Simulating Buoyancy
 

When simulating an air flow problem (with heating), I understand that I could include buoyancy in the problem through 1 of 2 ways: (1) use ideal gas and gravity or (2) use constant properties and Boussinesq approximation.

For my problem, I am using (2) and am finding that my flow field is not changing at all when I change temperatures through the model. The residuals are not even readjusting for the flow equations... I would think that changes to the energy solution should bring about changes also to the flow solution.

I have also tried (1) and always seem to find that this causes my flow to dive in the negative z-direction near the pressure outlet (of a pipe, for example).

Can anyone give me any insight as to why my flow residuals are not changing? Or why (1) gives me such poor results.

Thanks!
Dylan

for the method 2 using constant properties and boussinesq approximation you need to
- add gravity or there will be no buoyancy force
- ensure you set the thermal expansion coefficient since for air this property defaults to zero so the air does not expand with temperature

if these are not set then nothing drives any flow so there is nothing to solve and residuals wont drop since you already have a perfect solution of the flow

and you might also need to run in unsteady mode since buoyant flow can be very unstable

cant explain method 1 so you must have some other bad setting or boundary condition eg is gravity in the required direction etc

Thanks for that! I do have gravity turned on, and I thought that I had set the thermal expansion coefficient... but apparently that was the problem!

Dylan

Well, now that this has been fixed, I get a solution that has the same problem that (1) has. At the pressure outlet, there is backflow near the floor of the model and a very strong flow near the ceiling.

How should I be setting boundary conditions in this case?

Here is a sample of the type of model I am working with.

Attachment 31387

Your pressure condition should not be constant zero; the pressure should vary with height.