floating point exception in turbulent model

marco

dear cfd gurus, i'm running an analysis on a flow in a duct with diameter and angle variations; for the analysis i'm using k-omega turbulence and taking cavitation into consideration. After changing the model and outlet conditions from A (A:model ends at the exit surface of the duct with given pressure at this surface) to B (B:model ends with spherical camera after the exit surface of the duct and pressure is given on the surface of the camera) starcd started displaying unknown 'floating point ecxeption' errors due to possible initialization or meshing imprecisions. I've refined the camera mesh and tried to smooth conditions by scaling the inlet mass flow and giving appropriate initial conditions, furthermore the errors come out not only at the first iterations but even after few hundreds iterations. what could be the problem?