Which model can I choose to model natural convection?
 

Wen modelling natral covection,I use the models of SST k-w or realizable k-w,but all not converged.Can I use the model of Transition SST?Which model is the best choice?

Hello hotboy,
if you are simulating a closed system it's difficult to converge a natural convection phenomena: anyway I suggest using RNG-k-epsilon turbulence model.
In case of open domain you can use another turbulence model.
If you want to analyze the transition zone you have to activate some advanced models like 4-equations but it's not so easy to set-up (refers to Fluent Manual).

Best regards

I was simulating a closed system, it's really difficult to converged.

Bubt when I use the standard-k-epsilon turbulence model.It is converged .
Can I use the standard-k-epsilon turbulence model?Many people suggest that we can choose the model of RNG-k-epsilon ro sst-k-omega.

My geometry is complex when meshing and I do not have the BL mesh.Is the BL mesh very important ?

Hello hotboy,
standard k-epsilon is quite good if you haven't a complex geometry that involves recirculation, swirls, separation etc.
In heat transfer systems a BL control is suggested, of course: if you have problem in modelling a fine mesh you can accept a coarser mesh (without BL) and use wall functions in order to simulate your system accurately.
Another tip: usually natural convection in closed domain requires an unsteady solver to get convergence.

Best Regards