Regional Rail Proof of Concept
This Proof of Concept paper is the product of a collaborative effort by TransitMatters members led by Alon Levy and Ethan Finlan. We wish to acknowledge the contributions of the following TransitMatters members (in no particular order):
Josh Fairchild, Ted Pyne, Jim Aloisi, Tim Lawrence, Ari Ofsevit, Peter Brassard, Charles Li, Cindy Heredia, Zak Koehler
This Proof of Concept paper is the first in a series of supplements to the TransitMatters 2018 report entitled Regional Rail for Metropolitan Boston. Throughout this paper, our reference to the term “Regional Rail” refers to the vision we set forth in that 2018 Regional Rail report: a vision of fast, frequent, all day electrified train service for the Boston metropolitan region. Additionally in that report, we explicitly called for the cancellation of the proposed South Station expansion (“SSX”), a $2-3 billion project of practically no transportation value that will cement, perhaps irretrievably, outdated approaches to providing intercity rail service in the Commonwealth of Massachusetts. This Proof of Concept supplement is meant to elaborate on cheaper, more modern alternatives and provide the framework of an approach that we recommend the Massachusetts Bay Transportation Authority (“MBTA”) adopt to achieve our Regional Rail vision.
Specifically, this Proof of Concept highlights how modernized operating practices can maximize train throughput, thus adding capacity to the MBTA’s current commuter rail system. These methods and that additional capacity will permit a large expansion of commuter rail service at a significantly lower cost than many both inside and outside the MBTA and MassDOT presently assume. This capacity question is an urgent one for the Metro Boston region, as passenger volumes on commuter rail are up 20% over the last 6 years. Unfortunately, the MBTA’s only proposed capacity solutions to date are an ill-advised multibillion-dollar expansion of South Station, and an equally ill-advised new order of bilevel coaches to be powered by diesel locomotives that are approaching an average age of 40 years of service.
The adoption of global best practice operating protocols, combined with relatively minor track upgrades, would render the South Station Expansion (SSX) project completely unnecessary, even for the transformative service levels proposed in our Regional Rail Report. Moreover, single-level electric multiple unit trainsets (EMUs) combined with better frequency from faster turnaround times, would enhance capacity without the inherent drawbacks of bilevel trains (sometimes called double-decker coaches), which should be consigned to the dustbin of antiquated mid-20th century “9 to 5” commuter rail systems.
Maximizing Train Throughput:
A Low-Cost Approach to Increasing Capacity at South Station
Our guiding principle, as always, is organization before electronics before concrete. This means that before investing in anything else, the MBTA should immediately take steps to improve rail operations by maximizing train throughput. Some modifications are required to make sure trains can run more frequently, but these modifications involve better scheduling, more reliable electrical equipment, resignaling the terminal zone, and minor trackwork, all of which are significantly less expensive and more cost-effective than relocating property in Downtown Boston to expand the station footprint. With better operations, SSX is unnecessary and its budget can be reinvested in better projects, such as high-level platforms and electrification across the entire MBTA regional rail system. The resources exist; they simply need to be spent wisely.
The essential lesson is that the capacity of a terminal zone is dependent on how fast trains can enter and exit the station and its throat. At both South and North Stations, there are 10 mph speed limits for about half a mile out, which can be lifted to about 30 mph. There are also generous margins of error required by frequent failure rates endemic to the MBTA’s current diesel locomotive fleet, margins that can be specifically decreased with modern equipment.
Rather than run any meaningful level of reverse-peak service and midday service, the MBTA instead sends morning peak trains to sit idle in large, inefficient layover yards within Boston (the second most expensive real estate market on the east coast of North America.) This wasteful practice requires conflicting movements of trainsets across the main line, imposing further strain on downtown terminal capacity. The precious terminal capacity tied up by these practices could instead go toward providing transformative levels of regional rail service to the Greater Boston area.
TransitMatters proposes to eliminate these capacity-killing problems through a cost-effective combination of operational reforms and targeted investments. No station footprint expansion is needed. The trackwork required is at very small scale and entirely within the right-of-way. In the mid-to-long term, following implementation of these reforms and targeted investments, construction of the North-South Rail Link (NSRL) would provide a more direct trip to downtown Boston from the North Side, thus warranting higher frequency of service for all North Side lines and full realization of a transformative Regional Rail vision.
South Station has 13 platform-terminating tracks, which are utilized to serve 20 trains per hour (tph) per direction (inbound/outbound) at peak commute times. Based on current operations elsewhere in the world (as we will explain below), we believe that South Station could be optimized to serve 26 to 30 regional rail tph (that is, 26 to 30 tph in each direction) and North Station could serve 18 regional rail tph (in each direction, again). These capacity figures are only about half as high as these stations’ ultimate capacities. We propose the following as a realistic (but not maximum) capacity schedule in each direction at peak times of the day:
» Franklin Line: 4 tph, either all running via the Southwest Corridor or all interlining with Fairmount
» Haverhill Line: 4 tph
» Lowell Line: 4 tph
» Worcester Line: 8 tph
» Providence/Stoughton Line: 4 tph to Providence, 4 to Stoughton
» Fitchburg Line: 4 tph
» Old Colony Lines: 6 tph, 2 per branch
» Fairmount Line: 6-8 tph
» Newburyport/Rockport (Eastern) Line: 6 tph between the branches
These frequencies of trains per hour do not include Amtrak or other non-MBTA trains but leave sufficient room for them to operate. With our proposed speedup of the Providence Line, the infrastructure has room for 4 hourly slots for Amtrak’s Northeast Corridor services, for a total of 30-34 peak tph at South Station.
