Breaking: One confirmed dead as NJ Transit train crashes into Hoboken station

There is much debate over whether PTC is necessary (or possible, but that’s for another day) in terminals, but I’m confused why it’s debated whether PTC, if implemented, could prevent this. Of course it could, it’s express purpose is to enforce speed limits and prevent going through a red indication (physical signal or otherwise). Given those two things, why would PTC not have prevented overspeed here, as early indications describe, and/or not have been able to prevent continued overspeed as the train approached the bumper? Seems to me this is precisely what PTC does.

In any case, as I said before, you don’t need PTC to have prevented this. An accident like this would have been prevented by NYCT’s rudimentary present-day system of signals and trip cocks. I’m not familiat with NJT’s signaling, but either 1) they have no controls like this at Hoboken, or 2) those systems failed AND either there was mechanical failure on the train and/or there was operator error.

There are not many possible excuses for this accident at Hoboken. It’s a terminal, not a curve mid-route. You pass through a tunnel prior to getting there, so hard to miss. And then there’s lots of switching in the yard (in fact, I’m not sure, but I think there are a number of moves required to get from mainline to Track 5 at Hoboken, so again, not sure how the operator would have missed that).

Seems to me operator error (or medical issue) is likely here, but I’m more interested in why 1900s technology wasn’t in place to stop it.

So NASA could design the Space Shuttle to land itself, but we can’t design a train control system to bring trains into a station?

PTC and cab signals usually allow for manual override at sub- 10 MPH. This is because if there’s a signal malfunction faulting all traffic into a dead stop you need, for safety, enough engineer leeway to move the train away from a hazard like a grade crossing or switch and to be able to to crawl into a nearby station platform instead of being stranded. Inside yard limits the sub-10 override is a practicality because freight or passenger shunting activity between yard tracks becomes uselessly limited by all the extra signal waits, to point where the terminal can barely function. About the best you can do is go lower to a sub-5 MPH override, but that’s a balancing act that differs yard-by-yard, terminal-by-terminal for how low you can go before paralysis sets in on terminal traffic.

The engineer will still get dinged with an overspeeding penalty for exceeding the absolute limit of 10, but there’s additional latitude under that limit. Conductors have official supervisory duties inside yard limits to act as a check on the engineer to ensure that latitude is being handled responsibly. On this train the engineer did recall doing the mandated checklist upon entering yard limits: speedometer check, brake-tap test, and horn blow…all corroborated by the conductors. Since the signal system was found to be operating fully normally by the NTSB upon weekend inspection, overspeeds can be reliably ruled out everywhere except for that last 1200 ft. into the platform that appears to be the upper bounds of the area where something went horribly wrong.

Auto-stops aren’t bang-bang instantaneous, either, as they usually allow the engineer 1-2 seconds to react with a controlled brake application before the much more severe jolt of the brake penalty hits. So there are limits to how close that final-final stop signal can be placed to end-of-track before unimpeded inertia-of-motion is going to do its thing. You can’t completely eliminate the engineer override to crawl into the platform, otherwise the train will never physically reach the unloading spot for the first cars. The best you can do is set it at 5 MPH at the last signal, which isn’t going to 100% close the loophole for crashing through the bumper. For all we know that’s exactly how it’s set up within Hoboken, and the best feasible signal layout imaginable within the constraints of that terminal wouldn’t have changed a damn thing about the outcome.

All of this is why yard limits are the single most vulnerable place on any given railroad for train-on-train collisions or derailments from splitting switches mid-throw. About the best you can do is keep the speed limit to the tolerable minimum for the operator override within the constraints of the terminal district’s layout. But because the operator override has to exist to some degree–be it sub-10 MPH or sub-5 MPH–the “feature, not bug” part of the terminal district crawl is that the kinds of human-error fender benders and derailments possible at those override speeds are too slow and too direct a strike on the most fortified ends of the trains to cause serious onboard injury. If this had been an accident literally anywhere else on the approach other than that last 1200 ft. into the bumper post and roof supports the extent of injuries would’ve been a handuful of standees in neck braces. This incident just happened to hit an absolute hole-in-one at the singular point unmitigated by any of the usual design safeguards and probabilities therein, at a singular Hoboken-specific vulnerability where those roof supports are in easy strike distance of the bumper post. Collapsed roof lumped onto all the other damage sustained by the cab car from plowing through the bumper was just too much for survivability.

Maybe the only thing they truly can do is replace those flimsy little bumper posts with heavy hydraulic-compression bumpers (sample pic here: https://goo.gl/5NtfGA) rated strong enough to stop an unbraked 5 MPH train. Surprised they’re still using those nothing little ones at Hoboken given proximity of the roof supports. Other than that obvious change, maybe just moving the rebuilt roof supports as much as they can be moved away from the strike zone of a train that overpowers the best hydraulic bumpers money can buy. It’s still unlikely that the loophole can be perfectly 100.000% closed.