Amtrak signaling on the NEC

Jersey_Mike wrote:

So the bottom line is even in a cab signal with speed control territory when for whatever reason one operates with the safety system cut out, accidents can happen, which of course should come as no surprise to anyone I should think.

That's not the bottom line, its that accidents can happen even with multi-billion dollar safety systems present. After a certain point, additional safety systems are simply not worth the investment or present major operational difficulties (case in point SEPT'A's new Subway Surface CBTC).

We have not established yet that an accident has happened or was about to happen when a protection system was actually operational.Of course it is a tautology to say that accidents can happen where such is just installed, and no one can argue with that. I would be interested in learning about situations where an accident happened while the system was operating as designed and no one was working around it by switching parts of it off. Just for my education mind you. Not challenging anyone on anything, since I am probably one of the least informed on these matters on this board.

This is similar to a post I made in the Cali forum about the Metrolink crash

While im not intimately familar with the cab signal system that Amtrak uses in their Acela Express trains or what they use in their AEM7s/HHP8s or former Metroliner cab cars, I am familar with the system NJT uses, and how it works, and considering that NJT trains run on the NEC, this would be the "worse case scenario" - because unless all trains have a PTS system installed, its not that much safer.

Anyway, its a Cab Signal/Automatic Train Control/Automatic Train Stop system. Engineer must 1) physically acknowledge a cab signal change (ATS) and 2) physically put brakes on if traveling over the speed the cab signal system is now calling for/allowing (ATC). Now, you just can't put on a tiny little bit of brake, you have to achieve "Suppression" which is usually anywhere from a 17 to 25psi reduction in brake pipe (running at 110psi). The engineer can only release the brakes when under or at the speed the cab signals are allowing. If he does it before he gets to that speed (say 35mph whlie the cabs are allowing 30mph on an Appraoch in the cabs) it will immediately go into a penalty brake application. it will also go info a penalty if the enigneer doesn't get enough brake on in time (usually 6 to 10 seconds)

Also, in the case of what happened in Cali, if the engineer were to leave the station, and go full throttle, and not pay attnetion to his cab signals, as soon as the speed became higher than what the cab signals would allow, the cab signals would sound a warning and he has a few seconds to acknowledge and get 1) either his speed back under the speed allowed, or 2) apply the brakes enough to get Suppression.

So lets use Metropark going east on track 1 for example. You cannot ordinarly see Islen's home signal from the station. But your cab signals are at a restricting. That would allow the engineer to go a max of 20mph. If he full throttled out of hte station, as soon as he hit about 21 or 22mph, the cab signals would go off and he would have to slow down or put brake on (like i mentioned above) if he didn't, the train would go into a penalty brake application after about 6 seconds. So wiht a fully functioning cab signal system on the NEC, you could not just pull out of a station and blow past a stop siganl at 42mph, like the Metrolink crash (if the station is close to a home signal displaying a stop signal.)

However, with ATC/ATS you CAN still go past a stop signal, and it happens. The cab signals generally drop to a restricting (from Appraoch) about half way inbetween the Approach wayside signal and the Home signal (absolute signal). Not always and not in all locations, but MOST locations. So the engineer has to bring the train down to 20mph (or 15mph wiht some systems) - He can still go past that stop signal at up to 20mph (with a Restricting on the cab signal), and there is nothing to stop him when he passes that signal. (that would be PTS)

However, a PTS system woudln't necessarily stop the train in time. If he was going 20mph past a signal, and the PTS came on and stopped the train as soon as it went past the signal, that doesn't guarantee that the train would stop short of fouling a switch and could be struck by a train on the track that the train is now fouling.

So in short, what happened in California couldn't happen on the NEC with a train with a functioning cab signal system (they are very reliable). However, thats not to say that an accident where a train is fouling a switch because it ran a stop signal and got stuck either head on or nearly head on.

I think it's fair to conclude that for the metrolink accident to occur on the nec the engineer would have to have a pratically malicious intent, otherwise the cab signal system would go into penalty.

Well with reports that the engineer didn't make any attempt to slow down, even right before the collision, brings up to issue of some sort of medical problem. The typical dead man alerter as I have seen in the Metroliner cab cars goes off after something like 45 seconds of inactivity. That's more than enough time for a train to slide past a stop signal in the manner that happened here. The event recorder will hopefully show if the engineer was doing anything on approach to the Stop signal, but even a simple CSS system would have stopped the train had the engineer failed to acknowledge the code drop in advance of the stop.

