In my normally bucolic corner of Los Angeles, a northbound commuter train just had a head on collision with a southbound freight train. Numerous fatalities, an army of firefighters, EMS, cops and reporters.
The commuter train was going around a bend and then through a tunnel, as it headed towards the pass through the Simi Hills. I’m guessing the tunnel and/or the curve contributed to the accident.
I’m interested in what happened and why. What has to go wrong for two trains, going in opposite directions and using the same track, to collide?
Aren’t there numerous fail safes built in to the system, warning bells, alarms, requirements for visual confirmation? Can it all be put on operator error?
Yes, there are all these things, but they can and do go wrong. It is far, far too early to speculate on causes for the Californian accident, but the only thing I would say at this stage with any confidence is that there was probably more than one thing that went wrong.
Could have been dispatcher error, could have been driver error, could have been both, or neither. Systems fail, people forget or assume, sometimes there is malicious intent, etc, etc.
Danger Ahead! is a good site if you want to learn what causes rail accidents, and analysis of historical mishaps.
Check back in a few months’ time when the reports start rolling in. They are always fascinating reading.
I work for a freight railroad but we often cross other railroads territories.
Most railroad around here is governed by ABS, which is automatic blocking system and is signaled either in one direction or both. You have “interlockings” or “control points” where trains are held or diverted. And in between control points are automatic signals for each “block” of railroad. Each automatic signal gives you the condition and occupancy for the next two blocks, which are usually a mile or more long. If a train were two blocks ahead, you’d get an approach signal, warning you that the next signal might be a stop signal.
There’s also “Dark territory” which is not signaled at all. In this case a dispatcher would give a train written permission, usually copied via radio, to occupy a railroad. For Norac, this is called a Form D. Two trains cant have a form D for the same piece of railroad so in order for that accident to happen either one of the trains didn’t belong there or a dispatcher made a big mistake.
Another factor is that each railroad seems to be running things different than the other. Most railroads operate under NORAC rules. But I personally operate on five different railroads with three different sets of rules governing them.
And finally there are trains going right through stop signals and crashing into other trains. That usually happens if the crew happens to have fallen asleep. Or are half asleep. You’d be surprised how many people can and do run a train while sleep deprived.
In our case, we regularly work 12 hours or longer per shift and are called back to work as soon as we’re “rested”, which is ten ours after we walk out the door for home. I work an extra board, which doesn’t give you any off days. If we work 7 0r 8 days in a row (i forget which) we’re entitled to 24 hours undisturbed rest. It’s so sweet of them. I personally take off whenever I feel like it. …Didn’t mean for this to become a rant.
Thanks for the informative response. I will check out that site, and along with everyone else, I’ll be paying close attention to the post mortem.
I work in transportation planning for a big local agency, almost entirely with buses. Our particular office doesn’t schedule any subway trains. Extra boards and fatigued operators seem to be an issue across the transportation industry. I wonder if the airlines have some version of this issue too?
Sleep deprivation is a concern in so many industries. In TV and film production, we’d sometimes be on our feet, working for 16 to 20 hours. Then we’d have to drive home. Needless to say, some of us didn’t make it. On some shows, the producers would make a hotel room available if you needed it, but there was too much stigma attached to taking them up on their offer.
While it’s too early to tell, this LA Times article implies that the only indication that a train needs to pull over and let another train pass is a trackside signal. Seems a little low-tech in this day and age.
I don’t use many trains, but this makes me want to research what stretches of railroad are single-track and avoid those routes.
“Trackside signals” are what you have for another form of transportation - the car. Only there, we call them traffic lights. In theory at least, the railroad signal is safer because it is used only by professional drivers, but mainly because it performs the extra function of keeping trains on the same path apart (road traffic lights only keep cross-traffic apart from you, but leave it up to you to deal with your own street). Of course, there are still accidents occasionally. The best form of safety would be even lower tech (dual track everywhere and grade separated junctions), but in the end, it comes down to money rather than technology. There is high-tech stuff out there (Google “Automatic Train Protection”) like in-cab signalling, brakes that “trip” into full emergency mode if they pass a signal at danger, driver vigilance systems, etc, but how much you get depends on the traffic the line has.
I don’t think I’d be avoiding single track lines just yet, over one incident. Rail is still one of the safest forms of transport. Of course, if you are really worried, avoid the first and rear car (trains have “crumple zones” like cars, and a certain death toll is deemed acceptable for specific collision types).
Whatever the proximate cause, the underlying problem is too many trains and not enough tracks. As commuter rail and short-haul intercity train travel become more and more popular, the traffic expands tremendously–not only in the number people using the trains, but in the number of trains scheduled. Also, last I heard, the freight railroads have doing pretty good business lately, so that means more freight trains, or longer freight trains as well.
In 1975 Amtrak scheduled just three daily round trips between L.A. and San Diego, using the old Sante Fe tracks. There are at least six or seven, with more on some days, but all still using the same rails, which are single tracked in many places. North of Los Angeles, where the yesterday’s crash occurred, traffic is increasing as well, and Amtrak, Metrolink, and Union Pacific are all sharing the same tracks in many cases.
Two observations here: First, I’m surprised it took this long to happen.
Second, if I had to pick out a place in all the world that would become known for horrible passenger train disasters, I wouldn’t have picked Southern California!
According to current information, the driver of the passenger train missed a signal and plowed into the freight train at almost 40 MPH. He was sending text messages to a friend less than 60 seconds before impact. 25 confirmed dead, 135 wounded, dozen still in critical condition. They have stopped looking for bodies.
