Regardless, I would like to see the downloads from the event recorder and TIR for the locomotive. You can't fudge that data. I still think this either operator error, or deliberate.
My colleague spent a week in Pueblo, CO with a 48 KHz digital audio recorder connected to many identified defects on the "hospital train". They were cataloged, then characterized using DSP techniques. The bearing cone, cage, cup and roller frequencies can be extracted based on geometry and angular velocity. I turned his MATLAB work into real time monitoring with analysis and data reduction before saving to a web based database for trend analysis.
The bearing burn-off between hot bearing detectors in the rail bed was handled using sensors in the "stink bomb" holes on the bearing adapters. I can send the exact axle, side and car id to the locomotive engineer in real time...before a bearing burns off and derails the car. Inner and outer bearing adapter temperature measurements can expose wear requiring service ahead of a catastrophic failure. On a train that isn't in motion, I can discern which side of the train is illuminated by sunlight as the bearings are warmer than the shadowed side. Fun details gleaned from field testing.
I agree. We need to see the event recorder logs. All the yapping about the windshield being hit by a projectile appears to be intended to distract attention from the engineer's responsibility to actively control the speed.