Reading, PA Trip Report

I went on a signal trip to the Reading, PA area this past weekend where, for those of your who don't know, NS has a bit of a hub going on where their Northeast network was built on the skeleton of the former Reading Railroad.  In 2018 NS replaced most of the remaining Reading / Conrail signaling still present on the Reading and Harrisburg Lines.  However a late Conrail project to replace vintage Reading Railroad signaling in the late 90's has created a pocket of surviving Conrail signaling that looks to be safe from replacement for at least the time being.

The bulk of this "historic" signaling is located on the old Reading Belt Line between CP-BELT and CP-TULP.  In fact CP-BELT features Conrail painted NS style traffic light signals at one end that may indicate the project straddled the NS takeover. 

The island of Conrail signaling includes a single automatic at MP 6 with back to back distant masts on each track.  These replaced a pair of Reading brackets sometime after 1997.

CP-TULP features a paid of equilateral turnouts forming a "Conrail Crossover" that I believe dates from the Reading era as they exist due to a single track bridge over Tulp creek.

In other news the downtown triangle consisting of CP-OLEY, CP-CENTER and CP-WALNUT has been re-signaled to varying degrees.  CP-WALNUT has the same interlocking hardware with new signals, CP-OLEY has been completely re-signaled and CP-CENTER has been reconfigured with the western turnout in the "Conrail Crossover" being moved further west to eliminate a two track grade crossing.  This had the side effect of replacing a surviving Conrail dwarf stack with a new Darth mast.  I suspect the interlocking logic and the Reading era relay hut were also replaced, but I wasn't able to check on that one detail.

I also want to take a minute and plug the Saucony Creek Franklin Station Brewpub, which had taken up residence in the old Reading Franklin St Station.  I don't think they were the ones who restored the building, but they did lease it after plans to turn it into a transportation center fell through.  Great food and drink with a lot of history.

Pittsburgh Line Cab Signal Overlaps and Human Factors

A recent derailment between three NS freight trains near Labrobe, PA sent me down the rabbit hole of human factors and accident prevention once again.  The derailment was the result of a restricted speed rear end collision that caused cars to derail into the path of a third opposing freight train.  Straight away I would like to point out that this sort of accident is not PTC preventable and exactly the reason  why high crash standards are important even in the face of technical safety systems.

Occurring just a few months after NS changed the Pittsburgh Line from Rule 261 wayside signaling with Cab Signals to Cab Signals without fixed automatic wayside signals I immediately suspected that the loss of visual cues from wayside signals prompted a crew to misjudge the point where Restricted speed would take effect as under the traditional wayside system an obstruction can be located one foot beyond the signal displaying Restricted Proceed.

 In theory a cab signal should change from Approach to Restricting as one crosses the insulated rail joint into the block where the cab signal code is not reaching the following locomotive.  It is standard practice for a cab signal cut to be placed some distance in advance of an absolute signal, dropping the cab to restricting in time for a train Approaching at Approach to know that an absolute stop is necessary.  I have observed SEPTA placing cab signal cuts in advance of automatic signals on the Reading viaduct, but there the signals are spaced a half mile apart so and SEPTA had plans to convert the Regional Rail Division to transit style operation anyway.  However, in doing some research I learned that Conrail had implemented this same system on its entire Pittsburgh Line with cab signal cuts between every signaling location such that a train on Approach would see its cabs drop to Restricting a mile before encountering the Stop and Proceed signal.  This feature has since been confirmed from a head end video.

New Signal Indication Point (SIP) at MP 125.3

Old unlabeled Cab Signal change point at MP 125.3

This explains why NS was creating new "Signal Indication Points" every mile along the Pittsburgh Line as part of its re-signaling project.  It wasn't increasing capacity by increasing the number of blocks, just replicating the system Conrail had already installed.  I have long wondered how Conrail would prevent rear-end accidents on its Rule 562 territory (cab signals without fixed wayside signals) as there were no mentioned in the rulebook of crews traveling on Approach being provided with a stop target (as is the case on the Union Pacific Hiawatha Sub).  It turns out that they were likely using signal overlaps the whole time.  If this practice originated with the PRR's Conemaugh Line experiment remains to be seen.