From public accounts, I gather it was SOP for the commuter train to stop, wait for the freight train to clear that section of track, and then continue.
I live right by a Metro Link station and work right near the closest Metro Link station to the accident.
I don’t know who designed the warning systems for the crossing, but these systems suck.
Near my house the ML station is about 100 yards N of where it crosses the street. A northbound will cross the street and stop at the station. The crossing gates will stay in the down position until the train pulls out of the station. :rolleyes:
In Chatsworth the station is aout 1000 feet south of Devonshire St. A ML train pulling into the station an stopping will cause the gates at gates on Devonshire to close for about 1-2 minutes before the system figures out that the train actually did stop in the station.
IMHO this does not promote safety but rather works against it.
Now I could be full of shit, but it seems to me that a half way competent hacker armed with a TRS-80 could figure out a much better system.
Off the top of my head it does not seem to be rocket surgery for the system to understand that a train is stopping. But hey, what do I know?
FWIW, the NTSB is treating this report with caution. Apparently the same thing happened in Boston after an accident and the reports of the engineer text messenging were inaccurate.
If engineers miss or ignore signals (and if backup systems either don’t exist or fail to detect this), problems will happen at almost any level of traffic density. Unless those problems are addressed, you won’t achieve acceptable safety by building more track.
Metrolink has publicly taken responsibility for the accident.
It isn’t clear when the text messages were actually sent. The messages were received, by the teenage frothies, two minutes before the accident.
It does seem from the reports, that the lives of hundreds of people were at the mercy of one missed signal. I had thought there were more built in fail safes than that. Given that the current system is perfectly safe, over 99.99 % of the time, the decision not to institute more safety measures seems to be based on a cost benefit analysis.
Yep, “too many trains” is no excuse. Too many trains may slow things down, but with a safe system, the level of traffic has nothing to do with the level of safety. Once you have more than one train, you need a system for making sure they never occupy the same section of track.
TheLoadedDog’s traffic light analogy is a good one, except traffic light style protection isn’t really good enough for trains. Safe rail operations don’t trust the driver to see the red light and stop. They install safeguards, from a simple device that derails the train before it can enter the track with the other train potentially coming the other way, to elaborate protection systems that monitor the train’s speed and the indications of the signals and stop the train before it gets to the red light. No system is fool proof, but an effective system of safeworking can come very close to guaranteeing that a head on collision never occurs.
A crash like this one represents a failure of the entire safety system of the railroads in question. How good was that system to being with? Were there the engineering systems in place to prevent this or were they relying on the driver alone to see the red light and stop? Given that Metrolink had another head-on back in 2002, I’m betting the latter.
Mixing freight and passenger trains decreases efficiency, but might actually increase safety in a way. You can keep pumping more and more trains onto a given section of track up to the point where train headways are nearly but not at the limit of safety. Train headways are determined by the acceleration and braking (especially the latter) capabilities of the heaviest, most sluggish train (the weakest link) that ever uses the line.
If you have a rail system used only by relatively light, high acceleration and braking vehicles (say, a subway in a major city), you can bring the trains closer together. But add just one multi-thousand ton freighter into the mix (even if it’s only one a day or one a week), and the entire thing will be based on that freighter’s braking capacity. This means that the lighter commuter trains running the rest of the time might be, as an example, ten minutes apart rather than two minutes.
When we take our freight trains onto Amtraks railroad, we have what are called LSL’s. Amtrak requires them. I think it’s short for Locomotive speed limiters. What it does is makes sure that you’re conforming to the signal. So, if you get an approach signal (meaning that the next signal might be at STOP) you must conform to a certain speed withing a given amount of time. If you don’t, you get a penalty brake application (the train stops itself).
I don’t know how the crew on the freight train made out but freight trains hace considerably more weight behind them and are built to take a pretty decent impact.
On our trains there are almost always two people on the head end. Each must individually call the signals as the see them and confirm the signals with each other. They’re both supposed to be qualified on the physical characteristics of the railroad on which they operate. If the engineer fails to conform to the signals, the conductor is supposed to stop the train.
In passenger trains the conductors role is in the passenger cars I suppose.
I heard radio interviews where the teenagers in question stated they received text messages from Sanchez, the engineer. Considering that they were fans of Sanchez and fiercely defended him, I don’t get the impression they were lying, or had a reason to lie.
This is a very interesting thing in the way that danger can stem from human behaviour. In my city, there was a major rail crash about five years ago in which a train took a curve at an insanely high speed and derailed. It transpired that the driver had suffered a heart attack, and being a heavy man, the weight of his legs kept the deadman pedal in position after he lost consciousness or died. The focus then shifted to the guard (conductor) of the train. He was expected to apply the emergency brake if there was no response from the driver’s cab or if he felt the train was being operated in an unsafe manner.
But he didn’t. And the question was “why?”
Turns out, the guard - depending on whom you talk to - is the “boss” of the train. Driver is merely the controls monkey, but it’s the guard’s train. Nice in theory, but in practice the driver is more senior. In terms of employment, you become a guard first, and then a driver, and the driver thus tends to have higher seniority and a bigger pay packet. Also, he’s the driver, dammit.
The crash enquiry was scathing about this system; a system in which a young recruit guard off the street would be expected to question the skill of a driver who would likely be an older man and possibly have decades of experience, and for the safety of the train to be riding on this. It’s a bit like expecting a supermarket grunt to check the CEO’s bags at the checkout because he’s been told “no exceptions”. Chicken Wire? may be able to correct me on this, but my understanding is that the system is the same as it ever was, notwithstanding the inquiry.
