[zuba]

Hello


I have a problem with a simulation of a plume (dry air 60% + CO2 40%) into the atmosphere. I am running into problems partly because of my lack of experience with CFD and compressible fluids in particular. The model includes a wind of 1-15 m/s and a non-flat topography in the bottom. I use rhoReactingBuoyantFoam to simulate it without reactions and with the combustion off. I do so because I need a gas concentration varying with the time in each cell. The solver computes p_rgh instead of p. The simulation runs until the end (1000 s), but the pressure increases from 100kPa until ~200 kPa (2 atm). I would like to avoid this drift of the pressure, but I cannot figure out how or if it is possible even. I have tried optimizing the boundary conditions without any effect. I also tried a "brute force" approach where p_rgh is shifted so that its value is pRefValue at a reference cell in the end of each calculation loop. I do so adding the following line:

p_rgh += pRefValue - p_rgh0

pRefValue=dimensionedScalar with a reference value in Pa.
p_rgh0=dimensionedScalar with the value of p_rgh in a cell that I use as reference in the bottom of the calculation domain.

This last "trick" avoids the drift but it prevents the convergence of the outerloop of PIMPLE. This is not a problem for me because the "inner loops" converge well and I am only interested in the stationary state. However, I do not understand why does not converge. The only problem I could find in the output log is with the continuity error. The cumulative error goes to -6 very fast.

Does anybody know why the "brute force" method breaks the convergence of the outer loop of PIMPLE? The Navier-Stokes equation has only derivatives of the pressure and a constant should not change the equation.


Does anybody have any suggestion?

Thank you