[SjBos]

Hi,

I'm trying to simulate Stairmand cyclone in Fluent with cyclone diameter of 0.29 m. Some info on my setup:

Mesh and domain:
- Polyhedral mesh with inflation layers. First element height equal to roughly 1.5e-3 [m] to have y+ > 30, and use wall functions.
- Extended inlets and outlets by length of 1D.
- Different mesh refinements 100,000 to 500,000 cells.

BCs:
- Velocity inlet of 10 m/s at inlet.
- Two pressure outlets 0 Pa.

Simulation:
- Pseudo transient: time scale factor around 0.1 - 0.5.
- Coupled pressure-velocity
- Turbulence, start with SST curvature correction. After some initial convergence switch to RSM (linear pressure-strain)

Problems:
- The RSM improves the results and gets it closer to experimental data. However, in the center, the velocity profiles mainly show deviations in the center. I thought mesh refinement would improve it. But the max tangential velocity remains being lower than experimental data. Additionally, the convergence gets really bad and further refinement results in no convergence at all.
- I'm going to time scale factors of like 0.01 just to somehow improve convergence. This can't be right, is it?
- Also, is reverse flow over outlet common in cyclone simulations?

I'm wondering what could be my next steps:
- Switch to low y+ approach (<1) and stop using wall functions?
- Different simulation setup?
- Would it be related to my mesh?

Anyone experienced in cyclone simulations?

Thank you in advance!