Philadelphia, PA/USA

[crfriend]

As many have now heard, Amtrak train 188 took the ground in North Philadelphia, PA last night a little bit before 02:00 UTC. The formal investigation started this morning, and the recovery of the locomotive's event-recorder went without incident and the data have already been read out. It's a story that breaks hearts.

Normally, nobody ever wants to be called to an accident scene because it means that there's been a tragedy. In this case, it's at least seven dead and well over a hundred injured. There are myriad reasons why accidents happen; but the reason that no-one wants to hear or assign blame for is "human error" -- and that is because "human error" is the one cause that's usually 100% preventable in mission-critical operations.

And yet we have a train entering a curve with a 50 MPH speed limit at 106 MPH -- more than twice the rated limit. These limits have "fudge-factors" built into them for sure; nobody would rate something for 50 MPH if you'd derail at 55 -- but a factor of two... Of note is that the stretches of tangent track are rated for 70 MPH (mainly as deceleration-buffers for the 50 MPH curve) and the train was at 106 through the approach distance.

There's a failure here in human terms; I just wonder what the NTSB will find. I just hope a cell' 'phone and driver inattention weren't factors. Or, perhaps I do! It might be a rallying call to make sure that such technology doesn't work at more than walking speed. Oh what a wonderful world it'd be if car-drivers -- nevermind train-drivers -- paid attention to the road ahead of them instead of their little electronic distractions. (Of note, an MBTA Green-Line driver was killed and several passengers injured when that "driver" missed two restricting block-signals and rear-ended another train a couple of years ago in Boston.)

[dillon]

I cant understand in this age of simple and inexpensive GIS equipment, why passenger trains are not thusly equipped to triangulate speed, so an alarm system can be employed and a train brought to a halt remotely and safely. As you said, this is not the first fatal crash due to excessive rail speed lately.

[crfriend]

I cant understand in this age of simple and inexpensive GIS equipment, why passenger trains are not thusly equipped to triangulate speed, so an alarm system can be employed and a train brought to a halt remotely and safely.

There are a number of factors involved in this, and the primary bugaboo that's surfaced in the notion of "Communications-Based Train Control" (CBTC) is the fact that radio communication between train and central-control -- much less train-to-train -- is spotty at best thanks to things like tunnels, general surrounding terrain, and "urban canyons". So, if one cannot count on radio as a mission-critical resource one needs to go back to what one has on hand.

On a travelling train, you have a very good idea of your speed. There's an indicator for it on the panel in front of the driver. The driver is supposed to have a pretty good idea of where he is on the road, even in lieu of markers. If there are blind turns, or approaches where it might not be possible to see -- and stop to avoid -- another train, there should be track-side signals to prevent "conflicts" from occurring. In all modern systems, those signals are brought into the cab and displayed for the driver to see. In many instances, speed limits may also be displayed along with the controlling signal aspects.

CBTC is the "Holy Grail", and like the good lads from Monty Python, we are likely doomed to spending our time seeking it out instead of finding something "Good Enough." We've actually got something that, for the most part, is "Good Enough" and it doesn't depend on radio (which, recall, can be jammed and/or spoofed); we may just need to augment it.

My heart goes out to those affected by the recent calamity. I just hope it was a sincere human-factors problem and not something as stupid as inattention to detail.

[r.m.anderson]

From the news media (NBC-TV) it was reported that not all portions of the Northeast Rail Corridor have the safeguards installed especially
the curve in Philadelphia where this occurred.

Important Legislation is being hung up in Congress as Democrats and Republicans argue over further financing of Amtrak.
Democrats budget 2.1 Billion $ and Republican 1.1.

While this appears to be an engineer problem - failure to control the train - this area did not provide that engineer with the safe guards
to alert of approaching a reduced speed zone !
Cheap skate $#%@*& infernal Government trying to run a railroad on the extreme cheap side and like the recent air event with the
GermanWings airplane where is the 2nd person in the cab of the locomotive to keep the level of conduct on a safe business level.
I am not implying that the engineer is/was suicidal but with a 2nd person in the cab/cockpit accidents maybe reduced to ZERO !
AND if the necessary safeguards were installed and used these problems could be reduced to near ZERO !

