Amtrak 501 Derail in Washington State

[PNWRMNM]

7j43k

Hiatt:

"These guys were trained in darkness.  All of them.  They couldn't get availability to the track during the daytime, so that's part of the factor."

Except that the crash happened 19 minutes before sunrise.  That would appear to be the same darkness that they were trained in.

 

Ed

 

Ed,

I agree with you about the dark. In fact, I lean toward the idea that training in the dark would be best as it focuses attention on the details of the line.

I agree with Hiatt that each man should have made at least three descents of each of the two hills. This route goes from about 300 feet to near sea level on steep down grades at each end to an absolute stop at Tacoma Depot on north end, and a conditional stop before entering BNSF main at South end. Those are not forgiving line segments and totally unlike anything between Seattle and Portland on the BNSF main.

Operating this line safely is all within the capability of any engineer familiar with the line, but is a trap for the untrained, unfamiliar or unwary.

Mac

[Deggesty]

Apparently whoever was at the controls missed the warniing signal two miles ahead of the bridge--and he did not recognize anything else that cold have told him that he must reduce the speed of the train.

I have a real life story of a bus driver who missed a turn because a high hedge that he had used as a sign to turn (instead of a flashing overhead light and a wayside sign which warned of a right angle turn ahead in the highway) had been severely pruned to be a low hedge by one of my brothers. The bus started up the alley by our house before the driver realized that his personal marker was gone. As it was, there was no damage except, perhaps, to the driver's ego.

[Euclid]

PNWRMNM

 

Euclid

I understand what you are saying.  The over-speed indicator does not indicate speed limit.  In thinking about this further and reading what you said and what Overmod said below, I now realize that the overspeed indicator only indicates when the top speed of the line is exceeded. 

However, I am considering your point about the speed limit, prior to the curve limit of 30, not being known. What difference does this really make to my speculation on the engineer's reaction to the speed warning that I wrote above in the 6th post from the top?   You seem to be objecting to my comments there in your post which follows mine.   Obviously the engineer was exceeding the maxium speed limit on the line.  What difference does it make if the speed limit where he received the warning is actually lower than 79?

 

 

 

What I am objecting to is that in your post of 12-23 at 9:42PM you are assuming facts not in evidenece, specifically that the speed limit approaching the curve was 79 MPH.

If the speed limit was 60 MPH, it implies to me that the engineer was lost/distracted for longer than if the limit was 79.

The main issue is what went wrong.  The TV story about the Safety Investigator who was a locomotive engineer is highly, and in my opinion rightly assuming his tale is correct, critical of ATK's training practices as actually conducted on this route.

If what he is saying is true, I am not much suprised with the notion that the engineer was lost. If what he is saying is true, ATK bosses should be criminally liable for failure to train their engineers sufficiently. I also suspect it is likely that WSDOT paper pushers were in a "hurry up" mode that may have contributed to ATK training failure.

What the engineer did in the last few seconds is irrelevant EXCEPT that if he made a light application of the independent to shut up the overspeed indicator, it is further evidence that he was lost. By that time the curve was staring him in the face and should have gone to emergency, but it was too late to do anything to save the situation.

Mac

 

All I have done is offer a scenario that plausibly fits what we know.  It explains why he only made an application of the independent brake when he was actually in a situation that called for an emergency application.  My central point is that he either did not know that he was about to enter the curve at the 30 mph restricted zone, or for some other reason, he was heedless to the need to slow down for that curve.    

I agree with your point that I don’t know what the speed limit was when he received the over-speed warning.  However, that information is not at all essential to the point I was making with the plausible scenario, so I have rewritten it to eliminate any reference to the actual speed limit leading up to the 30 mph limit at the curve.  The main point is that he was going 80 and unprepared to slow down to 30 within 700 feet.

Also part of the point of this plausible scenario is to consider the rather incredible irony of being warned of an over-speed condition, responding to that warning, and yet not realizing that he was actually in a dramatically more serious over-speed situation at that very moment.

[Paul of Covington]

Deggesty

Apparently whoever was at the controls misesd the warniing signal two miles ahead of the bridge--and he did not recognize anything else that cold have told him that he must reduce the speed of the train.

   Johnny, I think you hit the nail on the head.   With four(?) people in the cab, including trainees, there was likely to be discussions going on, and in the few seconds that the diagonal "T30" sign went by, eveyone was distracted, and nobody noticed it.   When I am driving and there is someone else in the car, I have a hard time carrying on a conversation and paying attention to the road.

