Learning Forum

[Leonardo Provana]

Hello everyone,

I'm running a RANS steady-state simulation of a centrifugal compressor stage (impeller + vaned diffuser) by means of Ansys CFX. Between the rotor and the diffuser, I'm imposing a mixing plane. However, I'm experiencing a weird behavior of the Mach number on the mixing plane: at the 95% of the bladespan, it shows a stripe where it's null. I noticed that this stripe is at the boundary between the jet (flow with high momentum, directed towards the diffuser) and the wake (flow with low momentum, reversed towards the impeller outlet). My idea is that in this region the mass flow rate is null and it has a detrimental effect on the averaging process of the mixing plane.

Is there anybody who met the same problem?

[rfblumen]

Although it's appropriate in most cases to use a Stage interface for a Multiple Reference Frame (MFR) case like this, Stage can have issues if recirculation occurs across the interface, particularly if the the recirculation occurs across Stage bands.  That is likely what you're experiencing.  Some possible work-arounds:

-Try moving the interface, if possible, so that it's away from the recirculation.  We typically recommend placing the Stage interface halfway between the upstream trailing edge and the downstream leading edge.  However, if the location can be shifted such that it's biased torward either the upstream trailing edge or the downstream trailing edge to avoid/minimize the recirculation, I would try that.

-Using Frozen Rotor will avoid these issues occurring at the interface, but can generate other artifacts related to the way the rotor and stator are "locked together" at a particular orientation.  It may be worth trying to see how the solution looks.

-The most accurate approach, and one that will avoid this interface issue, is to run transient with a Transient Rotor Stator interface.  This will be a bit more expensive computationally but may be the best option in this case.  You might consider using also TBR methods.

[Leonardo Provana]

Dear rfblumen,

Thanks for your reply, it was really complete. I have a few answers/questions/doubts:

• Unluckily, the recirculation affects the entire vaneless diffuser, therefore moving the interface can't really help.
• About the Frozen Rotor, should the pitch be the same between vaned diffuser and impeller?
• For sure the TBR method is the best for this kind of problem. I was trying to avoid it because of the computational cost, but if it's unavoidable I will go for it.

Thanks once again for your help