A minor correction here for Jersey Mike. Positive Train Control would actually cost about $4.5 billion to implement on the Class I network (my estimate in a 2004 report to Congress). HOWEVER, the railroad industry is currently spending about $500 million annually on capital replacement of existing signal systems (which cover about 65,000 route miles). What they're replacing is old CTC or ABS, and what they're replacing it with is new CTC, unchanged in functionality. BNSF is actually removing the obsolete Santa Fe ATS system (which does not, in any case, comply with current FRA regs).

At the current rate, over the next 20 years or so the Class I railroads will replace all existing signal systems, at a total cost of about $9 billion. Now, some of this work would have to be done anyway, even if PTC were installed. You would still need signals at interlocks and at the ends of sidings, and of course you'd need field relay logic to control them and associated turnouts. But, bottom line, replacing CTC with PTC would actually SAVE the railroads about a billion and a half dollars over simply replacing the CTC in kind.

Now, PTC will essentially prevent "human factors" accidents because it will stop trains BEFORE they reach the end of their authorities, rather than letting a train overrun a red signal and then applying the brakes. PTC is indeed a proven system; I saw it demonstrated on Minnesota's Iron Range about 20 years ago (and that was with a 4800 baud digital radio link and the equivalent of an 8086 processor in the on-board computer. We can do a lot better than that today).

FRA has a "microprocessor-based train control" rule which lays out how a PTC system should work, and the performance it must offer in order to be in compliance. There are no longer any obstancles, regulatory or technical, to PTC implementation.

Oh, and its cost is certainly not justified by accident avoidance. Rather, better information on train location should lead to an increase in system average velocity and a corresponding improvement in car utilization. I calculated the benefits in the 2004 report, and they exceed $2 billion per year for the industry.

So what are the railroads waiting for?

Now, PTC will essentially prevent "human factors" accidents because it will stop trains BEFORE they reach the end of their authorities, rather than letting a train overrun a red signal and then applying the brakes. PTC is indeed a proven system; I saw it demonstrated on Minnesota's Iron Range about 20 years ago (and that was with a 4800 baud digital radio link and the equivalent of an 8086 processor in the on-board computer. We can do a lot better than that today).

CBTC Systems are inherently unreliable, even the kinds that employ large amounts of trackside beacons. I point to CTBC problems on the L in NYC and the SEPTA Surface Subway with the latter having doubled travel times from 20 to 40 minutes. The GPS PTC systems may be good for low density, currently un-signaled lines, but they are not a replacement for track circuit based signaling on main lines and might even be dangerous given the current ways freight trains are operated. Freight trains do not have a uniform braking curve and and improper brake applications can cause derailments. In the absence of a uniform braking curve freight trains have to be run under the worst case scenario. This is why freight trains have to run with extra speed restrictions on Amtrak and Metro North. This sort of stuff might fly in Europe where the railroads are basically a rapid transit network and the freight service is rather pitiful, but slowing trains down to 15mph for an Approach signal (a la Metro North) is just not going to cut it in North America.

The track circuit is the best form of rail safety system because it directly measures the integrity of the track. Trying to replace it with some sort of GPS based system that infers the condition of the track is a recipe for disaster. You are also assuming the constant availability of GPS and other satellite based systems and ignoring of signal interference issues with tunnels, canyons, power lines or GPS Spoofing. Pulse code CSS is in use by all 4 major Class 1's (and I think the FEC). It provides 95% of the safety of the more "advanced" systems and if you want absolute stop protection throw in some ACSES style transponders for that (or "upsidedown canoe" ATS inductors).

What you are suggesting is that every train broadcast its position 25k miles up to a geo sync satellite and then 25k back down to some central control facility where a vital system keeps track of everything and then sends back instructions (25k up and down) to the trains. Let's see here, we have a disruptable communications link, a central office that that is also vulnerable to outage and a round trip time of about 5 seconds for the messages. If you aren't using track circuits who knows what could be out on the tracks and if you are you need a parallel data link for that information. The alternative is pulse code CSS that's been in operation since the 20's and can work with a solar panel, a battery, three relays and a pocket sized solid state pulse code generator.

The consultants love to hock complex systems because there's more profit in it and when the system doesn't work properly they get even more fees. Yet time and time again everywhere people have tried to be smart and high tech its been a disaster. Save the SCADA networks for IT crowd and keep the field hardware simple and reliable. That means track circuits and pulse code/audio frequency cab signals with full wayside signals and absolute and distant signals for CSS failure fallback. Oh wait, that's exactly what Amtrak has and continues to implement on its high speed lines.

