[pchoopanya]

Dear CFD folks,

I am currently undertaking a 3D, steady-state simulation of a 19-stage axial compressor.

My approach is to utilise the axi-symmetry nature of the axial compressor hence I only need to mesh a single blade from each ring of blades (rotor and stator blades in one stage are meshed separately) and then add each blade one by one in the simulation through >>> Append >>> Mesh

I am using the mixing plane approach as demonstrated in one of the tutorials from ANSYS FLUENT.

However, I have just completed 4 stages which means there are 8 blades: R1, S1, R2, S2, R3, S3, R4, and S4 (R for rotor blade, and S for stator blade).

I then assembled all blades together, followed the guidance from Mixing Plane tutorial to set all the relevant parameters. My case does not converge. I played around with the settings about still no luck.

My questions are;

1. Could you provide me the best practices specifically for a multi-stage axial compressor simulation?

2. I notice a reversed flow occurring at mixing planes (inlet/outlet interfaces) and I believe this is the reason why the solution does not converge, am I correct? I've tried extending both inlet and outlet, both very short and very long, again, no luck. How can I fix this?

3. Since the compressor is to compress the air, meaning the static pressure must be higher at the outlet. This is why I set the R1 inlet pressure at atmospheric pressure (0 Pa) and S1 outlet at 73000 Pa. This value is the linear interpolation of a 19-stage axial compressor with compression ratio of 14. So, my rough calculation suggests the compressor would raise the static pressure 73000 Pa after each stage. Is this a good way to do this?

Does setting the inlet pressure at 0 Pa and a higher (positive) pressure at the outlet represent the realistic operation of a compressor?

In so doing, wouldn't this induce a reversed flow from outlet back to the inlet due to the pressure difference? Could you please suggest this, I am really confused.

4. Due to the complexity of the blade, I am unable to obtain good mesh with conformal mesh. Rather, the two side walls (they are rotationally/radially periodic walls) of the box containing the blade are meshed independently (non-conformal). How much extend does this contribute to the non-converging residuals? Seems like I only have to choose between, having a conformal mesh periodic interface but with highly skewed elements or non-conformal interface with better mesh quality. Which one is the best option? Or is there a better option from you?

5. Can you provide me more tutorials specifically designed towards turbomachineries/axial compressor/mixing plane simulations?

6. Regarding incompressible vs compressible flow, do I understand this correctly that since the compressor's job is to compress/squeeze the air. Then the flow must be compressible flow? I know one (rule of thumb) criterion is the mach number, if it is greater than 0.3M then I should switch to AIR (ideal gas)? What if the compressor squeezes the air but the flow velocity is low, below 0.3M? Can I still consider the air as compressible ideal gas?

From my experience, choosing constant density air gives better convergence rate but when switching to ideal gas the solution diverges.

Does the act of compressing/squeezing the air also dictate the choice of AIR as INCOMPRESSIBLE IDEAL GAS vs COMPRESSIBLE IDEAL GAS?

7. What is the best turbulence model for an axial compressor? Spalart-Almaras or k-epsilon, or else?

Could you please help me clarify my queries and any comments/suggestions would be much appreciated. Thank you very much.


Pat