   On our local highways, there is usually a "SPEED ZONE AHEAD" maybe a hundred yards or so before the sign indicating the speed limit, but I noticed something else many years ago.   If the speed limit is 65 MPH, and there is a 45 MPH lmit ahead, the highway dept. usually places a "65 MPH" sign about a hundred yards before the "SPEED ZONE AHEAD" sign.  At first I used to wonder why they bothered to post the speed limit when it was about to be changed, but then I realized that the first sign triggered my brain to be on the alert for the second one.   I wonder if it would be worthwhile to put a sign before the diagonal "T30" saying "SIGN AHEAD" OR "BURMA SHAVE" or anything or nothing, but just something to get the engineer's attention.   Chances would be a lot less that he would miss both of them.

[charlie hebdo]

Overmod

I could almost stop reading this story at John Hiatt's 'credentials' - but I am glad I stuck around for the Amtrak material at the end.

Best to wait for the NTSB formal report than to pile Hiatt-style BS deeper and deeper ... we have enough of that tendency here already.

 

 

More than enough of that.  One thing occurs to me.  The new stretch of track was how long?  It seems like it was only a short segmant in a longer SEA-EUG route already well known by engineers.  I would think training on this short segment alone would have been fairly quick to master for an experienced engineer.

[Lithonia Operator]

There were reference to 4-6 people in the cab during training. But IIRC, the NTSB has said there were 2 during this accident. Now, that could mean “at least 2,” but that’s not how I took it. My understanding at this time is that two persons were in the lead unit’s cab.

[Lithonia Operator]

It looked to me that there was a height-transition aspect to the car in front if the trailing unit. Pretty ugly, IMO. Is that type of equipment very common? And, is Talgo equipment particularly suited to this line, or is it more a case of that’s what they got a deal on?

[BaltACD]

The individual operating the controls was LOST both situationally and in fact.  All the data point to that fact!

[7j43k]

lost:

1.  not having a clue where one is, and wishing to know

2.  thinking one is in one place, but being in another

3.  not even wondering where one is, and being in the wrong place

 

Are there more varieties of being lost?

 

 

Ed

[BaltACD]

7j43k

lost:

1.  not having a clue where one is, and wishing to know

2.  thinking one is in one place, but being in another

3.  not even wondering where one is, and being in the wrong place 

Are there more varieties of being lost? 

Ed

And they ALL apply to who was at the operating controls.

[Overmod]

7j43k

Are there more varieties of being lost?  

Oh yes.  There is the variety of 'lost' correlating to having something, like a perfectly fine SC44 or Talgo consist, and then suddenly not having it available any more...

[Gramp]

Wonder if one of the Midwest Talgo's that are in storage will be heading "north by northwest".

[PNWRMNM]

LithoniaOperator

It looked to me that there was a height-transition aspect to the car in front if the trailing unit. Pretty ugly, IMO. Is that type of equipment very common? And, is Talgo equipment particularly suited to this line, or is it more a case of that’s what they got a deal on?

I am 99.9% certain that Washington and Oregon DOTs bought Talgo equipment because they are marginally faster on some curves. I  rode them and know that some curves between Seattle and Portland are posted something like this; T 70, P 60, F 50. Check ATK schedules to get an idea of how much faster they are than trains 11/14 which are conventional high level equipment.

There are two components to the answer to the question "How fast can/should I go around this curve?"

The answer to "can I" is that I do not want to have my equipment roll over and fly through the air. The math is reasonably simple vector analysis. The key variables are sharpness and super elevation of the curve, and height of center of gravity of the equipment. I am confident that the CG of Talgo equipment is lower than that of the locomotive. This wreck demonstrates the truth of that statement since the locomotive appears to have gone flying first, dragging the first part of the train off the track with it. Speed of the Talgo trains is thus controlled by the height of the center of gravity of the locomotives.

Why then are Talgos allowed higher speed on some curves? How fast "should I" go around this curve? The answer is that they tilt the carboday to minimize the passenger's perception of centripidal force, that is the force tending to move their body to the outside of the curve. The tilt does nothing to reduce the forces acting at the rail and wheel, in fact the shift, by moving the center of gravity toward the outside rail tends to reduce the overturn speed. This is offset by the low center of gravity.

What is controlling the speed in routine operation is passenger comfort, that is how much sideways force they will tollerate before becoming quasy or uncomfortably aware of the speed.

Mac 

[Overmod]

Gramp

Wonder if one of the Midwest Talgo's that are in storage will be heading "north by northwest"?

