I am trying to simulate the effect of a high impact load (about 300 g's) on a model with two offset hemispherical shells with rubber spacers between them (see attached pic). The shells are ABS polycarbonate (linear elastic isotropic), and the spacers are silicon rubber (hyperelastic, mooney rivlin, constants 4 and 4.5 MPa). I don't have access to lab equipment, so I determined the constants theoretically.

The force is applied from 0 to 0.2 sec, with a peak of 150 N at 0.1 sec. I changed the singularity elimination factor to 0, convergence tolerance to 0.005 and activated the large strain & displacement options. Using the direct sparse solver, the simulation always fails at 20-30 %. However, it solves completely when I reduce the load to 5 N. I am supposed to use impact loads of upto 500 N, so I'm not sure what's going wrong here. The force is huge but it only acts for a millisecond, so the spacers should spring back immediately. I need to know how much stress is transmitted to the inner shell. Any ideas on how I can go about this ?

Update: Specified a no penetration contact set between inner face of outer shell and outer face of inner shell. Adjusted spacers so that both the flat ends are coincident with the shell faces. I did an interference check and coincident interferences show up even if the box to show them is unchecked. Meshing fails as a result, without any actual interferences in the assembly.