[joaran]

Dear CFD Community,



I'm using OpenFoam 2.2.0, trying to solve the XiFoam tutorial (moriyoshiHomogeneous). The only file I modified is the fvSolution, where I changed the nOuterCorrectors to 200 and nCorrectors to 2, and added the ResidualControl and relaxationFactors subdicts. The file is attached. The problem is that in many time steps, all the fields' residuals start oscillating around a certain value, i.e. their mean values "flatten", and the PIMPLE algorithm does not converge to the tolerances I set for them (being the pressure field the most affected one, with the highest initial residuals). It doesn't seem to matter whether I refine the mesh, or coarsen it, or even change the time step or BCs. See attached a residual plot for the first time steps of the run. I also tried adjusting the relaxation factors for all the fields, each at a time, and altogether, and perhaps it converges well where it didn't before, but at some point it will show this same behaviour again.
I've visited many threads where this topic is discussed for different solvers, and everywhere the cause of the problem is pointed towards the quality of the mesh and/or appropiateness of BCs.

I don't consider it to be applicable for this case I'm bringing, because the mesh is just composed of squares, and there are no inlets or outlets!

I also noticed that other solvers using PIMPLE carry this "issue" too. Say, e.g. in Tobi's book*, 4th ed., figure 11.19. There, the second time step shows the same I'm experiencing. I also tried with reactingFoam, and the tutorial case of this solver also shows such issue.


Perhaps this behaviour is unavoidable, and it's something we must learn to live with. I'm not sure I'm comfortable though, ending the simulation with many time steps having initial residuals on the order of 10^-4 or 10^-3, or sometimes even 10^-2 just because I don't have the skills or knowledge to lessen them.



Any word of advice or help will be very much highly appreciated, folks!




Best,
Joaquín





*Tobias Holzmann. Mathematics, Numerics, Derivations and OpenFOAM(R), Holz-
mann CFD, Leoben, fourth edition, February 2017. URL www.holzmann-cfd.de.