Personally I would require some form of high visibility position light at all interlockings not in CSS territory.

I'm not a rail person, but I am an engineer of a different sort.

So let me see if I understand what I read correctly: The signaling information on the NEC is transmitted by some sort of pulse code conducted via the rails?

I get the benefit of cab signaling in that it is possible for the state of a block to change after a train has entered it.

pennstation wrote:I'm not a rail person, but I am an engineer of a different sort.

So let me see if I understand what I read correctly: The signaling information on the NEC is transmitted by some sort of pulse code conducted via the rails?

I get the benefit of cab signaling in that it is possible for the state of a block to change after a train has entered it.

The benefit is not if the signal changes while the train is in it. The purpose is to stop the engineer from "speeding" or passing a red signal.

One benefit *is* that if for any reason the track circuit is broken or shorted, for example by another train entering the block, the display in the cab will drop, whereas with only waysides you wouldn't know.

As for the case which I think you're referring to, if the signal becomes more favorable in the cab, say from Approach to Clear, you get the benefit of not having to wait until the next wayside signal to speed up. Here's the rule:

553. Cab Signal Changes Between Fixed Signals
If the cab signal changes between fixed signals, the cab signal will govern, subject to the following
restrictions:

1. Cab Signal Changes to Restricting
When the cab signal aspect changes to Restricting between fixed signals, the Engineer must take action
at once to reduce to Restricted Speed.

2. Interlocking Signal Requires Medium or Limited Speed, Cab Signal Changes to More Favorable Aspect
If an interlocking signal requires Medium or Limited Speed and the cab signal changes to a more favorable aspect, the speed must not be increased until the train has run its length.

3. Cab Signal Changes from Restricting to More Favorable
If the cab signal aspect changes from Restricting to a more favorable aspect, the speed must not be increased until the train has run its length or 500 feet, whichever distance is greater.

4. Cab Signal Changes from Clear to Approach Medium
If the cab signal changes from Clear to Approach Medium between fixed signals, trains must immediately begin reduction to Limited Speed, and must further reduce to Medium Speed, unless the next signal is seen to display a more favorable aspect.

EXCEPTION: If the cab signal does not conform to the fixed signal at the entrance to the block, and the fixed signal is more restrictive than the cab signal, the fixed signal will govern movement through the entire block.

One benefit *is* that if for any reason the track circuit is broken or shorted, for example by another train entering the block, the display in the cab will drop, whereas with only waysides you wouldn't know.

When there was that boxcar runaway at Canton Jct, the MBTA train's cabs dropped when the boxcar entered the block toward it allowing the train to stop. This is why you'll always need track circuits, which also detect flooding and many types of adjacent track derailments.

hi55us wrote:

pennstation wrote: I get the benefit of cab signaling in that it is possible for the state of a block to change after a train has entered it.

The benefit is not if the signal changes while the train is in it. The purpose is to stop the engineer from "speeding" or passing a red signal.

Just to be pedantic, as was pointed out to me by an NEC Amtrak Engineer on the railroad.americas netnews group, the cab signal does not do any enforcement, it just displays the current aspect in the cab. It is the speed control system that enforces the speed. The old PRR speed control system will not prevent a train from passing a red signal at restricting speed, i.e. 20mph or lower. For that you need the ACSES overlay (if on Amtrak or the ASES overlay if on NJT) which provides additional information to the train about exact location of signals using track mounted balise/transponders, and enforces the stop at red signals.

And yes, one additional benefit is that if signal changes while the train is within block the train gets the information immediately and can/must take appropriate action to comply specially if the change is to a more restrictive state. Moreover if it is to a more permissive state, then the train can comply after a short delay which is specified in the rules, thus improving efficiency of operation. Again all this is enforced by the speed control system based on the aspect displayed by the cab signal.

As pointed out by Jersey_Mike, in a system that has good train integrity detection system built in as in CSS or even axle-counter based systems as allowed in ETCS, even unplanned intrusions can be detected by the system and acted upon. In addition the CSS based systems also are able to detect rail breaks etc. and have the right thing happen, and hence in that sense they are superior to axle-counter based systems, and of course both of those are better than pure communication based systems. OTOH, a communication based overlay on CSS like ACSES II can give some additional efficiencies, but of course is more expensive.

