“When you consider investment priorities, safety comes last,” she said.
May 14, 2015 11:27 AM   Subscribe

Technology to reduce speeding trains absent from many train lines, including the section of the most recent derailment. Just one day after the Philadelphia train derailment, the House votes against increased Amtrak funding

From the first link:

In fact, the absence of the technology has come up repeatedly. Positive train control might have prevented the derailment of a Metro-North commuter train in the Bronx in December 2013 that killed four people and injured dozens, according to the safety board. Its investigation determined that the train’s engineer had fallen asleep and failed to slow a train that was traveling at 82 m.p.h. before it entered a curve limited to 30 m.p.h.

“Without it, everybody on a train is one human error away from an accident,” according to the safety board.
posted by sio42 (29 comments total)

This post was deleted for the following reason: Poster's Request -- frimble

When asked for comment, Louie Gohmert replied "Funding safety measures only makes it easier for Communist ISIS Terrorists from Kenya to sneak across our undefended border and sap our precious bodily fluids."

Okay, he didn't. But no one would be at all surprised.
posted by delfin at 11:34 AM on May 14, 2015 [17 favorites]

Infrastructure, schminfrastructure. Let's throw more money at this fighter plane which doesn't work that nobody wants.
posted by fifteen schnitzengruben is my limit at 11:39 AM on May 14, 2015 [36 favorites]

Plenty of money for funding airlines and airports and the military, though. Roads and tracks and education, though, not nearly important enough.

And this is why campaign finance reform is important and needed, so that we can get people who work for people instead of for corporations.
posted by mephron at 11:41 AM on May 14, 2015 [5 favorites]

Let's throw more money at this fighter plane which doesn't work that nobody wants.

You forgot the part about "...solely because after I leave office I'm looking forward to the 7-figure salary working or lobbying for the company that makes them."
posted by aught at 11:48 AM on May 14, 2015 [8 favorites]

$10 billion = safer tracks = Merck, Phizer, Apple, Boeing, Dow, CVS stock buybacks = replacement for Ohio nuke sub = graft in Indian jobs program = Vetran's Choice VA = Snapchat valuation.

Seems like the thing you'd have just lying around. The optics on the timing aren't great.
posted by Ogre Lawless at 11:59 AM on May 14, 2015 [1 favorite]

As if their week wasn't bad enough, an Amtrak engine just caught fire in Milwaukee.
posted by JoeZydeco at 12:01 PM on May 14, 2015

Positive Train Control (PTC) has been a law since September 2008, with the deadline of Dec. 15, 2015. It's expensive; from this HNTB report (PDF):
Current industry estimates are that PTC will cost between $9.5 billion and $13.2 billion over the next 20 years. This is an average of about $130,000 per mile. Estimates of the initial deployment costs are between $2.0 billion and $3.7 billion.

While the federal government has made an authorization of $50 million per year for 2009-2013 in funding to help support PTC with the Rail Safety Act of 2008, funding was not included in the Administration’s budget requests for Fiscal Year 2009, 2010, or 2011 and is absent in the 2012 budget as well.
But this is a requirement for the private train companies, not just Amtrak, to install such technology on their lines. It's not an issue of government investing in this, but companies spending money on something with a limited return on investment. So it's as much an issue of private companies dragging their feet as it is for the Feds to support safety for Amtrak.
posted by filthy light thief at 12:27 PM on May 14, 2015 [2 favorites]

(Metafilter's own) Eric Fischer has been collecting and storing data from Amtrak's realtime information feed, which includes the locations and speeds of trains, and he's just published his analysis of Train 188's speed around the time of the crash: https://www.mapbox.com/blog/amtrak-derailment/
As you can see, trains normally operate at about 80 mph west of the curve, slow to around 50 mph for the curve, and then accelerate to about 110 mph east of it. Train 188, shown with red circles and large numbers, appears to have accelerated to 106 mph before the curve instead of after it.
posted by jjwiseman at 12:52 PM on May 14, 2015 [11 favorites]

A friend's been writing about the requirement for PTC in the FT for years. He got back from the crash site the other night, and was pretty wiped out by what he saw.
posted by scruss at 12:58 PM on May 14, 2015

Why the *&^%$# does PTC cost $130k/mile? GPS will tell you where you are to close enough for speed control, there may be some reliability stuff there, but if you're just trying to keep trains from going into a 50MPH curve at 100MPH, this should be a few thousands per locomotive.

And if you're trying to keep trains from crashing into each other, take some of the dollars spent down the V2V automobile rat hole, which has tons of potential deployment issues which make it a really bad boondoggle, and play with those concepts on trains first (where you can so equip the entire fleet in a relatively short cycle, rather than a half a century).

More in the "how fast it was going" department: Surveillance video shown here, with timestamp, has the train passing in a little over 4 seconds. 7 85 foot units at 4.5 seconds is about 90MPH, it's probably not a full 4.5 seconds and I haven't got confirmation about if the locomotive counts in the 7 "cars" reported, but yeah, it's about 100MPH.