I am more interested in seeing what happens with the set that was in the derail incident, which we were discussing being trucked back for rebuilding.  That may have become a much higher-priority repair!

[Euclid]

BaltACD

 

7j43k

lost:

1.  not having a clue where one is, and wishing to know

2.  thinking one is in one place, but being in another

3.  not even wondering where one is, and being in the wrong place 

Are there more varieties of being lost? 

Ed

 

 

And they ALL apply to who was at the operating controls.

 

It is also possible that none of these applied.  He might have known exactly where he was on the map, and yet have fogotten about the curve and its speed zone.  He might have forgotten about that after being made aware of it in passing the 2-mile advance warning.  Or he may have forgotten about it days eariler and then missed the 2-mile warning in a moment of distraction. 

Asside from the capabilities of PTC, why isn't there a realtime display on a video screen in front of the engineer that clearly lays out the line, speed limits, and important features?  I assume that this would be part of PTC, but couldn't it have easily been implemented by now as a separate feature without having to wait for PTC?  Why coulnd't it just count distance from origin so it always shows current location?

[Norm48327]

Euclid

BaltACD

7j43k

lost:

1.  not having a clue where one is, and wishing to know

2.  thinking one is in one place, but being in another

3.  not even wondering where one is, and being in the wrong place 

Are there more varieties of being lost? 

Ed

And they ALL apply to who was at the operating controls.

It is also possible that none of these applied.  He might have known exactly where he was on the map, and yet have fogotten about the curve and its speed zone.  He might have forgotten about that after being made aware of it in passing the 2-mile advance warning.  Or he may have forgotten about it days eariler and then missed the 2-mile warning in a moment of distraction. 

Asside from the capabilities of PTC, why isn't there a realtime display on a video screen in front of the engineer that clearly lays out the line, speed limits, and important features?  I assume that this would be part of PTC, but couldn't it have easily been implemented by now as a separate feature without having to wait for PTC?  Why coulnd't it just count distance from origin so it always shows current location?

I see your point but your analysis fails to take into consideration human failure but you assume that computers can overcome that. Not until artificial intelligence is perfected beyond reproach will your pipe dream come true.

[Overmod]

Norm48327

I see your point but your analysis fails to take into consideration human failure but you assume that computers can overcome that. Not until artificial intelligence is perfected beyond reproach will your pipe dream come true.

That is almost the utter antithesis of what he really said.

A moving-map display showing a train's position (derived from sequential GPS fixes, relative to a map derived from GIS sources) has been practical for some time, and it is possible (as I noted earlier) to incorporate custom data or metadata to track proximity and speed of approach to 'points of interest' just as existing train-management systems can use it to determine when the end of a consist passes a restriction.  Most of the resolution, jitter, etc. problems that make such a system unreliable for positive train-control purposes (e.g., failure to resolve adjacent track centers correctly) don't apply to longitudinal distance to fixed landmarks, and a GPS system inherently 'calibrates itself' relative to the map coordinate system used, so there is no need for external transponders or markers  or signs for the system to work properly.

It is not, of course, a 'replacement' for an attentive engineer, and it will not 'substitute' for one any more than Tesla's AutoPilot would.  On the other hand, it would assuredly have caught the missed pending speed-reduction notification at Lakewood, and could easily be designed to provide persistent and augmenting alerts (or, if enabled, to actuate the brake a la ATC to bring the train to a stop, "penalty" or otherwise, if minimum detected safe response distance were violated). 

None of the necessary systems integration is particularly difficult; in fact, most of the electronics are already OTS available.  Were "PTC" not mandated as a single system, a device like this would be a useful subsystem worth implementing many places PTC is not required.  I would, on the other hand, be interested in your explanation why a simple position awareness reminder device is supposed to require AI 'perfected beyond reproach' or constitutes some kind of 'pipe dream' to implement.

[Euclid]

 

Yes, I am not looking for something that is so perfect that it can drive the train on its own and prevent every conceivable mishap.  But engineers are supposed to “know the territory” and we have seen problems with that.  Knowing the territory does not mean just going for a ride over the line a couple times to become familiar with it.  It means memorizing nearly every detail of mileage, stations, curves, grades, bridges, crossings, signals, etc.  So it all is in your mind for reference to any detail.  That is fine, but why not complement that human memory process with a real-time reference chart so to speak?  This was probably possible 15 years ago and only a tiny project compared to PTC.  It would speed up knowing the territory, and would be a natural safety device particularly for passenger trains where a lot of lives are at risk.

