What We Know About Amtrak 501

Earlier today, we saw the latest in a series of crashes that have plagued Amtrak and other US passenger rail providers over the last few years. This is, first and foremost, a human tragedy; but it is also an urgent concern of public policy. While trains–and all public transit–are on a population level much safer than driving, there is no need to accept any casualties at all, ever. While others–primarily the NTSB–will provide a full analysis in the weeks and years ahead, this is my attempt to reckon with what we know about this incident as of the same evening. I had intended this to be a series of bullet points but WordPress doesn’t like the formatting, so I’ve bolded every topic heading. 

Let’s keep in mind that the victims of this tragedy should be in our minds; I haven’t seen a casualty count since the morning, but we know there are fatalities and serious injuries. That shouldn’t have happened, and in addition to wishing their families comfort, this post is inspired by a sense that we–myself as a transportation professional and those who read this blog–should do all we can to prevent such things from happening.

Amtrak 501 was operating over–was, in fact, and somewhat remarkably, the very first revenue train over– the Point Defiance Bypass, a state/federal-funded project that moves passenger trains from a mudslide-prone, curvy waterfront route around Tacoma to a more direct, faster route.


wsdot project map

Source: WSDOT

While the tracks for the bypass have been in service, they have not carried passenger trains along their whole length until now. Trains have been running to test the line for months, but this was the first one to carry passengers.

As befits its purpose, the Point Defiance Bypass is mostly straight, easy 79-mph running, but the area where the train derailed is much trickier. Toward the southern end of the bypass, not far from rejoining the freight main at Nisqually Junction, the tracks flow into an S-curve with a bridge over I-5 in the middle.


derailment 3d

Looking south, in the direction of train travel.

Going into the curve southbound, the speed limit drops from the standard track speed of 79 mph to 30 mph, as confirmed by an Amtrak employee timetable I’ve been sent. 

amtrak timetableAccording to one report, there should have been an indicator sign two miles before the speed restriction indicating the drop in speed; certainly, there was a sign indicating the 30 mph restriction immediately before the curve.

The train was probably going too fast. Amtrak’s train tracking system doesn’t report train speed or location completely continuously (at least not publicly) but in this case it appears to have pinged the train immediately before the crash, reporting a speed of 81.1 mph at a position just 1400 feet east of the crash site. The system isn’t 100% reliable, so don’t worry about the report that the train was going two mph above the speed limit (which wouldn’t have made a difference in any case). transitdocs detail The same Seattle Times report quoted a motorist who said he was driving in the 60 mph range and the train was going faster. And the positioning of the crashed train–the lead locomotive taking a nearly straight route out of the curve, as if it didn’t follow the tracks at all–indicates a speeding train whose inertia carried it (or rather, part of it) forward. Remember, the train should have been going 30 mph going into that curve. There is no way for a passenger train to shed 50 mph in the space of 1400 feet.

If the reporting system data and eyewitness reports are at all accurate, this is pretty clearly a case of a train exceeding the speed it should have been operating at. Overspeed (as it is technically known) is, however, more a descriptor than an explanation; beyond that I strongly discourage speculation. There are too many causes to count: operator error; signal failure; equipment problems (the lead locomotive was a brand-new Siemens Charger); track problems (remember, this is new, or at least recently refreshed, infrastructure); or any number of other possibilities.

Though I discourage speculation about root causes, it’s impossible not to note the scary parallels between this crash and two other recent overspeed crashes, Amtrak 188 at Frankford Junction, Philadelphia in 2015 and Metro-North at Spuyten Duyvil in the Bronx in 2013.



Diagram of the Amtrak 188 crash at Frankford Junction. Source: http://www.businessinsider.com/here-everything-we-know-about-amtrak-188-train-wreck-in-philadelphia-2015-5



Metro-North crash at Spuyten Duyvil. Source: https://twitter.com/NTSB/status/407329136735027200/photo/1

Amtrak 188 entered a 50 mph curve at 106 mph; in a situation eerily similar to today’s the Metro-North train entered a 30 mph curve at 82 mph. We still don’t really know the root cause of the Frankford Junction crash, though most theories have centered around the engineer (who is suffering from amnesia from the accident) losing attention somehow, without his recollections it’s impossible to know for sure. At Spuyten Duyvil the engineer suffered from sleep apnea and was apparently asleep as the train went around the curve (the same issue has come up in several other, more minor commuter rail incidents recently, including at Hoboken and Atlantic Terminal). Whatever the cause, overspeed incidents are all too common on American railroads.

