Learning Forum

[Ehsan]

Hi,
My question is long but I will appreciate considering it.

I want to simulate a mixing tank filled with water, air, and glass beads in Ansys Flunet. The mixing tank is a coaxial mixer comprised of a central impeller, a wall scraping anchor, and a gas sparger. The fluid followa non-Newtonian fluids law (power law) and particles are patched at the bottom of the tank (5%-15%), and I want to model the solid suspention.
I am using three-phase Eulerian model:

1. Water: Continuous phase
2. Glass particles: Dispersed phase (Granular- constant diameter: 1mm, Granular viscosity: Gidaspow, Frictional viscosity: Schaeffer, Frictional pressure: based KTGF)
3. Air: Dispersed phase (constant diameter: 3mm)

Since I want to model solid suspension, Turbulence dispersion is important, So, I used Simonin model for Turbulence disperion and Schiller-Naumann coupled with Brucato modification for Drag coeeficient in all phases interactions.  Also, I used dispersed RNG k-epsilon turbulence model, transient solver, 0.0001s time step, gas velocity inlet 0.1 m/s, outlet as outflow, central impeller 300rpm, anchor impeller 10 rpm and sliding mesh method.

The problem I have is that some of my residuals (u-air, u-solid, v-solid, w-solid, w-fluid ) are not getting converge and are constant (actually they show a little divergence). I had continuity divergence at the beginning as well. But I solved it by mesh refinement, using coupled scheme, and reducing the relaxtion factors for pressure and momentum to 0.5. Moreover, I encounter backflow when the simulation goes on. 

I have some guesses myself:

1. KTGF model is better for transient fluids, however at the beginning of the mixing, the flow is not fully turbulent. Moreover, I did a couple of simulations without considering Frictional viscosity for solid phase and I saw weak solid suspension. 
2. I could not use degassing boundry condition since it is a 3-phase simulation and degassing is only availabe for 2 phase simulation. So, I had to consider back flow in boundry condition and that is why I am getting back flow in my consul as well. 

I don not think that my mesh needs modifications since I have worked on it a lot.

Any recommendations? 

 

[Rob]

I can't see where the inlet/outlet are. I'm assuming the outlet is a pressure outlet, do not use the older outflow condition. Convergence is poor, but I also don't know how the cell size is relative to the flow speed and time step: you may need a smaller time step. Have you used an inflation mesh? If so, how does the cell size compare to the bubble & particle size? What is the inflation aspect ratio like, and how does the flow change along a cell? How do the monitors look, and please post some images of the results.