Natural convection in liquid metal column
 

I am thinking about how to model the following problem:
I have a cylinder, filled with mercury. The cylinder walls are electrically isolating - from the bottom to the top an electric current is flowing. The bottom is cooled, the side-walls are at room temperature. The current heats the metal to about room temp.+30K and natural convection occurs. The speed of the convection is the thing I would like to know.

-> buoyantBoussinesqPimpleFoam is the right model?
-> can I simulate with 2 different phases, too? (e. g. water over mercury)
-> can I simulate the heat transfer through the wall, or do I have to use a wall with a thickness == 0 at a constant temperature?

Many thanks!

hello,

1) yes buoyantBoussinesqPimpleFoam would be the right solver for natural convection, but you need to add source term (current heat): customise the solver or look at explicitSetValue.

2) multiphase not miscible => interFoam, but current "official" solver are isothermal only, so you need to modifiy interFoam (simple if you stay incompressible).

3) you can use both, i.e simulate heat transfert or even try the "shell" model (look here:http://www.cfd-online.com/Forums/ope...tml#post320936

hope this help,
regards,
olivier