CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM Running, Solving & CFD (
-   -   InterFoam with boussinesq approximation (

sinusmontis March 28, 2007 09:35

Hello, I've been trying to

I've been trying to implement the boussinesq approximation into two phase solver interFoam. I have implemented density as a temperature dependent function in the twoPhaseMixture class and added
+ fvm::div(phi,T)
- fvm::laplacian(DT, T)

as the "energy" equation to be solved within the PISO-loop.

As the test case I use a right circular cylinder half filled with the liquid as a axis symmetric simulation (same setup has been used for isothermal simulation and working perfect). Gravity is acting along the symmetry axis and the contact between the liquid and wall is 90, so the liquid is at rest and no motion is induced by capillary forces.

I'm encountering a increase of the liquid fraction during the simulation when the wall that touches the contact line heats up the liquid.
To verify the cause of the phenomena I did some tests. When gravity is turned off and only diffusive terms play a role the liquid/gas ratio stays constant and one can observe the diffusive temperature transport in the liquid and gas. With gravity turned on and only heating the part of the wall that is in contact with the gas produces nice convection/diffusion transport in the gas phase and no change in the liquid/gas ratio is observed.

I'm guessing that the additional density gradient near the surface accounts for additional buoyancy with is not balanced by sigma*k (surface tension * curvature) which causes some flux normal to the liquid/gas interface which is then adjusted by the gamma equation, causing extra liquid volume.

I've tried to adjust the energy equation to account for no convective fluxes across the liquid/gas interface in same manner as in the gamma equation, but without any luck.

Unfortunately I'm out of ideas how to overcome this problem. Did anyone already implement two phase flow with boussinesq approximation?

I would real appreciate if anyone could give me a hint where to look at or point me to some papers that discuss the implementation of the boussinesq approximation in a two phase VOF-code.



All times are GMT -4. The time now is 19:48.