[mchurchf]

To whom can answer,

I am trying to understand why the following is occurring:

I am using a version of buoyantBoussinesqPisoFoam modified so that it can be used with LES models. I have a case in which I am trying to solve for atmospheric boundary layer flow and have a domain that is 1000m tall. I initialized the solution at cell centers and boundary faces such that p = 0 at the ground and p = -9810 m^2/s^2 at the top of the domain with a linear variation in between. (Note that the p variable is pressure/rho, I am using gravity = -9.81m/s^2, and since this is incompressible, all that really matters is pressure gradient--the pressure level is arbitrary, so it is okay to have negative pressure.) The initial solution also has some small random perturbations in velocity near the surface. Temperature is set to constant throughout the domain, so no buoyancy-induced turbulence generation should occur. I am not using a wall function for these initial trials to get the solver running correctly, just no-slip (note, this isn't really correct for atmospheric flow). All lateral boundary conditions are periodic.

If I set the pressure boundary conditions to fixedGradient, the solution become laminar after some time. If I set the pressure boundary conditions to fixedValue at 0 and -9810 at bottom and top respectively, the solution becomes very turbulent after the same amount of time. I can use pisoFoam (no gravity) with zeroGradient boundary conditions and get the same laminar result as using fixedGradient with buoyantBoussinesqPisoFoam.

Any idea why the use of fixedGradient versus fixedValue on p would affect the solution so greatly?

Thanks,

Matt