I found the reason for this problem. The MF2k-model was made by means of a source/sinks-coverage (MAP) with GHB and RIV-package. To apply saltconcentrations I duplicated this coverage, deleted all GHB/RIV-arcs but two to which salt concentrations were written. Result: crashing Seawat, but no error indications whatsoever. Solution: no additional coverage, but changing the two arcs into CHD and apply the source-concentration to these two arcs. Result: running Seawat.
In GMS 9.0.2 I have a good running modlfow and mt3d simulation. Seawat also works fine as long as the VDF is not activated, but that is not what it is made for. VDF, properly filled in with a minimum density and a slope (delta_density/delta_conc) with consistent units, the simation would not run for 1 split second. It crashes immediately. Does anyone have som advice?.