My soapbox tirade ends.

[skirted_in_SF]

The driver is supposed to have a pretty good idea of where he is on the road, even in lieu of markers. If there are blind turns, or approaches where it might not be possible to see

Rail engineers are tested and qualified for every segment of track they travel. If a train is rerouted off it's usual routing, a qualified engineer is provided by the owner of the rails to guide the train on that segment.

[Uncle Al]

Carl,

You mentioned CBTC(Communication Based Train Control). What about
the DCC(Digital Command Control) that model train enthusiasts use?
Block detectors? All of these 'controls' are operated via signals transmitted
through the rails.

I think(I know, that's dangerous ) that Amtrak should look at
the logical option of DCC. CBTC is prone to problems which DCC
does not have. The only way a DCC train can lose signal is if there
is a break in the track. The train would continue until it is signaled
to do something different. "Block Detectors" could trigger any signal
to the train that is required for safe train operation.

I'll put my away but that's MY $.02 worth

Uncle Al

[crfriend]

You mentioned CBTC(Communication Based Train Control). What about the DCC(Digital Command Control) that model train enthusiasts use? Block detectors? All of these 'controls' are operated via signals transmitted through the rails.

Actually, most vital signalling systems use the tracks as the communications medium, and this goes all the way back to DC-energised "track circuits" designed to control wayside signals to indicate whether there was a train in the block (the length of rail so carved up into circuits) ahead of you.

Most everything now is welded-rail and uses AC signalling (which is attenuated out over distance so adjacent blocks don't interfere with each other) including the in-cab signals which use another carrier modulated onto the rails by wayside gear that calls out signal aspects and speed limits. These are capable of working in zero-visibility conditions, but generally speaking one wants the driver to be able to at least slow down if there's something large blocking the rails ahead of him rather than navigating "by ear".

[Uncle Al]

If the signals are properly sent, the 'engineer' would only
be there to "Monitor" the signals. The train would be
operating without human control.

Uncle Al

[crfriend]

If the signals are properly sent, the 'engineer' would only be there to "Monitor" the signals. The train would be operating without human control.

That can produce other, unintended, consequences. Take, for example, the diminishing role of piloting in the commercial aviation sector; the upshot of having pilots monitoring the machines is that pilots forget how to operate the aircraft in manual mode and crash perfectly good fully-functional aeroplanes (The Asiana 777 crash on the West Coast a while back springs to mind).

If the intent is to take the human out of the loop, do so completely. Half-doing it is a recipe for problems.

There's also the notion of liability -- and that's going to come to a head sooner rather than later in the guise of, "Who is at fault when a "self-driving car" hits and kills somebody? Is it the manufacturer? The company which designed the software? The owner of the device in question? That's just for some hypothetical piddly car-crash. The debacle in Philadelphia will wind up costing millions of dollars, not tens or hundreds of thousands. With nobody at the controls, who pays for it? (Hint: It won't be the manufacturer.)

It's now a question of sitting down and patiently waiting for the NTSB to finish their investigation and publish the results. They're remarkably good at what they do.

[Uncle Al]

It's now a question of sitting down and patiently waiting for
the NTSB to finish their investigation and publish the results.
They're remarkably good at what they do.

And remarkably slow. I've seen what they're capable of
and waited 3 years for the results of an 'accident' which
killed a college friend. My oldest son is named after him.

The NTSB usually ends up calling the incident Pilot Error.
It's a good catch-all label

Uncle Al

[crfriend]

[The NTSB usually ends up calling the incident Pilot Error.
It's a good catch-all label

I'm sorry to hear of the case of your college friend.

In this case, it's important to read the actual report, not what gets sprayed out by the mass media which has an agenda all its own.

