Towards a Connecticut S-Bahn: the Waterbury Line

I’ve been neglecting blogging recently because I’ve been hugely focused on my ongoing senior paper writing process, which is a lot of fun and very rewarding but also very time-consuming. One of the joys of that process, though, is that it occasionally prompts thoughts about other planning issues on which my research touches. Such was the case with the news that Connecticut has managed to find some funds to invest in its ugly-stepchild Waterbury Branch, and that Metro-North is re-opening the temporary Devon Transfer station to allow track work on the main New Haven Line.

Some of my senior paper work focuses on the S-Bahn paradigm of regional rail services common in the German-speaking world and beyond. Like other systems that rely on mainline rail, S-Bahns in major urban areas combine lines on major trunk lines in urban cores to provide rapid-transit-like levels of frequency. One of the distinguishing marks of the S-Bahn paradigm, however, is its emphasis on precisely timed transfers at outlying stops across a wide region–up to and including a whole country–an approach known as the Integraler Takftfahrplan, or, roughly, “Integrated Pulse Schedule.” Scheduled thus, transit services can avoid running at expensive overwhelming frequency, and rely on precisely timed transfers to maximize rider utility.

Takt scheduling diagrams, from Maxwell 1999

One of the advantages of takt scheduling is that it can bring relatively frequent, useful transit to regions without a massive or dense population base. Indeed, some agencies have found the S-Bahn/takt system useful for serving polycentric, dispersed regions without one massive urban center. That got me thinking: where in the US, outside of core major urban areas, might such an approach be useful?

So, why not Connecticut? Connecticut has no dominant city; Stamford, Bridgeport, New Haven, and Hartford are all roughly in the same size class, and there’s a coherent second tier of smaller cities such as Danbury, Waterbury, New Britain, Norwich, and New London.

Urbanized areas on the New Haven Line, from RPA’s Getting Back on Track report.

The New Haven Line–arguably the country’s best commuter rail line, and one that I have argued before should be turned into a rapid transit line–ties together three of those top-tier cities, with a new service to Hartford starting in 2018 and connections to several of the second-tier cities. And yet, the state’s rail service is still predominantly conceptualized as “commuter” rail intended to shuttle passengers to office jobs in Fairfield County and New York City. Rather than providing everywhere-to-everywhere connections, the system all feeds toward the state’s southwestern extremity.

Also from RPA’s Getting Back on Track report.

Such is the case with the Waterbury Line. Currently, the branch operates only a skeleton schedule; completely unsignalized and nearly devoid of passing sidings, it can manage only one train every two hours in each direction. The diesel-powered trains offer transfers to other New Haven Line trains at Bridgeport, which eats up mainline capacity both because the short, slow Waterbury trains take up slots and because Bridgeport’s narrow, constrained station is a terrible place to turn trains. The new investment over the next few years will signalize the line and add a few sidings, bringing capacity to two trains per hour in each direction at peak.  That’s obviously a start, but what if we can make it better?

First, let’s think about where Waterbury Line passengers might actually be traveling. Here’s a look at the top 25 places where workers in the Naugatuck Valley are employed, courtesy of Census LEHD (the full spreadsheet is here for your perusal):

Place	Count	Share
Waterbury city, CT	17,215	12.7%
Milford city (balance), CT	9,421	7.0%
Shelton city, CT	7,752	5.7%
New Haven city, CT	7,093	5.2%
Stratford CDP, CT	6,497	4.8%
Bridgeport city, CT	6,343	4.7%
Naugatuck borough, CT	3,849	2.8%
New York city, NY	3,562	2.6%
Hartford city, CT	2,871	2.1%
Trumbull CDP, CT	2,569	1.9%
Stamford city, CT	2,494	1.8%
Derby city, CT	2,425	1.8%
Danbury city, CT	2,311	1.7%
Norwalk city, CT	2,226	1.6%
Orange CDP, CT	2,043	1.5%
West Haven city, CT	1,883	1.4%
North Haven CDP, CT	1,556	1.2%
Meriden city, CT	1,547	1.1%
Ansonia city, CT	1,369	1.0%
Bristol city, CT	1,140	0.8%
Torrington city, CT	1,094	0.8%
Westport CDP, CT	877	0.6%
New Britain city, CT	834	0.6%
Middletown city, CT	789	0.6%
Oakville CDP, CT	748	0.6%

