Forcite MD with Constraints

Hi All. I tried doing a Forcite NPT run with constraints. I got to know that in the current implementation of the code one cannot have a barostat with some fixed particle coordinates (x,y,z fixed). This makes sense, but I thought fixing fractional coordinates (for a set of atoms in the cell) is in principle something that is doable with a barostat, so I tried it. It seems that the code starts without any complaints (in contrast, the code actually complains in the previous case of fixing absolute coordinates), but then strangely, the fractional coordinates actually do change during the MD run, as if Forcite just ignores the fixed fractional coordinates constraint. Does anybody know if it is actually possible to fix some fractional coordinates (for a set of atoms) in a Forcite MD run?
P.S. the barostat I use is Souza-Martins.