Also interesting: that curve was called out as a hazard back in 1900(!).
posted by straw at 1:24 PM on May 14, 2015 [1 favorite]

Why the *&^%$# does PTC cost $130k/mile? GPS will tell you where you are to close enough for speed control, there may be some reliability stuff there, but if you're just trying to keep trains from going into a 50MPH curve at 100MPH, this should be a few thousands per locomotive.

Is it, in fact, more complicated than that. Implementing PTC as required by law requires a degree of interoperability between private carriers that didn't previously exist, and as such has been (and will be) expensive to attain, so it's also about making every train be able to talk to every other train. It's also about being able to enforce temporary speed restrictions -- due to work on the track, weather, hazards, accidents -- on any given train for any given section of track, which requires both a communications system (to get that information out to the trains) and a control system (to make those decisions and send them out). And it's also about being able to detect and transmit how track switching is set -- GPS can tell you where you are, but unless you know which way the switches ahead of you are set it can't tell you where you'll be going.

There are some other issues -- notably, but not limited to, redundancy (what happens when you can't get switch information on a train? do you stop it? keep it going? what if you can't get accurate speed data? what if you can't get updates on temporary speed changes? all of that needs to be planned for, and programmed in) and training (since this is a substantial change for drivers).

The other issue is that if we'd been investing more in rail infrastructure over time, the sticker-shock wouldn't be nearly as bad since it would be (one hopes) more of an incremental upgrade).
posted by cjelli at 2:09 PM on May 14, 2015 [10 favorites]

Also interesting: that curve was called out as a hazard back in 1900(!).

There was also a severe 1943 accident around the same location, though speeding through the curve was not the issue.
posted by Drinky Die at 2:16 PM on May 14, 2015

So, essentially, the problem with PTC is that the perfect (as required by law) has become the enemy of the incrementally better, and V2V/I2V in trains is not just a bad idea for all the reasons that it's a bad idea in automobiles (let's mandate that cows and children on tricycles be equipped with transponders!), it's also a bad idea because range needs to be miles, not hundreds of feet.

I realize that software people think everything is easy, until they try to implement it, and I definitely fall in that category, but... Optical lane deviation warnings and follow-distance systems have been deployed in commercial trucks for damned nigh two decades now, speed limit assistance by OCRing signs is standard in some cars, sure seems like a couple of million bucks and some time with OpenCV could bring similar driver assistance to locomotive cabs relatively quickly.

Especially since you don't have to worry about LDW so much.

Hell, a couple of million bucks even if you let Booz-Allen take half for administering the federal contract.

Thank you FRA (which has also made efficient fuel per passenger-mile US passenger rail super difficult to accomplish in the U.S.).
posted by straw at 2:39 PM on May 14, 2015

I'm posting this preemptively: Why Aren't Trains Autonomous?
posted by Monochrome at 2:48 PM on May 14, 2015 [3 favorites]

So how much would it cost to put co engineers in these cabs?
posted by notreally at 4:19 PM on May 14, 2015

How America lags behind rest of developed world on train safety

Some of the content is repeated from the NYT link, but it is still interesting.

Key quote: “It’s basic kit over here. Human error like this derailment should not happen in the 21st century".
posted by knapah at 4:23 PM on May 14, 2015 [1 favorite]

And "congressional sources", including Rep. Charlie Dent, R-PA and Andy Harris, R-MD , are claiming that PTC was installed on both that stretch of track and the locomotive, but was turned off.
...Yet the technology that could have slowed it had not been enabled – allegedly because the Federal Communications Commission told Amtrak there was not enough available radio bandwidth to reliably operate the system, Harris, Dent and a third congressional source say, citing communications between the Appropriations Committee and Amtrak.
posted by straw at 4:38 PM on May 14, 2015 [1 favorite]

“Human error like this derailment should not happen in the 21st century"

Similar to the incident where 79 people died at Santago de Compostela 2 years ago. That train derailed while travelling at 111mph through a curve rated for 50mph.
Discussed previously.
posted by boffin police at 5:24 PM on May 14, 2015 [1 favorite]

Okay, he didn't. But no one would be at all surprised.

Indeed. Some have already blamed the gays.

The lunatics have truly taken over the asylum.
posted by juiceCake at 6:19 PM on May 14, 2015

I have GPS in my car. It is accurate to plus or minus 1 MPH and knows the speed limit at any point on the road to an accuracy of about 20 feet. It puts up an alarm if I exceed the speed limit. It costs a little more than $100.

This GPS accurately covers about 2,500,000 miles of road. There are only 140,000 miles of railroad.
posted by JackFlash at 6:51 PM on May 14, 2015 [1 favorite]

This is an average of about $130,000 per mile.

Maybe they should get Garmin on the job... I've got this doohickey in my car that goes *bing* when I'm going too fast for the local speed limit. Cost me $50 off Amazon.

I'm only half joking when I wonder why putting speed limits for trains on a GPS receiver couldn't be done reasonably cheaply. Or put speed cameras on the tracks, with fines going to the local city.
posted by 43rdAnd9th at 7:04 PM on May 14, 2015

I'm not remotely a train expert, but it looks like some form of automatic breaking in response to traveling above the speed limit has existed since 1923 according to this wikipedia article.

