Monday, December 25, 2017

Too Fast and Too Slow: The Washington State Derailment and Positive Train Control


After more than a decade of planning and construction, a new section of track was opened for Amtrak passenger service south of Tacoma, Washington on Dec. 18, 2017.  The old route that Amtrak trains used to take went northwest from Tacoma along the coast of Puget Sound, around a peninsula named Point Defiance, and then down the coastline several miles until it crossed Interstate 5 south of the small town of DuPont and headed south inland.  The new shorter route uses a bypass track that goes southwest of Tacoma and hugs I-5 for the rest of the distance, crossing the interstate south of DuPont.  There is a long stretch of fairly straight track just north of I-5 past a golf course before the track makes a sharp left turn to the south to cross the bridge over the freeway.

The problem with the old route was that a number of sharp turns and single-track tunnels slowed the Amtrak passenger trains down, making the Point Defiance section something of a bottleneck.  The project map on the Washington State Department of Transportation website for the Point Defiance bypass bragged that the top speed allowed on the new route would be 79 miles per hour.

Rail fans and others interested in passenger rail transportation made plans to be on Amtrak 501 as it left the station in Tacoma on the new route.  The engineer, whose name has not yet been released, was training another railroad employee who rode with him in the cab. 

In most parts of the U. S., trains are not operated in a completely automatic mode, although in many regions a system called Positive Train Control (PTC) is in operation.  PTC is a kind of robotic supervisory system that, among other things, constantly monitors a train's speed and intervenes if the train goes too fast for a particular section of track.  About 60% of all Amtrak trains use PTC, but in order for PTC to work, the track has to have sensors installed along it, and the Point Defiance bypass was not one of those routes.  So the engineer was solely in charge.

Around 7:25 AM, the train was running on the long stretch of straight track before the turn to the bridge over I-5.  A properly trained engineer knows what speeds are safe for which parts of a route, and knows when to apply brakes in anticipation of a lower-speed area ahead, as passenger trains can take several miles to decelerate at a rate that doesn't unduly disturb the passengers.  A video exists of what was going on in the cab in the last few seconds before the train reached the I-5 bridge.  The train was still going at the maximum route speed of 78 MPH.  Six seconds before the bridge, the engineer commented about the excesssive speed of the train, but by then it was too late.  The engine and a dozen other cars left the tracks, killing three, injuring dozens, causing numerous highway-traffic crashes (none fatal), and closing Interstate 5 for many hours.  The maximum safe speed for negotiating the turn was posted as 30 MPH.

Although the National Transportation Safety Board (NTSB) will not issue its formal report on the investigation of this disaster for many months, the preliminary evidence is pretty clear that the accident was caused by human error.  Something—possibly distraction in conversing with the trainee, possibly plain forgetfulness—made the engineer neglect to slow the train before the I-5 curve.  As numerous reports emphasized after the wreck, if the train had been using PTC, it would have automatically slowed down for the curve if the engineer had done nothing, or even if he had tried to keep the speed high.  And we have no knowledge of how many wrecks of both freight and passenger trains have been prevented by PTC, because by definition such incidents that don't injure or kill anybody don't get reported.  But it is clear in this case that the absence of PTC was a contributory cause.

Congress mandated the installation of PTC after the worst train accident in the last thirty years, a 2008 wreck caused by operator error that killed 24 people.  The original deadline for all passenger trains to be using PTC was 2015.  But as the deadline approached and railroads were lagging behind in their rate of installations—in fairness to them, due to problems with government regulation of necessary radio frequencies as well as other causes—they told Congress that if the deadline wasn't extended, they would simply shut down.  How serious this threat was, we'll never know, because Congress caved and moved the deadline to the end of 2018.  And under the current business-friendly administration, we can expect if the railroads ask for another extension, they're likely to get it.

Statistically, rail passenger travel is very safe overall, with the number of fatalities most years hovering in the single digits.  Still, nobody wants to be one of the six or seven people who get killed in a train wreck or hit by lightning—dead is dead, no matter how you go. 

A utilitarian approach to the issue of PTC and passenger trains might conclude that, hey, given the low number of fatalities, let's just allow things to go the way they're going, and eventually we'll have PTC everywhere and we won't have to worry about it.  But the expense per life saved is so high with railroads that we'd be better off using political and monetary capital fighting automobile traffic accidents or promoting self-driving cars.

That's one approach.  But another approach says, "Look, here's this technological fix that will cost the railroads money and trouble, but will almost completely eliminate what is the last major remaining cause of railroad passenger fatalities:  human error.  Let's bite the bullet and make a special effort, even spend some extra money, to fix this thing once and for all."  Maybe that's the engineering approach, or even the perfectionist approach (many engineers have perfectionist tendencies).  Yes, the absolute numbers of fatalities are small.  But deaths in a train wreck share with deaths in plane crashes a peculiar horror, in that you are completely bereft of control of the situation.  And in the case of train fans who simply wanted to experience a new route for the first time and ended up paying for their hobby with their lives—well, some ironies are too much to contemplate.  I have a good friend who, if he was not otherwise engaged that day, might well have been on that train, because he simply likes to ride trains.

Better training (pardon the pun) of engineers and faster completion of the installation of PTC are needed.  And maybe if these things happen, this will be the last fatal accident involving train passengers for a long time.

Sources:  I referred to several news items on the accident, including CBS News at https://www.cbsnews.com/news/amtrak-derailment-dupont-washington-video-shows-crew-not-using-electronic-devices/, a government-run transportation statistics site at https://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/publications/national_transportation_statistics/html/table_02_42.html, a Washington State Department of Transportation map of the bypass route at https://www.wsdot.wa.gov/Projects/Rail/PNWRC_PtDefiance/Map.htm, and a report giving the time of the crash at https://www.washingtonpost.com/news/dr-gridlock/wp/2017/12/18/amtrak-train-derails-in-washington-state-rail-cars-fall-onto-interstate-5/. 

1 comment:

  1. Quote: About 60% of all Amtrak trains use PTC, but in order for PTC to work, the track has to have sensors installed along it, and the Point Defiance bypass was not one of those routes."

    Sensors with accompanying communication links? That seems grossly over-engineered. A GPS could determine where the train is and software know what the correct speed is for that stretch of track. There's already software that tells drivers if they're speeding along a particular stretch of road. This would be that but for rail lines.

    My hunch is that the railroads weren't clever enough to keep up with technological change. They came up with a fix more appropriate for the 1950s. That meant lots of unnecessary hardware.

    Egads, someone could create a smartphone app that'd at least alarm and tell a driver when to slow down.

    There's another issue here. What madness led Amtrack to task the engineer on the first schedule trip on this new track with also teaching someone else at the same time? He should have had time to learn the route before being expected to teach it to others.

    ReplyDelete