations.

Also the size of the cavitated region is often very small and the simulation well converged except at the cavitation bubble. You may have adequate convergence despite the residuals not achieving the target.