 

[Anonymous]

Euclid

This was probably possible 15 years ago and only a tiny project compared to PTC.

Intermittent train protection systems are in serve since the 1930s. It started with the Indusi system. With edition I60R in the 1980s it was able to deliver the protection you ask for. Not continously controlled by a computer but initiated by trackside inductors/antennas:  https://en.wikipedia.org/wiki/Punktf%C3%B6rmige_Zugbeeinflussung
Regard, Volker

[Euclid]

I am not even talking about monitoring conditions or variables along the route such as weather or train performance.  I am only suggesting something better than committment to pure human memory, which is fine after it has had enough time to be developed after many, many trips. 

Basically what I propose would be like a rolling chart of fixed information that unrolls as a changing displays that matches the train's location.  There is little or no imput from any trackside devices or onboard detectors.   

The alternative seems to be to give the new employee all of the route information, and make a rule that he must memorize it.   

[Overmod]

Volker, you haven't been listening.  We had inductive train control worked out long before InduSi, and in fact mandated ATS, or something better 'to be developed' as a result of the mandate, for passenger trains shortly after the First World War.  By the 1930s we had moved on to continuous code-based train control via multiple frequencies in the rails, which is inherently safer than any inductor-based intermittent system could possibly be -- this conclusion was reinforced quite recently with the multiple-unit collision about this time last year. 

But what Euclid is discussing is a system that requires no fixed inductors or magnets or balises or, in fact, any wayside markers or tercom optical tracking.  It's just the moving map from a system like LEADER, which is (or can easily be) coded to contain most-economical throttle up and down for train management based on upcoming grade profile (stored) and consist details (encoded, then stored).  It is literally trivial to add fixed restriction points to the metadata behind the moving map's GIS, and then have the system calculate effective power reduction or issue effective alerts relative to these just as to grade changes.

Note that only power changes, not 'automatic' or penalty brake applications, are involved here; in the 501 accident the system would merely have substituted -- quite effectively, in my opinion -- for the view of small signs down a tunnel of headlit track by a distracted and perhaps inexperienced cohort of people in the cab.  A few more lines of code would produce an 'estimated reduce to' speed indication, with its own alarm/alert (to eliminate as much common-mode suppression of alerts, as with the alerter whistle in the Chase wreck, as possible).  None of this involving confusion with PTC network communications, or a need for any wayside communication other than sync to GPS satellites in the constellation.

It does seem to me that such a system would be compatible with properly-designed modular PTC (which, of course, we aren't getting) simply by including 'hooks' for the other functionalities in the bundled mandate, via glue logic or an intelligent control system as necessary.  But that again has nothing to do with the functionality of what Euclid is proposing.

[Overmod]

Euclid

I am only suggesting something better than committment to pure human memory, which is fine after it has had enough time to be developed after many, many trips.

Let me throw out a possibly ominous concern.

I have been concerned for decades with the potential pernicious effect on traditional route learning that would come with almost any safety system that automatically detects and compensates for running conditions.  This remains a concern if a moving-map alert system becomes relied on to 'know' where the train is, as opposed to the engineer-in-command knowing that directly by observation.  I worry that there are any number of potential failures in the map system -- railroaders familiar with systems like LEADER can comment on how likely these are, and the form they assume) -- that could potentially induce a green crew or even a fatigued experienced one to 'trust the map' rather than reinforce what they know.

One such system might involve a rational version of that idiotic 'call and point' practice, where upcoming wayside signal alerts are tracked and the crew must then demonstrate they have looked out and seen the appropriate sign or signal, perhaps by calling it by naming it or its mileage and then the response to be taken (e.g., "T30 in 2 miles; reducing now")  If this is presented correctly it might be a memory aid to learning the territory, not a crutch substituting for it.

[Euclid]

Overmod

I have been concerned for decades with the potential pernicious effect on traditional route learning that would come with almost any safety system that automatically detects and compensates for running conditions. This remains a concern if a moving-map alert system becomes relied on to 'know' where the train is, as opposed to the engineer-in-command knowing that directly by observation. I worry that there are any number of potential failures in the map system -- railroaders familiar with systems like LEADER can comment on how likely these are, and the form they assume) -- that could potentially induce a green crew or even a fatigued experienced one to 'trust the map' rather than reinforce what they know.

 

I understand your caution about the rolling diagram being relied on and thus replacing human memory.  Basically, the problem is always that people can become dependent on safety systems and consequently let their guard down.  The safety system causes them to reduce their own natural wariness.  So people using the table saw that stops when it hits a finger might rely on it and then cut their finger off when using an ordinary table saw.

