Were you able to find an error message on one of the mesXXXX files?
You ran the model on how many cores? Can you try a different core count?
Have you tried running double precision?
Also, which MPI are you using (Intel MPI? Platform MPI?). Have you tried with a different MPI?
You can use the *CONTROL_MPP_IO_LSTC_REDUCE keyword to get the same results with different MPI. Here is what the user manual says:
"For numerical consistency, LSTC_REDUCE was developed to ensure a deterministic
summation order. This keyword gives identical results while using
the same decomposition and core counts but different MPI algorithms."
You can download the LS-DYNA manuals here:
Let me know how it goes.