Of course when it comes to human factor's nothing is what it seems.  In a world where one's cabs stop to Restricting halfway through a block one knows to be clear until the next wayside signal, crews might not take that Restricting cab signal so seriously until passing a wayside Stop and Proceed.  Over decades of service, these habits can become ingrained, especially if rules testing fails to cover the corner case.  A feature designed to provide crews with advance warning of a track obstruction may have instead created a false sense of security and a hidden reliance on wayside signals.

Additionally, although this episode both revealed and in part confirmed the use of cab signal overlaps on former Conrail territory, it now raises further questions about how the new NS SIP system will function compared to the Conrail system of waysides and cuts.  Will the "odd" SIPs still only act as Approach -> Restricting cut points or will they be sully featured automatic signal locations that help to reduce train separation?  I guess I'll have to put my feelers out ;-)

CBTC is a Scam and the MBTA Backs Me Up

Communications Based Train Control promises higher capacity at lower costs thanks to the magic of WIRELESS TECHNOLOGY!  However almost every real world situation that makes use of it seems to wind up costing a ridiculous amount, having major reliability problems or both.  It turns out that with signaling there is no free lunch, but when faced with overcrowded subway trains, planners can't help but get seduced by those lovely braking curved.  I mean fixed block, that's so 1890, surely we can do better!

In 2016 the MBTA was conducting a capacity study for the Red Line, which currently has a somewhat anemic throughput of 13tph in the peak period.  With new rolling stock on order, the (T) would literally have more cars than it could run.  The current signaling system made use of fix block audio frequency cab signals installed in the 1980's.  This is similar to many other transit systems such as the CTA, WMATA and BART.  Of course CTBC is the to go technology for capacity expansion and the study quickly confirmed this.  Oh wait, it didn't.

• A detailed analysis assuming a moving
block CBTC system on the Red Line was completed.

• Analysis found that a CBTC system would produce
an improvement of just one train per hour beyond
the improvement from the new cars and minor
system changes.

• Major Red Line capacity improvements can be
achieved without implementing very costly CBTC. 

• Long dwell times in the downtown area and close
spacing of stations limit CBTC as much as they
limit fixed block systems.

That's right, just like the costly NYC Subway L Train CBTC system only increased capacity by 2 trains per hour, applying CTBC to the Red Line would only improve capacity by 1 train per hour over a fixed block alternative.  Past a certain threshold capacity is limited by dwell time and the efficiency of terminal interlockings. The study also found...

• The shorter the block length, the closer the
system is to the ideal CBTC (moving block)
braking distance

• MBTA block lengths in the central subway already
average less than 500 feet (6 car trains are 416
feet long)

It's nice to see that for once a transit agency actually ran the numbers and decided that CBTC just wasn't worth it.  It turned out the best way to increase the capacity was simply to allow the new rolling stock to use updated braking curves that will result in later braking and more aggressive cab signal speed stepdowns.  Also the 1980's audio frequency cab signal system will have its components replaced with digital versions that have faster reaction time and thus allowances for less conservative block progression. 

The Alstom GM4000 - It's a Fake!

Excuse the click-bait-y title, but I just discovered something the other day that I wanted to bring to everybody's attention.  In North America there are effectivly two families of electric point machines,  The Union Switch and Signal M3 family and the General Railway Signal Model 5 Family.  You can order these in a variety of variants, single control, dual control, low profile, high profile, etc.  These two families have been around for so long that in the case of the M3 a variety of third parties are offering legit off-brand examples at a lower price point and without the US&S trademark.   Back in the 60's, GRS decided to get ahead of the copycats by designing a brand new point machine with a new set of associated patents and trade secrets.  Known as the Model 55, this cheap looking rectangle was relegated to North American transit systems, although it did kind of become the closest thing to a standard point machine that the UK has (so like I said, transit systems).

GRS Model 55 on the TTC's Scarborough Rapid Transit.

