Negative sand concentration
In t3dmix_tridiagonal_settling.h, when the number of substep is defined, a function NINT is used. This choice could lead to unstable concentration of sand .
A fix could be to use the function CEILING instead of NINT in https://gitlab.inria.fr/croco-ocean/croco/-/blob/dev_2022_mustang/MUSTANG/t3dmix_tridiagonal_settling.h#L126
Exemple to test : 1DV case without current, only settling of sediment. Use the attached file to compile and run (.pbs for datarmor, env variable $croco to adjust ) If you plot the vertical profile of sand concentration at time step number 20 you will see the instability like on the image