Lift & Turbulent Dispersion in Stirred Tank

ozgur

Hi all,

Interphase forces (Lift and VM) are difficult to converge for dispersed flows in stirred tanks as reported by G.L. Lane (PhD thesis, 2005) and N. Deen (PhD thesis, 2001).

I'm performing MRF and sliding mesh simulations, with both Eulerian and mixture multiphase models for an aerated stirred tank. A fixed bubble size of 3mm is assumed. Sparger is modeled via mass and momentum source terms. Free surface is modeled via de-gassing BC (i.e. symmetry condition for liquid, gas is free to leave the domain at its terminal velocity). Turbulence model is dispersed k-epsilon. Computational domain consists of only 1/6th of the full tank with symmetry BCs at the cross-section planes.

I have no problem with VM. However, including the lift force (C_L=0.5), simulations blow up. The same happens when turbulent dispersion model of Simonin et. al. is included (i.e. drift velocity terms included both in momentum and turbulence equations) in Eulerian simulations. No problem with turbulent dispersion in mixture model.

I'm getting errors of type: /(divide):invalid argument[1]:wrong type[not a number] and divergence detected in AMG solver:k

I've tried numerious ways to get it worked: -lowering underrelax. factors for k, epsilon, turb. viscosity, momentum and volume fraction -lowering gas flow rate -switching on interphase forces only after converged 2-phase solution reached -starting by low lift coeff. C_L, and increasing gradually step by step but no success!

Bubble column simulations run just smoothly with Lift and turbulent dispersion forces without any problem.

I don't expect that these forces have significant effects at near-impeller regions, however I like to see their effect at far-impeller regions. Is there anybody in this group who was more lucky than me with this type of simulations, or at least have a good idea to recommend?

Regards, Ozgur