Eastern Railroad Discussion > Technical question - roller bearing failure

Does anybody know how long a failed roller bearing usually travels before triggering a wayside hotbox detector? Just how long can these things go in terms of mileage before it gets noticed? I also assume the load/empty status of the car has an impact on distance its able to travel before burning off (since the force of friction is a function of weight, failed bearings on heavier cars heat up sooner than on empties).

Given the fact that most detectors are spaced 15-20 miles apart, I assume that once the temperature of the steel reaches the threshold reading needed to trigger the detector, then the bearing has less than 30-40 miles before it burns off completely. But just how long do they heat up (smoke and spark) before they ever get noticed by the detectors? What is that average elapsed time?

I saw an empty coal train today where one car was developing a hotbox. Mild smoke, the smell, that distinctive 'ratchet' sound (which audible hotbox detectors listen for)... When the train passed the next detector 3 miles up the line, no defects were reported. It very well could have been a situation where the bearing wasn't hot enough yet to trip it. But therein lies the reason for my questions.

Anybody who works for Timken on here who can shed some light on their products a bit?

The bearings you refer to are a "Timken Boxcar Bearing," they are also used on the "W.R. Stamler Conveyor Belt Feeder," which is a 36" primary crusher used in both surface and underground mines.
You have to remember that this bearing is pressed on to a large diameter shaft that supports the wheel assembly. This being the case the bearing during failure will develop a lot of heat but it also will
dissipate a lot of heat as well, so it takes a long time for a bearing of this type- "cup and cone" to finally destroy itself.

The old brass journals had a rather long failure curve, leaving billows of smoke for miles before reaching failure. On many railroads, hotbox detectors are still spaced on the basis of that failure curve.

Unfortunately, roller bearings have a very steep failure curve, in some cases being only a few miles from any human sensual indication to catastrophic failure. It's only a few years ago that this has finally gotten the attention it needs.

Detector manufacturers and the AAR are busy trying to find earlier warning methods which would permit a train to stop and set out the car before there was nothing left to set out! Therin lies the move toward acoustic and other new technology detectors. Early warning of roller bearing failure has a large data base to draw from in general industry, fortunately. Much is known about acoustic signals and vibration signatures at the very earliest stages of bearing degradation. The challenge, as always, is to put this knowledge to use in a railroad environment, except for outer space or deep sea, possibly the toughest around.

If there is a characteristic acoustic signature of a failing bearing, is it loud enough / persistent enough to be detected at the end of the train? Would an acoustic sensor tuned to the right range of frequencies have the sensitivity to catch the sound? If so, could a FRED be modified to pick up the sound and set the brakes from the rear? Not the same as a set of eyes at the back, but nothing else is.

I've heard of bearings failing within 5 miles. There was an incident on the CP in which a couple of propane tanks "BLEVED" in the subsequent derailment

-Eric

A couple of years ago I met a retired NKP hogger who told of the time he had a roller failure on the rear tender truck of his ex-Wheeling Berk and couldn't convince anyone in Brewster that his iron horse had really come up lame. Finally, a road foreman and a trainmaster were sent to the site to see for themselves.

See! this is just one more good reason why we want our cabeese back, and we wants them now! No sub for a good set of human peepers on the EOT.