I have been working on simulating hypervelocity impacts on layered materials in abaqus and I am getting some results that I am concerned might indicate my model is inaccurate. My simulation consists of a panel 36mm by 36mm with 11 layers and is clamped around the edge and a total thickness of 1.55 mm with Hashin failure criteria. The panel is impacted by an aluminum impactor with a diameter of 3.6 mm and Johnson Cook damage evolution and plastic deformation where the impactor is moving towards the panel at 20 km/s. The general interaction properties are hard contact and friction with a 0.3 frictional coefficient. The mesh size on the impactor is 0.5 mm and the mesh size on the panel is 0.75 mm. The simulation runs for 50 micro seconds as a dynamic explicitly model though the impact occurs in the first time step of 0.5 microseconds. I am aware that this is pushing the limits of what can be done with abaqus so I checked the model energy and work components over the simulation time and discovered that of the 1500 J converted away from kinetic energy in the impact the largest amount (1106 J) is converted into Internal work by penalty contact (ALLPW) which value is higher than the internal energy of the model (19 J) remains constant even after the initial impact. When talking with my mentor he indicated that this could be a problem as contact energy should return to 0 after the impact has ended and that contact energy should have a lower magnitude that the internal energy (the ALLPW value was  read as -1106 J) however he is less familiar with how energy is calculated in abaqus so if anyone is familiar with the what goes into these values and if they are reasonable your input would be appreciated.