Hello everyone,

I'm kind of confused with setting up boundary conditions for a simple natural convection problem. The situation is that a copper cube (which is a volumetric heat source)is cooled by surrounding air (can be considered infinite, and ambient temperature is 25 degrees C) through natural convection. In Gambit, I modeled the copper cube together with the air. For the air domain, I used a big cube which is 64 times bigger than the copper cube. Now my question is what boundary types or conditions to assign on the boundaries of the air domain. Can anyone kindly explain to me?

Thank you,

Ken

1) All six faces ( or whicever face air is available to enter/leave the domain) in that air volume will have pressure BC ( air can enter or leave the domain) 2) Assign temperature of free air 3) setup gravity

If you have not assigned any face with pressure BC and you leave it then Gambit will assign it as wall ( no fluid can pass through)

velocity will be generated as the heat gets tranefrred to the air...

Good luck... AJ

Hello AJ,

Thank you first for your answers. They're clear. Now I think my concern boils down to point (1) you mentioned. Could you explain how to determine which faces are pressure inlet and which are pressure outlet depending on real situation? For the problem I stated in the previous message, I'm not quite sure.

Ken

You could use velocity inlet for one face, and pressure far field BC for other faces.

Hello AAS,

Thank you for your help here. For that velocity inlet, what velocity should I assign? And does pressure far field BC mean the pressure at infinite distance?

Ken

Hello,

Actually I misinformed you. You should in fact use pressure farfield for all of your outer faces (and not velocity inlet for one face.) Do this only if the fluid you are modeling obeys the ideal gas law. With pressure far field you specify the free stream boundary values for static temperature, pressure, and also free stream Mach number. The fluent and gambit user manuals contain a wealth of information for problems such as this. As to the paramters you should input for this BC, that depends on the specific problem you are trying to solve.

Good luck A A S

Thank you, A A S. But are you saying the pressure farfield BC's do not apply to other methods like Boussinesq approximation?

Ken

Hello AAS,

I just tried using pressure farfield BC's with static temperature of 25C, gauge pressure of 0, and Mach number 0. And I used ideal gas law for density and set operating pressure to 101325 Pa. Gravity was defined in z direction with -9.8 m/s^2. But Fluent gave me "Error: Floating point error: invalid number". Do you have any idea why it is?

Thanks,

Ken

Once again, I apologize. I thought you were modeling forced convection (I misread your first post.) I urge you to consult the Fluent user manual for bouyancy driven flows. The pressure far field BC is useless in your case and pressure inlet and outlet BC is all you need.

Thank you, A A S

Yep! With pressure inlet/outlet BC's, Fluent is calculating now. But is there a physical reason for assigning one boundary as pressure inlet or outlet? In my problem, a heat generating copper block is surrounded by air. I'm not sure how to decide pressure inlet or outlet on every boundary of the air domain. Is it ok to assign inlet or outlet randomly?

Ken

If I were in your place, I would assign Pressure Inlet to the bottom of the big cube, Pressure Outlet to the top face and Symmetry for the lateral faces (all four). Actually, this problem could be easily reduced because of the symmetrical construction of the flow domain, so you should cut only, let's say, one quarter of the flow domain (one corner of the copper cube), and also assign Symmetry BC for all the vertical faces (except the cube's faces of course). The external boundaries should be placed at a sufficient distance from the copper cube, at least 10 times the cube's height, this way not influencing the flow near the cube.

Also, beware that if the temperatures on the cube are high compared to the surrounding air, Boussinesq approximation will be erroneus, so you better use ideal gas law.

All the best,

Razvan

Hello Razvan,

Thank you very much for your answers. But I just don't understand one thing. What's the reason for assigning Symmetry BC for all four lateral faces of the air domain (in the original un-simplified model)?

Regards,

Ken

Hi Ken,

The reason for this is very simple: in this type of simulation, the most difficult equation to converge is continuity, and to ease its convergence one has to make sure that the inflow and outflow boundaries are distinct and also that there is no reversed flow through these boundaries. One way to achieve this is to put the inflow and outflow boundaries as far as possible, and use symmetriy BCs for the other surfaces of the exterior boundary. Of course, this has to be done in such way that the analyzed phenomena will not be influenced, or at least in a very small percent.

In your specific case, it is perfectly justified to assume that the flow is naturally going from the bottom face to the top face, and the lateral faces have very little contribution, but I repeat, this is true only if the lateral boundaries are far enough from the region of interest.

All the best,

Razvan

Hello Razvan,

I got it. Thank you again.

Ken

[engahmed]

Hi Ken,
I know this problem has been long time ago but now I am facing the same problem, I am modelling natural convection from a heated horizontal pipe embedded in a box (2D), symmetric. So if you imagine, I am setting the bottom, and side as pressure inlet and the top as a pressure outlet. My question simplifies to what values of pressure do I enter for the inlets and outlet while setting my operating pressure as 101325 pa??
Thanx in advance.