Discussions about these kinds of things always come back to Positive Train Control. Originally mandated by Congress after the 2008 Chatsworth crashnot an overspeed incident, for what it’s worth–PTC implementation was an unfunded mandate, has suffered extreme resistance from the railroad industry, and has been painfully slow. As at Frankford Junction and Spuyten Duyvil, PTC was not in operation on the Point Defiance Bypass today; as far as I can tell, it is intended for operational status later this year (as indeed it was at Frankford Junction…ouch). Yes, barring some kind of drastic equipment failure, PTC likely would have stopped this crash. But it’s worth noting that it’s not the only technology available to stop a speeding train headed into a slow zone; various forms of Automatic Train Stop have been able to do so for almost 100 years. So while the increasing series of crashes is absolutely making a cumulative case for cracking down on the rail industry’s PTC slowness, we should keep in mind that failures like this implicate not only the PTC mandate, but the entire safety culture of American railroading.

Let’s talk about safety culture. Jason Laughlin of the Philadelphia Inquirer just published a piece yesterday (literally not kidding) building off of the NTSB’s scathing assessment of Amtrak’s “safety culture,” stemming from yet another fatal crash, this one at Chester, PA in 2016. Let’s just take a moment to appreciate that the two maintenance-of-way workers killed in the crash and the train engineer involved all tested positive for drugs, and yet that was not found to be a necessary contributing factor to the crash. Similar assessments of commuter railroads have been, while perhaps not as bad, not encouraging either. American railroading has a lot of pathologies–a reactionary culture; toxic labor-management relations; an inability to accept innovation or new ideas–but few have the potential to affect riders as directly as the dysfunctional attitude that it sometimes seems everyone from the top down takes toward safety. It’s a problem that pervades both management and labor, and no one should escape the recriminations, when they come, unscathed. Alex Forrest has a good thread about the cultural contrasts between American and Japanese attitudes toward rail safety; but let’s just say the challenge of 21st century American railroading will be to change a culture where the idea that a train will go on the ground every so often is acceptable rather than unimaginable.

The train’s equipment–a new Siemens Charger locomotive and articulated, lightweight Talgo coaches–is fairly unusual by US standards, but there’s no indication it played any role in the crash. Here, you can see the Charger sitting on the freeway south of the bridge, the 12 Talgo coaches in various geometric arrangements across the crash site, and the trailing P42 (presumably included as insurance for the new locomotive) still sitting on the tracks. 

Don’t freak out. Train crashes get a lot of attention because they’re unusual, visually spectacular, good media content, and a grand American tradition going back to the 19th century. That doesn’t mean they’re actually common. You’re still a lot safer on the train than in a car. I’m obviously mad at American railroad safety culture–and you should be too–but that shouldn’t get in the way of data-oriented reality, even in moments where it’s tempting. Because ultimately, this is all about getting our casualties from mobility down to precisely zero–and we have a lot more work to do on the car side than the transit side.  

Featured Image source: https://www.seattletimes.com/seattle-news/photos-from-amtrak-train-derailment-near-olympia/

18 thoughts on “What We Know About Amtrak 501

  1. I think it’s worth noting that this part of the rail corridor is quite old — probably over 100 years. The tracks have been refurbished as part of the bypass project, but these 30 mph curves go back a long way (possibly only to the time when Interstate 5 was put in, but that’s at least back before 1969, according to historicaerials.com).

    I’ll be interested to learn how fast this type of equipment can slow down. An advance warning 2 miles out is almost too far a way for a passenger train, in my opinion, since they can slow much more quickly than freights. I need to ride the Northstar train here in the Twin Cities again to take some measurements of how quickly it stops in service from ~80 mph, since it has certainly felt like much less than a mile when I’ve taken it in the past. Of course, that question is largely moot since it sounds like the Cascades train was continuing at 80 up until the point it left the tracks, though it could play into any discussions we have about the engineer being distracted and at what point they would want/need to take action to slow for the turn.

  2. Pingback: Today’s Headlines – Streetsblog Los Angeles

  3. Pingback: What the Amtrak 501 Derailment Says About America’s Failure to Prevent Train Crashes – Streetsblog USA

  4. Pingback: What the Amtrak 501 Derailment Says About America’s Failure to Prevent Train Crashes – Streetsblog New York City

  5. Pingback: Casualties in Point Defiance Bypass Derailment

  6. Pingback: What is positive train control and why wasn’t it being used in the Amtrak train crash?

  7. Pingback: What the Amtrak 501 Derailment Says About America’s Failure to Prevent Train Crashes – Streetsblog USA

  8. A lot of the conversation around railroad safety focuses on Positive Train Control, which is a system that automatically overrides humans to slow down a train that is operating above the speed limit. At the rate that we are going now, this system will save lives every year. Congress-persons have been pushing for this and obviously it is worth pursuing for that reason alone. But meanwhile, speeding motorists are probably killing 10,000 times that number of people per year. Where is the political conversation about establishing a system for motorists that will automatically prevent them from driving above the posted speed limits? A positive train control for drivers would save tens of thousands of lives a year. And no one would dare even suggest it. At a minimum, where is the political will to establish stricter enforcement against speeding? If anything, politicians are standing in the way of this goal by preventing speed camera enforcement. As a society we seem to passively waiting for self-driving cars to save us? But if anything, legislation is standing in the way of that too.

