[Julian121]

Hi All,

I have recently started simulating the surge and stall of an axial compressor by CFX. In particluar, I need to simulate a vane recessed casing treatment and then compare the results with the experimental data in the literature. I need to show that the casing treatment would improve the stall margin and possibly the efficiency. I am familiar with Fluent but have never worked with CFX.
I wonder if anyone who has simulated a similar case in CFX, could give me some advice and share his experience.

Here are my questions:
1. What is the ideal point in simulating the stall/surge? Should a whole stage be modeled or just a blade is enough?
2. I have read in the CFX tutorials that rotor/stator can be modeled as frozen rotor. So, should the rotor be frozen in the simulations? I do not understand how this assumption can be true.
3. How far can the simulations be done? I mean, is it possible to have the simulations after the stall point or it is not possible after the stall?
4. And finally, is the assumption of unsteady flow true? or surge and stall should be modeled as steady flow? Why?

Thanks in advance!

[ghorrocks]

Have you read the bets practises guide in the CFX documentation reference manual? They describe a lot of more detailed information about how to go about some of these sort of simulations.

[Julian121]

Originally Posted by ghorrocks

Have you read the bets practises guide in the CFX documentation reference manual? They describe a lot of more detailed information about how to go about some of these sort of simulations.

Thank you for your reply. Yes, I have read that. However, it is not very detailed. Has anyone ever simulated the stall and surge of an axial compressor?

[ghorrocks]

I am no expert on turbomachinery but I will try to answer your questions:
1) If the stall occurs on all blades then a single blade is OK. If the stall varies around the machine then consider a whole stage. It really comes down to whether the assumption of periodicity is valid.
2) The frozen rotor approach includes most of the rotational forces but on a stationary mesh to make the simulation easier. Read the thoery documentation on frozen rotor so you understand the implications of the model.
3) Yes, you can model after stall. But often the assumption of a 2-eqn turbulence model is no longer very accurate. You may need LES, DES or SAS or some other more sophisticated model.
4) You almost certainly have to model surge and stall with transient models. See answer 3).

[Julian121]

Originally Posted by ghorrocks

I am no expert on turbomachinery but I will try to answer your questions:
1) If the stall occurs on all blades then a single blade is OK. If the stall varies around the machine then consider a whole stage. It really comes down to whether the assumption of periodicity is valid.
2) The frozen rotor approach includes most of the rotational forces but on a stationary mesh to make the simulation easier. Read the thoery documentation on frozen rotor so you understand the implications of the model.
3) Yes, you can model after stall. But often the assumption of a 2-eqn turbulence model is no longer very accurate. You may need LES, DES or SAS or some other more sophisticated model.
4) You almost certainly have to model surge and stall with transient models. See answer 3).

Thank you for your thorough reply.