how to set Boussinesq model for homogeneous binary mixture
 

I am using HBM for solidification with natural convection. Instead of the full buoyancy model automatically set by CFX, I'd like to use Boussinesq model for the liquid but I couldn't find the way to activate this option because CFX seems to treat the HBM as an individual fluid. I've defined the HBM as a mixture of liquid and solid phases for the same material, specified the single HBM as the fluid in the domain, and added momentum source (sink) to prevent the solid component from moving in the fluid. Any suggestions are highly appreciated!

Sounds like you will get convergence problems with this. Also note you don't include any phase change physics (sub cooling, heat release etc) so be careful that this model is appropriate.

Two questions:
1) If you have a momentum source to stop the solid moving then why will it move under buoyancy? Won't the momentum source stop the motion as well?
2) An alternate option would be to not use the buoyancy model and do it yourself with a source term. You could make the source term dependent on the material so if it is solid the buoyancy source term is not applied.

Thanks Glenn for your advice!

1) I am trying to model the buoyancy in the liquid phase. I already specified the liquid density as a function of expansion coeff. but the flow velocity is still too low and the liquid temperature is way below the melting point.
2) You are right. This is what I am thinking of to resolve this model setup problem because I have been trying both GUI and CCL to activate Boussinesq but none is working.

Here is a brief description of my case for your comments/instructions:
The liquid phase component (liquid metal) is applied with an internal volumetric heat source and cooled at the vessel boundary where solidification would occur.

First, I did single phase simulation for the liquid with assumed solid layer thickness in the model. Boussinesq was used and the resultant liquid temperature was quite uniform throughout the domain due to buoyancy. The results are also reasonable comparing to existing data.

Then, I defined the HBM and selected it in the fluid list. In this case, the buoyancy difference model was automatically selected by CFX with the specification of re. density. The results, however, always showed very low temperature below the melting point and the whole domain was filled with the solid phase component, no matter if the momentum source (sink) is added or not, and how I adjusted the latent heat.

It seems to me only the buoyancy due to density difference between the two phases is accounted for, rather than that in the liquid phase alone. That is why I'd like to specifiy Boussinesq for the liquid phase, similar to the single phase case I have done. Another reason might be the momentum sink, which could be too big to stop the liquid flow.

Any comments are highly appreciated!

I am not sure your approach is valid at all. The HBM model does not account for latent heat so growth rate of the solidification front will be totally wrong. I would talk to CFX support about solidification modelling.