    The human mind is a curious thing. Commensurate with fatality and injury statistics, for every article about train safety, for every press release from elected officials calling for positive train control or greater rail safety, there should be 1,000 articles and press releases looking into what can be done to pursue motorist safety. Is licensing up to par? What can be done to tweak public messaging campaigns? What is the latest in enforcement? Etc. Etc. Etc. One thousand articles and press releases. But no one has the mental capacity for that.

    Given the order of magnitude of the danger posed by motor vehicles, the single best way to pursue public safety is mode shift – shifting trips made by people in motor vehicles onto trains, buses, walking, biking. It’s why New York has the safest per-capita driving rate in the country. Everyone rides the subway. How many lives does Amtrak save per year by getting people out of their cars? It’s impossible to say, but it might be able to estimate based on passenger-miles traveled. What IS possible to say, and what everyone talks about, is how many people Amtrak kills per year, and by focusing on that number we’re perhaps focusing on the wrong solutions.

    So for the sake of safety, people should be encouraged to ride the train. I fear that unfunded mandates, even for safety like Positive Train Control, may result in the opposite. If Amtrak and commuter railroads have to raise fares to pay for Positive Train Control, and some proportion of riders shift away to make the trip by car instead, and those people crash their cars proportionally, the unfunded mandate of PTC will have the perverse effect of eroding safety. I guess it would take an academic study to figure that out.

  9. It seems that PTC consists of multiple, complementary technologies. Is the GPS part separable? Tracks don’t move around, so using a GPS-based signal to compute the maximum speed for an area (and upcoming areas) seems like a less expensive proposition and potentially shorter-term milestone on the way to full PTC.

    Is there a case for partial PTC via GPS, as a well-defined phase of full PTC installation?

  10. Pingback: What the Amtrak 501 Derailment Says About America’s Failure to Prevent Train Crashes – Streetsblog USA

  11. I’m not clear what sleep apnea has to do with this. If a driver is asleep at the wheel, they’re asleep, full stop. Sleep apnea, as I understand it, affects only whether they’re able to breathe readily while sleeping. Did you perhaps mean narcolepsy instead, which causes someone to fall asleep unpredictably (and should be a disqualifier for a job like driving a train)?

  12. Most of the recent train accidents have been the fault of the train enginers who for some reason have not slowed down the train when they were supposed to and cost innocent person there lives! There should be more training involved and a certain number of hours behind the control of the train! These sensless errors in judgment cannot continue to happen!

  13. Pingback: Qué sabemos sobre el tren de Amtrak 501 | El Pasajero

  14. “While others-primarily the NTSB-will provide a full analysis in the weeks and years ahead, this is my attempt to reckon with what we know about this incident as of the same evening.”

    In that you mentioned the incident at Chatsworth, I am curious to know whether you studied that incident in-depth, an avoidable tragedy that claimed 25 lives including that of Metrolink train 111 engineer, Robert Sanchez, and maimed more than 130.

    The reason for my curiosity stems from the fact that there are five inconsistencies related to the way the NTSB’s Railroad Accident Report RAR – 10/01 was written to draw definitive conclusions regarding these inconsistencies.

    For example, the NTSB stated in the report, that the signals in question (CP Topanga – west end) were operating correctly at the time of the collision (4:22:57 p.m.) after having undergone a battery of tests. The NTSB found, based on employed testing procedures and analysis, the signals to be working properly “at the time of the accident.”

    The problem with using language like that is that “at the time of the accident” suggests a time after the Metrolink train proceeded past the signal at CP Topanga and where the collision actually occurred. With signals working properly, all signals would show red or a stop indication. That is what was found to be the case when said signals were tested.

    But, that is not what was called into question. What was really called into question is what the color of the signal at CP Topanga – west end – was when Metrolink train 111 was departing the Chatsworth Station – a time prior to this train passing that signal. Was it red as most who have an interest in this incident seem to believe? Or was it green, as four eyewitnesses that day looking at that signal from the station platform claim – in this case, two railroad enthusiasts, one security guard and the conductor on train 111 that day. What should have been tested for was signal color at CP Topanga when train 111 was departing the station and prior to passing that signal. It was not, according to information presented in the report.

    The first question I would have is: Why not? That information would be crucial!

    People who read and study the report in depth should recognize four other inconsistencies as well. By my count, there were five in all in the accident report write-up in question.

    The fact that the tragedy happened and mobile device use was an alleged contributing factor in this incident, it should be noted that that alleged factor was neither the only one nor do I believe in my heart of hearts, it was the main factor.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s