Intermittent anomaly? What's that? Well, it's not the scientific name for a personality disorder. Least of all mine. Nothing intermittent about my personality ordered or disordered.
"Intermittent anomaly" is the probable cause ascertained by NTSB in its investigation into the train-to-train collision that occurred at SEPTA's 69th Street Transportation Center on August 22, 2017.
I urge everyone to take the time to read the complete accident docket. Every item. It will become clear why it took NTSB four years to release this report and it isn't Covid.
Then I'll ask you all to think of the best railroad officer you ever worked for, and imagine taking that cause-- intermittent anomaly-- to that person. Then imagine what would happen next.
Enough of that. In the "leaf season," November, of 2006, my railroad was caught between the rock and the flat space, the wheel flat space. We had placed more than 300 M-7 EMUs into revenue service and as November slipped into December, more than 250 were out-of-service with wheel flats.
We quickly ran out of shop space, and then yard space. And we stood up a lot of people.
What was the problem? Low adhesion? Sure but not exactly. The vehicles were equipped with two systems that not only weren't integrated but worked against each other-- namely the train control and the wheel spin/slide protection systems.
The railroad's train control system was based on continuously coded cab signals, supplemented by automatic train (speed) control. Fixed wayside signals were utilized at interlockings, and, of course, the largest interlocking on the system, Grand Central Terminal.
The M-7 on-board train control system, upon receipt of a cab signal downgrade mandating a deceleration from that current train speed, required the locomotive engineer to acknowledge the downgrade and apply the train brakes to achieve a specific rate of deceleration, identified as the "underspeed brake rate."
Failure to achieve the underspeed brake rate in X seconds would trigger a penalty service application of the brakes.
Further failure to satisfy the underspeed brake rate in a specific time with penalty service braking would initiate an emergency brake application.
Meanwhile, of course, the ever-vigilant wheel spin/slide protection system was ready to reduce braking effort to zero or near zero when slide was detected.
In the circumstance that would most likely trigger the WSP system, low adhesion with a braking requirement automatically enforced, the WSP system was "blind" to the mandatory deceleration, prevented satisfaction of the underspeed brake rate, triggering the service penaltly which intensified the slide activity, which extended the duration of the WSP system "protective" actions, thus triggering an emergency brake application.
Because my railroad was very cautious, or as we put it less delicately-- "because we weren't out of our @#$%ing minds like they were across the Atlantic"-- the WSP system was disabled by an emergency application of the brakes.
We were right, then. We are right, now. Better wheel flats than a signal overrun.
So, in almost the sole circumstance where WSP would be activated, the activation triggered a sequence of actions and reactions that resulted in exactly the condition that the WSP was supposed to prevent, wheel flats.
Better wheel flats than signal overruns, sure. But better yet, no wheel flats and no signal overruns.
That much we knew. We didn't know exactly how to do that.
The president of the railroad, in a slightly annoyed state over the shrinkage in the available fleet, but always generous, gave us 11 months to figure it out.
First and foremost we compared the equipment underspeed brake rate to our signal design distance required rate of deceleration. Satisfying the underspeed brake rate required achieving (if memory serves me) deceleration of 1.4 mph per second. Our signal design distance was based on a deceleration of .84 mph per second. Why the discrepancy? Because the equipment was designed to meet the requirement of our sister railroad's signal design distance.
Our order was piggybacked on that sister railroad's design specs. We were able to specify underrunning third rail shoes, but other than that our options were more than limited.
Then we pulled the event recorder downloads on every train experiencing a wheel slide event resulting in flat spots, and examined the data.
Three-quarters of the events leading to the penalties leading to the wheel flats occurred on the downgrade from a "nornal cab" speed indication and an average speed of less than 65 mph, to a "limited cab" and an authorized speed of 45 mph.
Now the actual risk to the train does not occur if the train doesn't reach 45 mph. It occurs when the train doesn't reach restricted speed, not exceeding 15 mph. The penalties were initiated on average with approximately 3700 feet of braking distance available before receiving a (theoretical) restricting signal. The penalty brake application was truly punitive not preventive.
One-quarter of the penalties resulting in wheel flats occurred on the downgrade from limited cab to medium cab allowing a speed no greater than 30 mph. Further examination led us right to the same conclusion.
So we developed a plan, a "program" as we like to call it. First, upon report of wheel spin slide we would reduce the MAS to 55 mph for those area(s) of low adhesion.
The M-7s were capable of using cellular frequencies to transmit data on vehicle performance, the deputy-chief mechanical officer enlisted the services of our IT department. The IT department created a program in which those messages registered as "hot spots" on a system map.
We also changed the penalty feature such that the "progression" to an emergency brake application was no longer automatic. The decision to initiate the emergency brake application was put back into the hands, literally, of the locomotive engineers.
Penalty service brake still occurred, but they were not irrecoverable, did not exhaust the brake pipe to zero, did not disable WSP, and could not produce the emergency brake application that disabled the WSP.
We were, in a very real sense, de-automating the conflict between the systems of speed control and wheel protection.
We understood that this was counter to the impulse behind automatic enforcement, but automatic enforcement of this specific type had come all too close to disabling the entire service.
We spent a lot of time screening candidates for locomotive engineers and even more time and money training locomotive engineers. We begin that process by making sure any candidate with homocidal or suicidal tendencies is prevented from ever getting his or her hands on the throttle. If we didn't have confidence in that process, then all the flat spots in the world werent' going to save us.
We had our plan for the "leaf season"--November 2007. Did it work? One PM rush hour, looking at the map of hot spots and listening to the reports from the engineers, we imposed the 55 mph MAS sections throughout the entirety of the electric service zone,
My phone rang and I picked it up to hear the president of the railroad, slightly annoyed but still generous, saying "You're at it again. Slowing down my railroad."
I said, "You're so right. We're going to slow your railroad down to the best on-time performance of any November ever."
We did just that. Best OTP for any November; fewest cars out-of-service for wheel flats for any November; and no signal overruns.
We were right then, and we are right now. Wheel flats are preferable to signal overruns. WSP systems cannot be allowed to nullify emergency brake applications, nor should they be allowed to create the need for an emergency brake application.
No wheel flats and no signal runs are even better.
And there's nothing intermittent or anomalous in that.
David Schanoes
February 8, 2022
Name That Tune
I couldn't sleep at all last night
Copyright 2012 Ten90 Solutions LLC. All rights reserved.