But in the case of this rolling diagram/map, I would expect it to simply reinforce the natural commitment to human memory the way training wheels help to learn how to ride a bicycle.  When running a train down the railroad, the details of the road will end up in your human memory no matter whether they are being tossed at you from their actual presence on the right of way, or being tossed at you by the rolling map.  Either way, they will end up in the human memory.  But I agree with your point that failure of the map system must not happen, or there will be an induced danger from relying on it.

I have always been fascinated by the “point and call system of safety.”  I think it is actually brilliant, but it seems to carry a natural stigma related to the feeling of the practice looking stupid to outsiders.  But nevertheless, it is a mandated, mechanized process of paying attention to pertinent details regarding safe work.  I have found myself performing the practice just as a natural inclination during critical work.  I point at it and mentally frame it as an important point. When you think about it, simple things like “Measure twice and cut once,” is a form of “point and call.”

[petitnj]

You take the track diagram and put it in your lap. It shows you grade crossings and elevations and speed changes. Then you stare out the window the watch for the speed changes. And the other person in the cab reads trackside signs to you to be verified. 

 

[Anonymous]

Overmod

Volker, you haven't been listening. We had inductive train control worked out long before InduSi, and in fact mandated ATS, or something better 'to be developed' as a result of the mandate, for passenger trains shortly after the First World War.

Sorry if I don't know every American train safety system that was invented. I didn't try to show that we were earlier but that there were systems available earlier than the 15 years Euclid mentioned.

Overmod

But what Euclid is discussing is a system that requires no fixed inductors or magnets or balises or, in fact, any wayside markers or tercom optical tracking.

I did understand it. I just wanted to show that systems were available long ago.

Overmod

By the 1930s we had moved on to continuous code-based train control via multiple frequencies in the rails, which is inherently safer than any inductor-based intermittent system could possibly be -- this conclusion was reinforced quite recently with the multiple-unit collision about this time last year.

Which collision do yo think of? If it is Bad Aibling Germany in February 2016 it was caused by the negligence of the local traffic controler using his cell phone fore games and misjudging the train locations when he overrode the PZB90 system. Had he left alone the system nothing would have happened. The traffic controler was sentenced to 3 years and 6 month in prison.

None of these systems are 100% perfect. PZB90 was deemed safe enough up to 100 mph and was implemented system-wide. For higher speeds there is the LZB now being replaced by ETCS.
Regards, Volker

[ROBIN LUETHE]

Res a display showing tracks ahead and noting signs. Garmins on cars and just about every boating chart plotter does that.

[7j43k]

It's starting to feel strange that the crew in the cab has yet to be identified.  Brian Bostian was named two days after the crash.  It's been about a week.

Neither do I recall any mention of their physical condition.  Or what they have said in response to an interview.  If there was an interview.

 

 

Ed

[Euclid]

7j43k

It's starting to feel strange that the crew in the cab has yet to be identified.  Brian Bostian was named two days after the crash.  It's been about a week.

Neither do I recall any mention of their physical condition.  Or what they have said in response to an interview.  If there was an interview.

 

 

Ed

 

I feel the same way.  The one thing that could shed light on why the train was speeding is the explantion by the guy who was driving it.  We were told by the NTSB last tuesday, on the day after the wreck that they would interview the engineer in one or two days.  They also stressed that they wanted to have the interview while everything is fresh in the engineer's mind.

It has now been seven days and we would kind of like to know who was in the engine cab, what their titles and purposes were, and who was running the engine.  When we were told one or two days, I had a very strong hunch it was going to be at least a few weeks.  Frankly, I also have a strong hunch that they interviewed the people who were in the cab right after the accident happened.  It feels to me like the information surrounding these questions has been found to be too sensitive to release without a lot more preparation.    

[BaltACD]

7j43k

It's starting to feel strange that the crew in the cab has yet to be identified.  Brian Bostian was named two days after the crash.  It's been about a week.

Neither do I recall any mention of their physical condition.  Or what they have said in response to an interview.  If there was an interview. 

Ed

I have been thinking the same thing - we don't even know the extent of their injuries if any.

[Euclid]

Maybe we are hearing nothing further about the role of the engineer because NTSB is interested in helping Amtrak shape the view that it is not the engineer, but us the general public who are at fault for not funding PTC.

https://www.nytimes.com/2017/12/24/opinion/amtrak-derailment-seattle.html