[sbesarati]

Dear friends,

I am working on heat transfer simulation of a tank filled with liquid salt using Fluent. The tank is cylindrical and there are four cylindrical heaters penetrating from the side walls to the tank (perpendicular to the side walls). The size of the heaters are much smaller than the size of the tank. I am trying to model natural convection inside the system. Only 1/4 of the geometry is modeled and periodic boundary condition is used to simulated the remaining 3/4 of the geometry.

The question I have is that for calculating the Grashof number which characteristic length should I use? If the height of the side wall is chosen, the flow is turbulent. Another approach is to only use the diameter of the cylindrical heater, which results in a laminar flow. I really appreciate if you could please help me in this regard.

Thanks

[FMDenaro]

Originally Posted by sbesarati

Dear friends,

I am working on heat transfer simulation of a tank filled with liquid salt using Fluent. The tank is cylindrical and there are four cylindrical heaters penetrating from the side walls to the tank (perpendicular to the side walls). The size of the heaters are much smaller than the size of the tank. I am trying to model natural convection inside the system. Only 1/4 of the geometry is modeled and periodic boundary condition is used to simulated the remaining 3/4 of the geometry.

The question I have is that for calculating the Grashof number which characteristic length should I use? If the height of the side wall is chosen, the flow is turbulent. Another approach is to only use the diameter of the cylindrical heater, which results in a laminar flow. I really appreciate if you could please help me in this regard.

Thanks

What you are evaluating is correct, two different values for the Gr numbers that depend on two different lenghts. Both are correct. That said simply that you must consider that laminar, transitional and turbulent flow condition can cohexist in your flow problem.

[sbesarati]

Thanks for your reply.

Then, the question is which model I should use to simulate this problem, laminar or turbulent?

[FMDenaro]

Originally Posted by sbesarati

Thanks for your reply.

Then, the question is which model I should use to simulate this problem, laminar or turbulent?

laminar is not a model....

think first about the best formulation you can adopt for the physics of your problem (DNS/LES/RANS)