I have delayed the replay, because there is an arithmetic problem in Axle force and speed computation (NaN - Not a Number); as I can't reproduce the problem, I'm trying to find a bypass to the crash and am still studying what can be an acceptable solution.
A workaround for this is now present in ORNYMG 152.1. When a NaN occurs in the code section that generated the crash, an error message is logged and a temporary switch to simple adhesion occurs.