Around 39,000, or 28.9% of the total, commute to Fairfield County, as the Waterbury Line is set up to serve. Another 2,500, or 1.8%, commute into Manhattan, and 700, or 0.5%, commute to Westchester. That’s a total of about 31.2%, as opposed to 65,255, or 48.3%, who stay within New Haven County–in the Valley itself, in the Shoreline towns, or in New Haven proper–and a further 1,600 who commute to Middlesex County and 1,163 New London County, further to the east. Which is to say: the Waterbury’s line’s emphasis on direct service to the southeast isn’t useless, but it’s not serving a majority of work-based trips particularly well. Can the area’s rail infrastructure help with that?

Perhaps the place to start is an emphasis on the power of connections. Offering a connection to mainline trains at Devon, rather than wasting crew and equipment time and mainline slots with a trip to Bridgeport, could free up the Waterbury Line to function more freely. Instead of rebuilding platforms all along the branch to fit level boarding for mainline rolling stock, the branch could use a dedicated fleet of European-style low-floor DMUs, making platform rebuilding much cheaper. (there is very little freight on the line, and a past study has found that electrification would only shave one minute off schedules because of the line’s extreme curviness) With trip times from Waterbury to Devon well under an hour, the line could run a train every half hour in each direction with four vehicles, plus one in reserve.

This is not, of course, a new idea. “Fixing” the branch by severing it at Devon is a common topic of discussion among railfans. A past study envisioned a reconfigured station at Devon looking something like this:

Alternatively, the “new” Devon could be moved a little east, and lose the “T” structure:

Of these two alternatives, I prefer the first–a two-track terminal offers more flexibility for frequent service, and the T-shaped platforms allow branch trains to operate different equipment than mainline trains. It wouldn’t be cheap–a substation located in the middle of the diamond would have to be relocated–but it also wouldn’t be as expensive as the study’s estimated $134 million price tag (the second alternative is projected at $73 million), mainly because there is absolutely no reason to build a parking garage under I-95. Instead of building a new pedestrian passage under or over the tracks, a future Devon station could retain the T format but rebuild the narrow Naugatuck Avenue bridge for use as an overpass.

With mainline trains now running every half-hour-albeit on a weird schedule with one 20-minute and one 40-minute gap–at off-peak times, the timing works out perfectly for branch trains to meet a mainline train at Devon every half hour throughout the day. The branch trains would, as in a takt system, be scheduled around their meeting time with a mainline train. With enough scheduling work, and as mainline frequencies increase as promised, the connection could become a three-way meet, with branch trains offering connections to mainline trains in both directions, thus increasing rider utility again.

One of the beauties of takt scheduling is that it can also offer connections to local transit. In this case, the trains meeting at Devon could also be met by a local bus feeding riders to the station from high-density  (by local standards) apartment developments near Walnut Beach in one direction, and from parts of Milford in the other:

Thus, a person arriving at Devon Station at the half-hourly takt mark would be able to choose to travel on transit in any one of four different directions aside from the one they came from.  Such a system requires hard scheduling work and good reliability of transit–but it is doable.

Turning the Waterbury Line entirely into a timed-transfer branch at Devon may or may not be the right concept. On the positive side, it would:

• increase branch and mainline reliability by ending mixing of branch and mainline trains
• allow timed transfers towards New Haven, where a not-inconsiderable number of Naugatuck Valley passengers are bound, without the extra travel to and from Bridgeport, as well as toward New York,
• allow operation of low-floor DMUs on the branch
• provide a rationale for a new connecting bus route

On the possible negative side, it would:

• eliminate one-seat rides for branch passengers to Bridgeport
• means that getting to an NYC-bound express train, which would stop at Bridgeport but presumably not at Devon, would require a second transfer
• struggle to attract walk-up traffic, since the station itself is isolated, bordered by the Housatonic river on one side and I-95 on the other

So there’s a need for further study, hopefully with more realistic cost estimates than including a massive, stupid garage when there’s a nice commuter lot available to build on at Stratford less than two miles away. But as a thought experiment, and a way to illustrate the feasibility and desirability of takt scheduling and the S-Bahn rail concept, I think it works nicely. Just don’t build the damn garage.

