Inconsistent results in cyclone modeling
 

Hi all,

for my graduation project I have to model a gas cyclone. The first step i'm taking is validating a model (=closure model, boundary conditions and other pre-settings) by two reference cyclones. In velocity measurements on these reference cyclones, a decrease in axial velocity at the core of the cyclone is observed. In my validation, I have managed to model this for one of the two cyclones, but for the other one it seems to be impossible. Even if exactly the same settings are used as for the other cyclone, the velocity decrease at the core is not observed. Does this sound familiar to anyone, or is there anyone who thinks he might know the cause for this strange behavior?

My settings are:
Transient simulation
Total time: 0,5s (but I also have tried longer)
Timestep: 0,005s
Fluid: Air at 298K
Closure model: SSG Reynolds Stress
Wall boundary: No slip
Inlet boundary: Normal velocity
Top outlet boundary: Opening with relative pressure and flow 'Normal to boundary condition'
Bottom outlet boundary: Closed, no slip
Advection scheme: High resolution
Transient scheme: Second Order Backward Euler
Residual target: 10^-4
Mesh: >10^6 elements with 5 inflation layers (y+ is in the range 20-200 for the complete domain)

regards,
Boris

Are you using the high resolution turbulence numerics? This is a new option in V12.

You may need an LES approach to get the core correct.

I'm using version 11, so I cannot use new options from version 12.

I don't think LES is necessary, because the core is modeled as it should be for one of the cyclones. Strange thing is that it doesn't work for the other one.

Well it sounds like a good excuse to upgrade. The high resolution turbulence numerics in V12 may help.

Maybe the misbehaving cyclone needs LES but the others are OK with RANS? I am no expert on cyclones but from what I have seen the vortex core is frequently transient and that will be a challenge for a RANS model (even Reynolds Stress ones). I have seen many people using a LES approach to get the vortex core motion to address this problem.

Hi Glenn,

thank you for your replies.

The problem seems to be in the time step I used. Since the misbehaving cyclone has a higher inlet velocity, it requires a smaller timestep to capture all core effects.

regards,
Boris