The ETCS axle counting malarkey would never work in North American due to the prevalence of on-line sidings and other non-interlocked entrances to main lines. European resistance to track circuits is absolutely inexplicable. They spend 100 years clinging to manual block stations with hand worked semaphores and telegraph block instruments and when it finally comes time to leave the 19th century they opt for a track occupancy device that omits the most important innovation in railroad safety, the track circuit. It's like trying to build a car that completely replaces seatbelts with airbags (and BTW yes that was tried, it didn't work out very well).

Anyway, let's hope all this reactionary PTC legislation dies down and Metrolink and Caltrain just install basic CSS. Look for Caltrain to get it first tho, they run an isolated network that already uses a unique (for the West Coast) speed signaling system.

As a side note, NJT does NOT have the ASES in service - the system that is in service is the "old fashioned" PRR system, but with additional Cab Speed 60 and CS80 aspects.

You can still go by a stop signal at 20mph with a Resticting in the cabs - there is NO stop signal enforcement.

Again, basic definitions are still in effect here

Cab signals - show the real time status of the railroad inbetween wayside signals by current sent through the rail and is displayed in the operating cab by some kind of device which shows the status by various means (could be small position light signals, could be color lights, it could be green bars, etc)

Automatic Train STOP - ATS - requires the engineer to acknowledge a change in the cab signals by pushing a button and doing it within a certain amount of time - if not, the system will put train in a penalty brake application and the brakes CANNOT be recovered until the train is stopped.

Automatic Train CONTROL - ATC - requires the engineer to take action by putting on a pre-determined amount of brake (called suppression) and bringing the train to or below the speed shown (allowed) by the cab signal system.

They are three SEPERATE systems - however, they can and do work together - think of it as "a la cart" - you can pick and choose what you want - Cab signals with NO ATC or ATS, Cabs with ATC only or Cabs with ATC and ATS, whatever. The NEC has CSS with ATC and ATS.

Im not to familar with Amtraks system on their Acela and locos on the NEC (ACSES), as I don't work with it. However, the system is only as "safe" as its weakest link, which would be the non ACSES systems of other trains that operate on the NEC - NJT, Septa, Conrail, NS, CSX, whoever, whereever. BTW, im NOT saying that the other systems are unsafe, but do NOT have a PTS/stop signal enforcement system working or inservice.

And considering that Amtrak is a relatively small percentage of the number of total trains run on the NEC, the "standard" system is much more prevalent.

Jtgshu wrote: Im not to familar with Amtraks system on their Acela and locos on the NEC (ACSES), as I don't work with it. However, the system is only as "safe" as its weakest link, which would be the non ACSES systems of other trains that operate on the NEC - NJT, Septa, Conrail, NS, CSX, whoever, whereever. BTW, im NOT saying that the other systems are unsafe, but do NOT have a PTS/stop signal enforcement system working or inservice.

And considering that Amtrak is a relatively small percentage of the number of total trains run on the NEC, the "standard" system is much more prevalent.

I believe that in the segments where 150mph is allowed under ACSES, anything running on that piece of track is required to have ACSES. This would imply that the MBTA trains on the Providence line are equipped with ACSES, as are of course all Amtrak trains.

OTOH, where ACSES is used in 135mph segments south/west of New York (County to Ham, and possibly Ragan to Prince) other trains are not required to have ACSES at present, and hence NJT I don;t believe has ACSES on its trains. I am given to understand that ASES and ACSES are mutually compatible, so as NJT trains get ASES installed they will be able to work under ACSES. Of course I am sure someone will correct me if I am wrong in any of my understanding, and that is always welcome.

On the matter of European, Jersey_Mike raises an interesting issue. Interestingly the TVM 300 and TVM 430 systems used by TGVs as well as the ETCS system installed on LGV Est I believe are both based on coded track circuits. ETCS uses track mounted balises/transponders (like ACSES or ASES), whereas TVM uses more information through the coded track circuit to achieve the same goals. For example I believe under TVM length of block information is transmitted via coded track circuit which is used by the train together with its own braking curve information and signal aspect information to decide how much brake to apply and where to stop. My impression is that it is possibly the Germans that are more allergic to track circuits, and at the end of the day it is probably the same motivation that motivates the pure CBTC based PTC crowd, which is cost.

My impression is that it is possibly the Germans that are more allergic to track circuits, and at the end of the day it is probably the same motivation that motivates the pure CBTC based PTC crowd, which is cost.

Once your deal with the debugging and all of places where the signal drops out you end up with as many (or more) wireless transponders as you would have had track circuit boundaries.