I tried that, as well as intermediate values (multiply the diffusion coefficient by 10^-1, 10^-2 or 10^-3) but in all cases the simulation blows up after 50-70 iterations, no matter what is the value of the time step.nAnother aspect that I would like to clarify is that I have checked that the UDRGM is well implemented, as I have compared the result given by my model with the ones given by ANSYS Fluent in some cases of single-phase flow, as well as experimental results of two-phase flows without energy equation. Of course, there can still be a typo somewhere, but I have checked all the formulas many times.n