New Haven Line Penn Station Access, Faster and Cheaper

The topic of bringing trains from Metro-North’s into Penn Station on the West Side of Manhattan has been the subject of endless studies and public attention for the past 15 years or more.

Penn Station Access studies, 2002, including Hudson, Harlem, and New Haven Lines

Over the years, the expensive option of Hudson Line Access, which would involve an extensive rebuilding of the current, done-on-the-cheap Empire Connection, has been pushed off into the hard-to-envision future. Current plans revolve around the less capital-intensive option of bringing New Haven Line trains into Penn Station via the Hell Gate Bridge and the East River Tunnels. The idea garnered particular attention when it was included in the 2015-2019 MTA Capital Program,  then singled out–in distinct contrast to the rest of the MTA’s capital needs–by the Cuomo administration in the executive draft of the 2016-2016 state budget, funding $250 million of the projected $1 billion cost.

And therein lies the rub. There is, reasonably speaking, no real reason New Haven Line Penn Station Access (hereforth referred to as just PSA) should cost anything close to $1 billion. Though details are sketchy, the project as currently conceived appears to involve essentially the construction of four stations in the Bronx, a short extension of third rail in Queens to close a gap where Metro-North’s M8 EMUs can’t operate…and that’s it.

2014 proposed PSA alignment, with stations

Documentation in the initial 2015-2019 MTA Capital Program suggests that the budgeted cost for PSA was $743 million, still incomprehensibly high, but somehow also $250 million below the number included in publicity this year:

The Capital Program budgets $188 million for the four stations in the Bronx–close to in line with the $41.3 million construction cost for West Haven, the most recent New Haven Line infill station. But that’s only the second-largest section of expenditures. The program also forecasts, very confusingly, $264 million for “track and structures.” That’s confusing because the whole point of Penn Station Access is that literally no track work is required, as Amtrak trains demonstrate every day. Alon has made the case for grade-separating Shell Interlocking, where the Hell Gate Line splits off from the Metro-North tracks to Grand Central, and that should definitely be done, but there’s no indication that that’s where the $264 million is going here. Perhaps some of it is going to the planned reconstruction of Herald interlocking in Sunnyside Yard, but that’s far more necessary for East Side Access than PSA. Perhaps some of ESA’s spiraling costs are being shifted onto PSA?

The other potential scenario is that Amtrak is demanding MTA restore some additional tracks onto the Hell Gate Line. The line has a four-track right-of-way that currently carries only two passenger tracks, with stretches of a non-electrified third track for (very limited) freight service. Amtrak hasn’t exactly been an easy partner with regard to East Side Access, so there’s no reason to assume they’d make the MTA’s life easy when it comes to PSA either. In any case, unless massive levels of service are planned for PSA, there’s no reason to add more tracks to the Hell Gate Line–the existing two tracks are plenty to handle Amtrak traffic plus a few additional Metro-North trains. But the point is the public doesn’t know where this significant expenditure is going. Maybe it’s actually being spent well. Maybe there are real needs I and other transit bloggers am not aware of. Or maybe not. In the meantime, it certainly looks bad.

Speaking of service: one of the other incomprehensible things about PSA has been the vocal insistence from MTA and the Cuomo administration that service cannot begin until some many LIRR trains are diverted to Grand Central by the opening of East Side Access. Presumably, this is their way of heading off conflicts with Long Island legislators who have previously gone to war to preserve parochial geographic privileges within the limited platform slots available at Penn Station, but it’s not, well, strictly necessary.

It has become very common and fashionable for transit advocates and bloggers to call for commuter trains to run through Penn Station rather than terminating there as a solution to the station’s growing capacity problems. With the very limited exception of the joint MNR/NJT Train to the Game Service, this has not yet happened, nor do the operating agencies show any apparent interest in making it happen, aside from vague references to through-running cooperation in dense documents.

