On Wednesday, February 27, the TTC presented a report on the derailment at Kennedy on February 4.
My comments here are a paraphase of the narration at the Commission Meeting.
The derailment happened late at night on a train leaving Kennedy westbound. The front truck of the last car in the train derailed with the car moving about a foot off of the rails. As it was dragged about 300 feet through the crossover, the track, signals and other plant in the tunnel were damaged. Getting the train back on the track and out of the tunnel, plus repairs sufficient to allow operations to resume, took a day and a half. Since then, a 15 kph speed restriction on the crossover was needed for safety as the track fastening to the tunnel floor was returned to good condition for regular operations.
This work is expected to complete on February 29 (Friday).
Page 3 shows the normal configuration of an H6 truck. For those unfamiliar with subway car design, a brief explanation. There are two axles and two motors on each truck with the motors mounted at right angles to the axles. The link between the motor and the gearbox is supported by the torque arm assembly (light blue). This link is flexible and the torque arm can pivot on the truck as we will see in following drawings.
In the event this link fails, the supporting torque arm is supposed to be caught by the safety pin (yellow).
Page 4 shows the initial failure of the torque arm link which allowed the entire assembly to drop down onto the safety pin. Where this happened is unknown, but presumably somewhere on the last eastbound trip toward Kennedy.
Page 5 shows the second failure where the safety pin fell out somewhere between Sherbourne and Castle Frank. This allowed the torque arm to fall to the trackbed. From that point on, the assembly started to break apart losing bits and pieces as the train proceeded east. However, note that the torque arm is in a “trailing” position where it can bouce along without snagging on equipment at track level.
Page 6 shows the problem as the train reversed at Kennedy. Now the gearbox is facing into the direction of travel, and when it reaches the crossover, it snags on the rails. The assembly pivots around and lifts the truck clear of the track so that is now dragging along the tunnel about a foot away from the rails. Meanwhile, the other gearbox runs into a restraining rail and pivots 180 degrees from its standard position. The centre pin holding the car into the truck sheared off and the carbody moved about two feet from its normal position relative to the truck.
Page 7 shows the comparison between a new torque arm link and the one that failed. Page 8 shows a photograph of the torque arm itself and the mark made when it fell onto the safety pin. Page 9 shows how the safety pin and its retaining system failed.
Page 10 shows how the assembly has been redesigned. This change has been retrofitted on the entire H6 fleet (the only cars using this truck design). Note how the new restraining plate locks the safety pin into place.
Page 11 shows the new design for the torque arm link. This has been modified to strengthen the part and to eliminate possible wear points that could lead to failure. Replacement parts are now on order and will be retrofitted to all cars.
Pages 13 and 14 contrast the design of the torque arms and their safety pins on the H6 and T1 cars showing the improved design on the T1.
I have already commented at length on how the TTC managed service following this accident and won’t rehash that discussion here. The TTC was extraordinarily lucky that this failure occured late at night in a location and manner where there was no injury, and where the temporary loss of a station did not block the entire line.
I have left this item open for comments, but do not intend to engage in a discussion about car design or maintenance as this is well beyond the technical competence of most of us and I do not want to engage in speculation about such a serious matter.
It may be a minor point, but I’m not clear on just what happened to gearbox #4 at Kennedy station.
I presume that gearbox #4 would have its own torque link keeping it from rotating. And yet it rotated. All I can think is that the impact with the retaining rail (which I guess are those safety rails found inside of the running rails at switches, bridges, etc.) was severe enough to break the torque link or associated pieces for #4, making a complete mess of the truck.
Also, was there any discussion of why the failure east of Sherbourne went undetected? As soon as the “immediate catastrophic driveshaft failure” occurs, then:
train performance decreases
the now-disconnected motor can free-run (would this not lead to an overspeed condition? or, if there are controls for this, an problem indication?)
various bad noises as the torque arm bounces along the trackbed eastbound, with the bent link banging against the truck and then the torque arm falling off (I wonder if the TTC was also lucky that the debris from the failing truck didn’t derail the train when it was eastbound, presumably with a good load of passengers)
As 5857 was the lead car eastbound, the operator would have been in that car. Were there enough noises and vibrations from the failure that the operator should have figured something was up?
Steve: Re gearbox #4, you are correct. The issue of noise came up in discussion of the report, and apparently the operator heard nothing unusual. It would have been the truck at the other end of the car from the cab.
I have not read much comment regarding the shuttle buses operated during the original shutdown following the derailment. What I cannot understand is why the TTC insisted on running shuttle buses in the first place. I read it was 30 buses. Since most people were going to take the subway beyond Warden, why not simply extend all routes normally ending at Kennedy to Warden? 2 or 3 extras on each route would have handled the load. 3 buses on 10 routes, same resource, much better service. As for the repair work itself and the extensive delay it getting it repaired just points to their inability to respond to serious situations adequately.
I took an indirect route home one evening so I could experience the shuttle service (rode it westbound and watched things at Warden for 10 minutes or so), and it seemed to be running pretty smoothly. Buses were departing every one or two minutes — basically, enough time to pull up, load with passengers and close the doors. At Warden, there was a constant stream of passengers loading onto the buses, but there never seemed to be a queue waiting since as soon as one bus left, another was waiting to take its place. The actual trip was pretty smooth as well (thanks in part to paid duty officers giving extra time for left-turning buses at Eglinton/Kennedy); the only problem was queues of buses exiting Kennedy and entering Warden, which ended up taking up half the travel time.
