There are different mean to define different conditions on different domains. For this case I suggest to add a second Initial Value node Heat Transfer interface. The define the selection where you want to apply it. Thanks to that you can use 2 different expressions to define the initial conditions. ]]>

The nonisothermal flow node couples the heat and the flow interfaces and provides options to account for viscous dissipation and pressure work. It is possible to define these couplings manually but using the predefined coupling is simpler. In addition the nonisothermal flow node tunes the stabilization of the flow and the heat interfaces for the coupling. Finally, it defines the thermal wall functions when wall functions are used by the turbulent flow model. ]]>

In COMSOL Multiphysics v5.2a we have a predefined option to include gravity in the single phase flow interface. You just have to select this option to model natural convection in a non-isothermal flow. ]]>

i try to modling natural conviction in power transformer but i have probleme with gravity , please i need your helpe ]]>

Both interfaces can be used (they offer same functionalities but have different default settings). For your application I recommend to use the Conjugate Heat Transfer multiphysics interface. In COMSOL Multiphysics 5.2 you can choose between following turbulence models:

- Algebraic yPlus

- L-VEL

- k-ε

- k-ω

- SST

- Low Re – k-ε

- Spalart-Allmaras

I am trying to study time dependent heat transfer b/w hot water and solid material. flow is turbulent i am confused which module should be used (conjugate or non isothermal) and need guidance. can anyone help.

]]>The Conjugate Heat Transfer interface available in the Heat Transfer module and CFD module is suitable to model oscillating non-isothermal flow past a heated cylinder. ]]>

In COMSOL Multiphysics 5.1, it is possible to simulate porous media flow using the Conjugate Heat Transfer interface. It’s definitely a good option. ]]>