"Human failure" (aka "pilot error") can happen for any number of reasons. A classic case of this is the flight that went down in the Florida Everglades many years ago now where everybody in the cockpit was so focused on what a light-bulb should have been doing that nobody was actually flying the aircraft, and by the time the ground-proximity warning sounded it was too late. That was listed as "Controlled Flight Into Terrain" ("pilot error" in the mass media) but got a strong callout about the failure of the entire crew and not just the Captain. That accident brought about the common use of "Cockpit Resource Management" that's in use today and which guarantees that *somebody* is actually flying and not trying to fix the problem.

One needs to actively dig to find the actual data; if one relies on what the mass media feeds us we're doomed to failure because by nature we are pre-disposed to *blame* some*body* for calamity -- and the placement of *blame* certainly dovetails in with the need of the mass media to sell advertising. A juicy story always sells more than a rather dry piece that actually points up important factual evidence.

Sometimes we'll never know what happened in a particular incident. However, in most, we can get to the bottom of the matter and try to learn from our mistakes. I love the fact that PTC ("Positive Train Control") wasn't deployed on the stretch of track that contains a particularly nasty curve -- a curve that's among the sharpest on the Northeast Corridor mainline -- and even with the bodies still warm the elites in Washington are voting to reduce Amtrak funding yet again. This is the "I've got mine, go screw yourself." mentality that the Elites in the USA now have and who hold power. They don't use Amtrak, why should they pay a dime toward it? Private jets and helicopters work; the plebes be damned.

But yet, in this case, what happened to a fairly young chap who was chasing a dream to become a train-driver to make him lose sight of what his charge was doing underneath him? After all, this is somebody who had not only glimpsed a dream, but actually went after, and captured it! (He is not so different from I in that regard.) Was he distracted by something? Was something eating at his mind and troubling him to the point where he could not function rationally? What's *behind* the "blame"? That's what needs to come out into the light of day.

I've had enough of crucifying individuals; I'd had a bellyful of that at my last sad place of employment where "Root Cause Analysis" translated into "Club The Survivors". It's time to listen to the very quiet signal that points up truth in preference to the "everything louder than everything else" world of the mass media. The world's a quiet place, save for a few loudmouths; take the time to listen carefully. You may be surprised by what you hear.

[dillon]

Clearly we don't want trains operated remotely. There is always the risk of malfunctions that could be disastrous or of hacking by malicious jerks or terrorists. But obviously there needs to be some remote monitoring and possibility of control remotely, if only to slow or halt the train. So both ends of the system should have override capability, though only to slow or stop the train until an issue can be resolved. And probably, having a co-engineer, as aircraft have copilots, seems also in order.

[dillon]

I should have continued my previous post to say that such systems are needed if the US is to ever make progress in rail transit like the rest of the advanced economies on this planet. With mag-lev trains able to reach 300mph plus speeds, and increasing congestion and delay, not to mention discomfort, in air travel, we need to look again at rails. Ironic that gas pipelines can be planned, routed and constructed with officials waiving most of the normal public hearing and oversight processes, but mention trying to obtain right-of-way for rail and it's Katy-bar-the-door!

On the issue of airline discomfort, I work in an intensively agricultural region of the US known for swine farming. Pigs, when transported on that class graduation trip where the buses go out full and come back empty, must be allowed a certain square footage of space by regulation, or else it is considered animal cruelty. Ironically, no such regulations appear to be imposed on the airlines for transporting humans!

[Caultron]

Assorted thoughts:

1. Automating the entire US rail network would be a massively expensive job, and it's not clear who would pay for it. And if it ever got hacked, the entire US rail network could be disabled.

2. Cities in the United States tend to be too far apart for rail travel, even on bullet trains, to be attractive.

3. Excluding subsidies, it's not clear that bullet train tickets would be cheaper than plane tickets.

4. I don't believe Amtrak owns any actual track. The freight railroads like BNSF, Canadian National, Canadian Pacific, and Union Pacific own the tracks and Amtrak just pays to use it.

5. Maybe they should work on better control of the engineers rather than control of the trains or tracks.

[Uncle Al]

To think that the railroads helped to build America.



When/Where did we(America) go wrong?

Trains in China vs America 2015-05-15.jpg

Uncle Al
Train Buff