I don’t think extending Kennedy-bound buses to Warden would have helped much. The delays on the shuttle were attributable to bus congestion, which would have been there regardless whether the extra buses were on the subway shuttle or on the extended 43, 57, 86, 116 etc. The transfer penalty from shuttle to bus at Kennedy wouldn’t have been much different since the shuttles were running every minute or so. Keep in mind also that you’d have to run a shuttle anyway to handle SRT riders, and then compare passengers trying to track down the correct route in a single bus bay at Warden in a confusing and unusual situation vs. the efficiency of directing everyone to the single shuttle service. Keeping things simple was the best way to go, and at least when I was out there in the evening rush hour it seemed to be running fairly well, considering the circumstances.
I am wondering if ‘dragging equipment’ detectors, like the railways often have installed with hotbox detectors, might be in the future for the TTC.
Steve: As far as I know, this is the first incident of this kind. It is important that we not spend a lot of money trying to detect a repeat of a single incident when there are so many more conceivable, if unlikely, ways equipment might fail.
On a cautionary note, it only takes one incident to end up with a lawsuit. As has already been said, the TTC was extraordinarily lucky with both the time and place of this incident, someone up there likes the TTC (at least a little). Timmy’s puts “Caution: Hot” on its coffees after a lawsuit. The TTC would likely save money through avoided litigation by investing in preventative measures than taking a reactive approach to potential equipment failures, however unlikely (and I agree they are unlikely, engineers are not stupid). They may not be as lucky should there ever be a next time. Are we supposed to wait for a sequel to Russell Hill (which this incident could potentially have led to with different timing (and speed))? Of course not. I think Calvin’s point is not excessive concern. Unreasonable people do ride the subway, after all.
I know that there was a similar incident not long after the B/D was extended to Islington as an H-1 car lost a drive shaft and it was pole vaulted to the ceiling and dropped about two feet to the right side while entering the x-over at Islington. The train could not have been travelling very fast for there was little indication of track damage beyond a few feet of the incident’s occurrence.
With this said, I concur with your comment to Calvin, incident rates are low so is it really necessary to spend that much on even more technology?
This serves as an example of something I’ve been saying for a long time about the TTC bus fleet.
All of a sudden after 20-some years of good service, the H6’s develop a problem, and need a part replaced. Thank goodness it was just one accident. If there had been another, regardless of if there was injuries, they might have had to take all the H6’s (around 25% of the subway fleet) out of service for a time.
While the subway fleet will never be uniform given it’s size and the constant advancement of technology, the bus and streetcar fleet can be. 100% of our streetcars were made from the same technology. 60% of our buses are of the same Orion VII bus model (with some changes for newer generations of the same product). What if there was some similar critical fault with these VII buses that didn’t pop up until 10 or so years of service? By then we might have 90% of the bus fleet be these buses and all of a sudden we need to pull them all off the road to be fixed. Warrenty or not, its a bad idea IMHO to put all your eggs in one basket and this is an excellent example of why.
20 years of good service? The H-6 fleet has been problematic since day one. The first cars didn’t enter service until at least a year after they were delivered. The fleet was withdrawn at least twice while in the process of delivery, and has had a number of equipment replacements since, including, I believe, the truck frames. The interiors haven’t held up well either, with the panels beside the doors needing to be painted and/or replaced due to unsightly “head spots” forming because the panels were not arborite/equivalent, but painted sheet metal. The H-6s were built during a period of transitional ownership and other problems at UTDC/Lavalin and the technology used in them was new and troublesome. They jolt when they stop and they often sound like they’re failing as the blowers go out and the lights dim. I’d sooner they keep the H-5s around for another 10 years when the new cars arrive and junk the H-6s instead.
In response to Nick:
The concern is real, in fact the New Flyer LF40s had a steering problem that did pull them from service a few years back (at least in Waterloo region, not sure about other operators), and there was that mess with the Combino, but honestly there’s not much that can be done. Even a fleet split three ways is going to be devastated by a type pulled from service, while going with one type has significant and ongoing savings for both capital and operational budgets; in the long term the little resilience gained from a split fleet just isn’t worth the cost.
Catastrophic events are NEVER the result of one event, but always the result of several events and situations aligning themselves.
Until now, there appears to have been only one incident involving something dragging (see Mr. D’s comment). Even in this case, it did not cause an accident until the train changed direction at Kennedy, long after the possibility of detecting the problem could have been done. The timing of this incident worked in the favour of the outcome, just by chance.
The Russell Hill accident was also not supposed to happen and had an equally low incident rate. A significant amount of money was spent to upgrade the signal system to allow an operator to know the difference between a red light that was red due to block occupancy from a red light that was red due to grade timing. I suspect that the installation of dragging equipment detectors in the minimal number of locations (likely just in advance of track switches at an interlocking) would not be a great cost.
It would no doubt be less costly than one lawsuit – especially now that they’ve had a dragging equipment incident.
Is the link you have in the article reproducing the official TTC report? I’ve looked on the TTC site without luck.
Steve: The original report was on the TTC site as a 60MB Powerpoint very briefly. I downloaded it and converted it into a PDF (all of 355K) which is the report on my site.
What makes me wonder is how long a properly-secured safety pin would have lasted. Because the motors are regenerative, there’s torque in both acceleration and deceleration. With a broken torque link, the arm will rise up in one mode, and fall back onto the safety pin in the other mode. This would happen at least once per station.
Are all the trucks inspected daily? Presuming the safety pin holds, how long could a broken torque link escape detection? Hopefully not so long that the safety pin fatigues and breaks.