The Metro-North train that derailed yesterday morning was traveling at speeds of 82 miles per hour as it entered the Spuyten Duyvil curve, the National Transportation Safety Board just announced. Speed limits on the curve are just 30 miles per hour, and the speed limit on the straightaway north of the curve is 70. The NTSB noted that they do not yet know if human error or mechanical malfunction caused the deadly incident.
The NTSB noted that six seconds prior to the train coming to a stop, the throttle went to idle, and one second later, break pressure dropped to zero. “We do not yet know the initiating event for the throttle going to idle or the brake pressure dropping to 0 psi,” the agency said in a statement.
As of now, the NTSB has noted that there were no prior problems with the brakes, and the safety investigators will continue interviews with the engineer and three other crew members. The rail cars and locomotive have been removed to a secure location for further study, and the tracks have been turned back over to Metro-North. Yet, extensive service changes remain in place for the afternoon commute and morning rush. I’ll have more as this story develops.
14 comments
Gives a bit of credence to the operator’s statement that the brakes were not working correctly. I can’t imagine a veteran motorman with 20 years experience taking that curve at 83 mph willingly, unless he was impaired in some way.
I have no doubt it was not a deliberate decision to speed.
But nothing and no one is infallible.
The operator could have had a micro-sleep episode, or about two minute’s brain fade during exactly the wrong two mintutes
or
??? some rare mechanical failure sequence caused the train to ignore the commands from the control cab, but finally respond to an e-brake command which was made too late.
btw, if the brake pipe went to zero 5 seconds before the curve, how much would that have slowed the train? I thought passenger trains had a reasonably good decel on e-brakes. PCC cars can do 7mph/second, but they have track brakes.
A decent passenger car can do nearly 30 mph/second.
Stopping distances for cars assume about half this braking rate. And this is only feasible because seat belts and because cars are held to lower standards than trains.
For mainline and subway trains, emergency deceleration is about 1.4 m/s^2. The basic difference between cars and trains is that trains have much less rolling friction, which means they can run at much higher speeds and be much more energy-efficient, but also have worse acceleration and deceleration and longer braking distances. Tires-on-asphalt can’t provide smooth enough ride to safely allow for train-length buses, and steel-on-steel can’t provide large enough capacity in terms of vehicles per hour to economically allow for car-length railed vehicles.
To be safe, you should not assume 30 mph/second. A randomly chosen vehicle will likely fail that standard, and even if the car is capable that amount of grip may not be available when you want it.
But it’s achievable:
Not including the human reaction time/distance, many cars can halt from 60 mph in 110 feet under ideal conditions.
Example:
A 2014 Chevrolet SS sedan can halt from 60mph in 110-115 feet. That’s medium+ sedan, weighing about a hair under 2 tons. It is reasonable to assume a nearly constant rate of deceleration from 88 to 0 ft/sec, so the event takes about 2.5 seconds, or 25mph/second average decel. A smaller, lighter car with more tire per pound (eg, a Boxter) can potentially do better.
This sort of decel is utterly impractical for a train, as standing passengers would pile onto one another.
Better to pile on other passengers than to be impaled by a tree while derailing.
You’re underestimating how dangerous these deceleration rates are for standing passengers. Derailments are perfectly survivable, when the windows don’t fail and the train stays upright.
I thought MNR had some sort of automatic train stop system? Or an alerter if you passed a speed restriction?
LIRR has it. I doubt Metro-North has it when you go from 70 to 30 on that rapid decline.
If there was no brake failure, then the answer is easy.
But if there was, I hope they follow through and check what’s wrong with the rolling stock. In Germany the rolling stock maintenance isn’t great (DB needs to look more profitable for privatization), and in Berlin they discovered a brake flaw that required them to pull three quarters of the S-Bahn trains on a few hours’ notice.
If there’s something potentially dangerous in the connection between the locos and the coaches in the push-pull sets, there may need to be a major recall. Fortunately a large majority of the Metro-North stock would be unaffected; the unelectrified part of the Hudson Line, which is by far the busiest unelectrified Metro-North line, could run into Penn Station with Amtrak dual-modes, or maybe LIRR ones if their shoes can get out of the way of the Metro-North third rail (I’ve read that they can’t).
That train almost made it.
Note that the lead cars had mostly changed direction consonant with the curve, and the trailing units followed the tracks but derailed.
That curve is very mild; when I’ve been a passenger going around it I could not sense any sideways force. At 70 mph it’s about five times as severe; I suspect a Camaro Z28 etc could handle it easily, though with some tire and passenger noise.
There is a design defect of all passenger trains: the wheels are too close together. The train is around fifteen feet high but perched on a wheel track four and half feet wide. Of course it’s going to fall over, the only question is how often.
First, much more than five times as severe. The outward acceleration in the horizontal plane is indeed five times as high, but the acceleration in the plane of the tracks is much more than five times as high, since the inward acceleration produced by cant is constant. Even the baseline is more than five times as high, since (82/30)^2 = 7.5.
To pull numbers somewhat ex recto, if the radius of the curve is 175 meters and cant is 75 mm, then the allowable 30 mph (=13 m/s) speed will produce 75 mm of cant deficiency (=0.5 m/s^2 in the plane of the tracks), whereas 82 mph with the same cant will produce a meter (=7 m/s^2) in the plane of the tracks.
Second, lateral acceleration rates on freeways and on trains are actually very similar. Interstate design standards have curves of radius 350 meters (“5 degrees”) and 5 degrees of cant (equivalent to 130 mm, or 0.85 m/s^2 lateral acceleration in the plane of the road surface), so at design speed of 70 mph (31 m/s) this is 2 m/s^2 lateral acceleration in the plane of the road surface, comparable to the top end of what can be achieved with tilting trains. Of course the acceleration felt by passengers on tilting trains is only about a third of that, but the train’s bogies feel the full 2 m/s^2, and although the maintenance costs of such trains are very high, they run safely.
And third, in safety regimes that emphasize proper maintenance of tracks and trains and automatic train protection, trains do not in fact fall over. In all Shinkansen and TGV derailments, some at speeds approaching 300 km/h, the train stayed upright. Most of the worst rail accidents recently were not about trains jackknifing, but rather about a bridge falling on the train (Eschede), the train falling from a bridge (Wenzhou), or the train running into a building (Amagasaki). Instead of trying to make taking curves at lateral acceleration rates approaching 1 G safe, good safety regimes make sure that such lateral acceleration rates will never happen.
Next time the MTA union reps insist we need multi-person train operation for “safety,” remember that the engineer that the union rep Bottalico called “diligent and competent” was asleep while the train was barrelling around a curve at 82mph, and the other THREE six-figure-salary union employees on this train didn’t notice a single thing amiss.
http://nymag.com/daily/intelli.....north.html
Automated train operation (of course in tandem with a smaller number of on-board employees as a back-up) would reduce such human error, and maybe we could put these union bums to work on something productive.
ATO wouldn’t work when you also have to deal freight traffic.
The engineer should have drinking some coffee before operating the time.