[pchoopanya]

Dear folks,

I am running a 2D compressor cascade model. The conditions are as follows;

1. 7 blades aligned in a row with angle of attack of 30 degrees
2. Steady-state solver
3. Inlet velocity is low; velocity-inlet from left to right at 4 m/s (Reynolds number is below 1000)
3. Incompressible ideal gas (viscosity using Sutherland method) since max local Mach number is far below 0.3 (max local velocity is below 10 m/s)
4. Pressure-outlet
5. Spalart-Allmaras as turbulence model
6. My mesh at the surface of the blades is, I believe, sufficiently thin (I checked the y+ values at the blade surfaces and at two walls are well below 1)
7. The residuals are at very low value. Though the energy residual stays flat at around 10^-16
8. Other parameters used as vertix monitor are flat signifying convergence.

So, I believe, with the present mesh density, I have arrived at a fully converged solution. And my questions are...


1. Look at the velocity contour, the flow field between each blade passage show periodicity except the one at the bottom. I notice a very large re-circulation zone underneath the bottom blade. Why is this one much larger than the other? I have tried to think of the reasons or the physics explanation behind this but without success. Should someone shed some light on this for me?

2. I tried to reduce the number of blade, leaving more and more gap between the bottom-most blade and the bottom wall hoping that I would be able to eliminate the wall-effect. Still I find the re-circulation zone underneath the bottom-most blade very large compared to the others. I do not think I can leave a very large area between the bottom blade and the bottom wall since I will need to build a test bench out of this configuration. I need to keep this blade "cascade" configuration. Is this something to do with the wall? If so, how does this effect the result?

3. Is there something wrong with my result? Anyone has run the similar blade cascade simulation and found a different result?

4. I am using a steady-state solver. Could this large re-circulation zone at the bottom blade be due to it is a transient phenomenon? How do I decide which (steady or transient) solver I need to use?

5. Should I be worried about the energy residual? Though it is very low (10^-16) but it does not drop.

Any comments are welcome as this will help strengthen my understanding to my problem of interest. Thanks in advance.

[pchoopanya]

I have another question too,

Look at the big recirculation zone of the bottom blade, The zone extends beyond the outlet. However, I have managed to get a fully converged solution (judging the residual graph and monitoring variables) . My question is, it this acceptable? Should I extend the outlet section, of course enlarging the computational domain increasing number of mesh elements, so that the entire recirculation zone is entrapped inside my domain?

[pchoopanya]

I have tried another model, using 7 blade in a row.
Now I reduced the angle of attack to zero degrees.

As expect, there is no large re-circulation zone underneath the blades.

However, from the velocity contour. I also notice that the velocity in each blade passage are decreasing from top to bottom.

Especially, when one compares the velocity inside the blade passage of the top passage (between blade 7 and top wall) to the bottom passage (the first blade and the bottom wall)

Why is the velocity at the top passage is higher than the velocity at the bottom passage even the gap (flow cross-section area) at the top is larger than the bottom?

I recall the conservation of mass equation;

A1 x v1 = A2 x v2

(A is the cross sectional area; v - velocity, 1 - location 1, 2 - location 2)

Wouldn't the velocity at the top passage suppose to be lower as the gap is bigger?