An article in today's Inquirer said an automatic breaking system was installed years ago on the southbound tracks of the curve, but not the northbound:

"A veteran SEPTA engineer who frequently operates trains on the corridor said engineers in southbound trains have six seconds to start braking after they are alerted by both lights and sounds in the train cab if they are going too fast. ... If the engineer does not slow to 45 miles per hour, the train is automatically stopped, he said. No such alerts or automatic braking occurs for northbound trains approaching the curve. For northbound trains, the speed limit for Amtrak trains leaving Philadelphia rises from 60 to 65 to 80 miles an hour before the Frankford curve, where the limit drops to 50 miles per hour. For southbound trains, the speed limit before the curve is 110 miles per hour. Boardman said trains could negotiate the curve at 80 miles an hour, so planners apparently decided it was unnecessary to install the "code change" that would have triggered emergency braking if a northbound train did not slow down from the previous speed limit."

I used to work with a guy who was very involved in the construction of the Broad Street Line cars, and he told me that US regulations require train cars to be significantly more resistant to crashes than in Europe, which leads to heavier cars and more difficulty reaching high speeds. In this case I wonder if it worked as intended and helped reduce the number of deaths - seeing those photos I am shocked that only 8 people died.
posted by sepviva at 7:49 PM on May 14, 2015 [2 favorites]

Knowing how fast the train is going and where it is and even if it's going too fast for where it is are the easy parts of the problem of how to prevent a train from exceeding local speed limits by like, an embarrassing margin. The reason why stuff like this is so difficult is because the interface between the no-speeding device's control software and the train's control software/systems has to account for every possible valid state of all of the physical components of the train and also include fault handling routines for as many invalid states as can be foreseen. This kind of thing simply takes times and costs money. That said, I know nothing about this project in specific so it may well be too expensive. I do however know all too well how easy and tempting it is to underestimate the complexity of a technology problem on the basis of it seeming easy on first glance.
posted by feloniousmonk at 10:37 PM on May 14, 2015 [2 favorites]

A suspicious man might say that it seems as though oil-funded politicians don't really want there to be an alternative to everyone piling into their own car. A suspicious man might say that.
posted by Legomancer at 5:53 AM on May 15, 2015 [2 favorites]

The Philadelphia tragedy has basically aborted the dream of American high-speed rail.
posted by Renoroc at 7:34 AM on May 15, 2015

I have GPS in my car. It is accurate to plus or minus 1 MPH and knows the speed limit at any point on the road to an accuracy of about 20 feet. It puts up an alarm if I exceed the speed limit. It costs a little more than $100.

This sounds an awful lot like "You need how many thousand dollars to replace that server? I can get a computer for $199 at Walmart!"
posted by one more dead town's last parade at 7:53 AM on May 15, 2015 [2 favorites]

On the FRA standards mandating much heavier passenger rail cars in the U.S. and safety: Europe runs .38 deaths per billion passenger miles, Amtrak .4. And that includes the 2013 Alvia disaster that killed 79 people.

I'd have to dig into the numbers more, but I'd also guess that many rail related deaths are suicides.

So, yeah: Heavier rail cars are basically killing our rail transport energy per passenger mile numbers, not making us safer.
posted by straw at 9:22 AM on May 15, 2015 [2 favorites]

Oh yes, didn't mean to argue for the heavier cars were better, and neither did the engineer I knew. There are few enough train accidents the trade off doesn't seem worth it to me. Especially when compared to car travel, trains already seem very safe.
posted by sepviva at 11:33 AM on May 15, 2015

Yeah. Heavier cars are certainly not necessarily better. Filling locomotives with concrete to meet the weight standards ceratainly isn't better.

Meanwhile, the FRA is also aware that modern car designs can dramatically improve safety in a collision. In light of these advances (which are akin to crumple zones in cars), it's pretty shocking that we're not rushing to replace or refit all of our existing rolling stock, or even requiring new cars to be fitted with these technologies.

I'm not quite ready to call out America's inferiority when it comes to overall rail safety. The passenger car weight restrictions are certainly dumb, but American signaling systems are arguably (as a whole) more advanced than most of their international counterparts, and have been for quite some time. The NEC has had cab signaling since the 1920s, which still isn't common in many parts of the world (and I find it odd that cab signals didn't prevent the recent accident -- the NTSB report will be interesting).

While passenger rail in the US generally sucks, we move a ton of freight very efficiently, across an awful lot of trackage. The mandate and deployment of PTC is a big deal, even from an international perspective. $10 billion isn't a huge amount of capital to spend, considering that it's going to affect every single mile of railroad in the US. The commuter railroads that have already been deploying ATC systems will not incur significant extra costs to comply with the PTC mandate -- any reasonably modern railroad already meets most of the criteria for PTC.
posted by schmod at 9:43 PM on May 18, 2015 [1 favorite]

« Older (there is sometimes cake and tarot readings)   |   Bones! Newer »

This thread has been archived and is closed to new comments