Needless to say the Model 55 was never very popular and even some of its enthusiastic early adopters, namely the Washington Area Metro and South Jersey's Port Authority Rapid Transit, eventually ultimately their Model 55's for US&S M3's.  I don't know why this was the case, but I suspect a lack of support was part of the reason as GRS had a new European parent in the form of Alstom and daddy Alstom was intent on making its own new M3 "killer" along with the associated patents.

Low profile US&S M3 on the NYCTA (7) Line at Queensboro Plaza

Backtracking a few decades, to take on the GRS Model 4 and GRS Model 55's electric transit switch machines, US&S had developed a low profile version of is venerable M3 because third rail shoes are a thing on transit. Basically it looks just like a regular M3 only flattened down by an inch or two.  So when the following appeared at Amtrak's rebuilt CORK interlocking in 2005, I thought they had opted for some sort of off brand low profile M3.

However something always bugged me about that shape and style and my instinct was indeed correct as that is not a low profile M3, but an Alstom GM4000A switch machine.  The spiritual successor to the GRS Model 55 and, apparently to some comments I got, just as failure prone.  Nevertheless a bunch of railroads bought GM4000s about 15 years ago and I happened to run into some of the survivors for the first time last December at Conrail's CP-JOHN in Morrisville, PA.

Outwardly presenting as a US&S low profile M3, the GM4000A is given away by having just two segments on top instead of three,  However if you check out the manual for the original model GM4000, it does indeed have the three segments.

I'm not here to speculate about why this inferior modern replacement of tried and true 1930's technology is an inferior modern replacement of tried and true 1930's technology.  That's a question for a C&S purchasing officer.  What I want people to take away is the knowledge that GM4000 family point machines are a thing, how at first glance they may resemble low profile M3's and that they might soon vanish from the scene so go take some photos while you can.  In this day and age every bit of diversity is a plus. 

PRR Main Line Signal Documentation Summary

On Nov 25th I am going to be presenting my work to document the PRR Main Line signaling between CP-HARRIS and CP-PENN at my local NRHS chapter meeting.  While the project unofficially started in 2009 and gained urgency in 2012 and beyond, the bulk of the photos were taken starting in 2014.  While I was not able to document everything I wanted to or even everything I would have been able to document assuming perfect planning I am happy with my work and I figured I should present a summary of the results here.

I will be using terms like Fully Document, Partially Document and Survey.  These mean got photos of all components of a signal location, only some of the components and then only photos from a passing train.



The Pittsburgh Line has a total of 38 Conrail era interlockings of which I was able to fully document 22, partly document 6 and survey 10.  The biggest misses were CP-TUNNEL, CP-MG and CP-SO.  CP-MARY, CP-AO, CP-JW and CP-W had all already been re-signaled to varying degrees and were simply not a priority.

The Pittsburgh Line has 82 main line signal locations of which I was able to fully document 38 with the remainder being surveyed, although when only counting position lit automatic signals I got 33 of 59.  I regrettably missed two signals on the track 3 turkey path between CP-C and CP-CONPIT entirely.  There are also 2 or 3 yard signal locations (2 at Altoona, one previously at Harrisburg) that were also missed.

  Total number of photos taken was around 14,000, including photos of trains and other Main Line infrastructure.  Remember this is just between Harrisburg and Pittsburgh and doesn't count work I have done on the Amtrak portion of the line.

The photos were compiled over 25 trips, 4 of which were surveys on Amtrak's Pennsylvanian, 2 of which I would consider failures.  There was an additional trip on a fall foliage special that allowed me to cover the Rockville wye.  The other 20 trips were of the road trip variety, two doing double duty for Buffalo Line photos.  Six were trips to Pittsburgh.  Four were trips to Clarion, PA.  Three were dedicated Altoona trips and the other seven day trips along the Middle Division, often involving the always wonderful HARRIS tower.  Arguably my first trip on this series was the one I took to ALTO tower in 2004 so that would make this a 15 year effort.

All these numbers are probably off by a few, but they are the best I can figure without a full blown audit.  I plan to do a live stream of my presentation sometime on or about Nov 22rd to practice and content and see how it goes for time (Draft presentation stream can be found here).