open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello!

I just want some advice on the selection of pre-conditioning options. I noticed that my free surface results show differences of ~0.08m at the inlet for the same run when altering preconditioning on the diffusion for k-epsilon , where one used the default option 2 (diagonal preconditioning) and the other used option 17 (exact solution along each vertical direction.)

Clearly the selection is important, but what criteria can you use to make sure you select the right one? Is one better for accuracy or case-dependent? For example, option 17 seems to be a lot more sensitive to changes. I couldn't find anything on the forum that suggested how to guage the right choice.

Many thanks!