Gratuitous YouTube break, demonstrating that New Jersey Transit trains can, in fact, run through to the New Haven Line

There are genuine technical reasons that through-running is hard. While NJT’s dual-mode and electric locomotives can operate throughout the corridor, the New Haven Line’s M8 EMUs cannot operate on the 12 kV/25 Hz electrification system installed on the Northeast Corridor between Gate interlocking (on the Queens side of the Hell Gate bridge) and Washington, DC.  There are a lot more of the EMUs, and they’re much preferable to loco-hauled trains, since they accelerate faster.

That being said, the gap between the end of 12.5 kV/60 Hz electrification at GATE and the beginning of LIRR’s 750 V DC, third rail electrification–which M8s can operate over–at Harold Interlocking is less than two miles. The third rail then extends through Penn Station to the west portal of the Hudson River tunnels. From there, it’s less than a five-mile gap of NEC-style electrification to Kearny Interlocking. There, NJT’s Morris & Essex Lines split off. Since 1984, they’ve been electrified at 25 kV/60Hz–a system under which the M8s can also run.

In other words, a perfect through-running partner for PSA service already exists on the Jersey side of the river–a line on which both NJT and Metro-North equipment can operate freely. The only technical barrier is the very manageable gaps in third-rail coverage.

Gaps in M8-friendly electrification highlighted in red.

From some Google Maps scouting, it appears that a total of about 16 track-miles of new third rail would be required, give or take some since I don’t know exactly where various electrification standards begin and end. Estimates as to the cost of new third rail vary, but $3 million per track-mile seems reasonable, perhaps even conservative. At $3 million per mile and 16 track-miles, you’d end up with a cost of right around $50 million for the needed third-rail extensions–very, very reasonable for the capacity improvement it represents.

So for just $50 million, we can run any New Haven Line train we want through to Gladstone, Dover, or Montclair State University. There are additional costs, of course. While all three M&E Lines terminal stations (in electrified territory) have high-level platforms, relatively few of the other stops do, and M8s have no traps for low-level platforms. I count a total of 58 platforms that would need to be high-leveled on all three branches. At a cost of $5 million per platform–again, conservative–that’s a further investment of $290 million. Most likely, you could knock off $90 million of that by not bothering with the ten stations of the very rural Gladstone Branch, and you could establish skeleton express service to Newark Broad Street, Summit and Dover on the Morristown Line and Bay Street and Montclair State on the Montclair-Boonton Line without any modifications at all. And, of course, existing NJT equipment can handle any and all platforms.

So where does that leave us? Costs for a barebones proof-of-concept run-through system could look something like this:

• $50 million for closing gaps in electrification
• $200 million for all four Bronx stations, politically the most important part of the project
• $200 million for high-level platforms on the Morristown and Montclair-Boonton Lines
• Presumably up to $100 million in various signal, yard modification, and other miscellaneous costs

For those counting at home, that’s about $550 million. For that money, you’d get:

• direct access from the Eastern Bronx to the West Side of Manhattan and job markets in New Jersey, including Newark
• a one-seat ride from eastern Westchester and Connecticut to Newark, and vice versa
• more efficient use of existing train slots at Penn Station–“free” capacity improvement that doesn’t detract from any other line’s service
• proof that running through Penn Station is both technically and politically feasible.

This vision of PSA and through-running at Penn Station might not be the highest priority we can dream about, but it is likely the most easily achievable. Given ESA’s ever-accumulating delays, PSA might not happen until 2025 if it has to wait for the other project. What I’m offering here may be barebones, but it offers the opportunity to make an innovative, somewhat important project happen far faster than otherwise planned.

Of course, this is the US, and more specifically the Tri-State region, so the real barriers aren’t technical but political and bureaucratic. With Albany and Trenton both mired in scandal, and a New York gubernatorial administration that for some reason seems determined to sandbag PSA, this kind of a scheme is unlikely to come to pass. Getting the various railroads involved here to work with each other is notoriously difficult, and given that Amtrak owns much of the infrastructure involved, heads would probably need to be knocked at the federal level (paging Senator Schumer…) The attitude from government so far has largely been to out-spend fundamental organizational problems (something that can be send of many, many aspects of transit in the NYC area), but let’s try for something better. In an era of fiscal constraint, low-investment, high-impact sure sounds nice, doesn’t it?