What's the possible reason for the large energy error


Hi all,
A drop simulation was solved by radioss. The energy error was negative at initial, but it increased slowly and became positive. At time 19.97s, the energy error increased suddenly to 99.9%, and the energy increased several times. The .out file was attached below.
Looking at .out file, the Interface 29 controlled the time step. In the model, the Interface 29 was self contact of foam material.
/INTER/TYPE7/29
DiPao_SelfContact
156 157
0.01
000
/GRNOD/PART/156
GrnodPartForInterfaceId_29
86
/SURF/PART/EXT/157
SurfPartForInterfaceId_29
86
What's the possible reason for this energy error problem (I think the energy error should be negative and the absolute value should be less than 5%)? How to solve it?
Best Wishes
Roy