Last months, I designed a high-speed centrifugal pump that is composed by an inducer +
centrifugal pump + volute + conical diffuser in this flow sequence.
It rotates 30000rpm. This is very common in turbopumps of Liquid Rocket Engines.
In my case I just have to simulate the centrifugal pump. Of course it is not the ideal solution. It must simulate the inducer together because the rotational component of velocity before the inlet of centrifugal pump. Any way I did the wrong way. The centrifugal pump was designed considering the component velocity.
In this simulation using Fluent v6.3, I considered the flow axially the inlet plane of centrifugal pump and without volute and conical diffuser, just an extension at inlet and outlet to improve the convergence of solution.
I know the volumetric flow and inlet pressure and the required outlet pressure at design condition.
My boundary conditions were inlet pressure at inlet and outlet pressure at outlet. Instead of it, if I considered mass flow inlet, the solution gave me wrong inlet pressure.
I got good convergence solution using steady, k-epsilon, SIMPLE, double-precision and hybrid mesh (boundary layer + tet) for large volumetric flow (overload conditon). Decreasing the volumetric flow (the Fluent gave the mass flow rate) I could not obtain convergence of solution. The pressure fluctuates too much. I think of that is the presence of unsteady phenomena rotating stall.
So if anyone can help, my questions are:
I think you cannot avoid inducer from the problem, since avoiding it will give wrong boundary condition to rest of CV.
In my opinion right boundary condition would be mass flow rate at inlet and pressure at outlet.
You need not go for unsteady formulation , instead you can try steady state in 'Moving Reference Frame' technique.
|All times are GMT -4. The time now is 17:21.|