[sirsammie72]

Hello,
I am simulating a cyclone separator and trying to match theoretical pressure drop values and achieve mesh independence - using Ansys Fluent.

I am using the static pressure value under volume integral to compute pressure drop in CFD.

The theoretical pressure drop for this cyclone is calculated to be 100 Pa. However, the CFD pressure drop is consistently around 10 Pa with any mesh size. I have double-checked units.

I have checked air flow properties, gravity, etc. The outlet velocity of the cyclone matches the theoretical predictions. It is just pressure drop that is off. Any thoughts on why this could be?

Thanks

[karachun]

Pressure Drop is difference in Total Pressures, not Static Pressure.

[sirsammie72]

I have revised my method and am now taking the difference in pressure at surface of inlet and outlet, using total pressure. However, the value is still quite low.

[karachun]

Do you perform mesh independence study?

[jsm]

Hello,

Cyclone separator is somewhat unique to solve and get the accurate results.

Can you answer following questions.

Is your mesh is hexa or tetra?
How much have you refined your mesh nearer to wall & inner core of cyclone separator?
Which turbulence model are you using?
Have you extended the pressure inlet & outlet sufficient length to avoid back flow effect?

Please let me know your answer for these questions and post geometry image if possible for better understanding.

[sirsammie72]

Hello, thank you for your response.

Mesh is tetra. We have not refined the mesh more near the wall and inner core, just mostly around the inlet. We are using RSM, and have enabled prevention of back-flow at the inlet.

Any tips for convergence with a finer mesh? We cannot reach convergence with fine mesh even with very small time step

[jsm]

Tetra mesh is NOT good choice for RSM model. It will give very poor convergence.
Remember that RSM model solves six Reynolds's stress equation for each stress tensor. As tetra mesh is not single direction oriented, it will create unnecessary artificial diffusion and mislead the solution.

I would recommend to do hexa mesh with highly refined boundary layers nearer to wall and refined core region of cyclone separator. Y+ should be close to one. As RSM turbulence model doesn't like variation in mesh size, keep very less mesh growth rate for better convergence.