South Station’s 13 terminal tracks are sufficient to permit separating the four trunks heading into the station, thereby keeping the Worcester Line on two tracks, the Old Colony on two, Fairmount on two, and the Northeast Corridor (including Needham, Franklin, Providence/Stoughton and Amtrak) on the remaining tracks. (See Figure 1)
At North Station, the current infrastructure is a barrier to completely separating the lines. However, a project to add approach tracks, which would permit separating the Lowell and Fitchburg Lines, is already funded. Unlike SSX, this North Station project is worthwhile, because it removes an infrastructure-based constraint to improve operations. Regardless, traffic at North Station is low enough that the current infrastructure provides more than enough capacity to meet the needs of Regional Rail.
The Terminal Interlockings and Speed
Our projected schedules have trains traveling between South Station and Back Bay in 2.5 minutes, and between South Station and Ruggles in 4.5 minutes. Today, trains are timetabled to take 5 and 8 minutes respectively. This difference is due in part to the assumption of electrification, and partly to speeding up the slowest part of the route - namely, the South Station approaches and terminal capacity limits. While city center terminals such as North and South Station will always face inflexible constraints absent costly expansion, the switches can support much higher speeds than the current 10 mph limit.
There are two primary reasons for today’s conservative 10 mph speed limit. First, as is typical in the United States, many passenger train speed limits are simply too low, a legacy of the steam era, and have never been revised. For example, the extent of legally allowable centrifugal force on a train moving through a curve, which in turn governs its maximum speed, is based on a passenger comfort experiment conducted in the 1950s with New Haven Railroad trains; this outdated requirement reduces allowable speed on curves by 15-30% relative to best practice. Thankfully, the regulations were recently superseded by the Federal Railroad Administration (“FRA”), but the MBTA has not taken advantage of the change.
Second, the design of American switches (or “turnouts”) is handed down from a bygone era and does not properly control for the change in acceleration experienced by a diverging train. Based again on steam-era standards, current American industry standards for switches require the diverging rail to be straight where it crosses the straight rail, a point called the “frog.” In contrast, for example, German switches are curved through the frog and are designed for smoother transition between the straight segments and the curved ones, enabling greater speed through the curves. The point being, our switch turnouts are significantly slower than other world-class train systems. These are resolvable barriers to better train speed and throughput.
It is hard to overstate the importance of removing the slowest speed restrictions, which are in place at both North and South Stations. A half mile at 10 mph takes 3 minutes to traverse. In contrast, at 30 mph, with dedicated tracks to improve reliability, that time is cut by two-thirds; trains in that same half-mile approach would spend a minute going into a station terminal and a minute going out. This is not done today as a result of suboptimal switch design and antiquated signaling circuits, some of which hail from the steam era. These slow zones are unnecessary and relatively easy to fix, alongside fixes to terminal capacity. To add perspective to the relative cost and impact of our proposal, the modest investments in reliability and switch design that we propose can save more time in the last half mile into North or South Station than would Amtrak’s $450 million project to increase top speed in New Jersey from 135 to 160 mph.
The pinch point in the South Station throat is an interlocking called Tower 1. It features a complex of switches called a ladder track: trains from tracks at one end can take the diverging path on a series of switches, thus gaining access to all of the different South Station terminal track options. North Station has an interlocking called Tower A with similar characteristics that allow trains from any track to access any other track at the North Station terminal. While on the surface this seems practical, it is one more vestige of prior century railroading and it slows the trains down, thereby limiting the capacity of each terminal. With all day service by reliable trains and separated track assignments for each line, there is no need for trains to have infinite track options.
Fortunately, the process of reconfiguring the switches to allow smoother, faster travel, called “kinematic gauge optimization”, does not require infrastructure modifications beyond the rails themselves. The switches do not need to be made longer. Modifying the switches to smooth the transition to the curve requires track geometry changes so subtle they can be done within the right of way, without hitting various utility and catenary poles. The project requires laying rails but does not require any of the usual difficult sitework complicating capital construction. Thus, this improvement is relatively inexpensive and has also been recently undertaken and completed by one of our neighbor railroads: Metro-North Railroad recently upgraded 40 mph turnouts to 65 mph at a cost of only a few hundred thousand dollars each.
With the switches so modified, trains could enter and exit the terminals at speeds up to 30-35 mph, allowing trains to clear the station throats rapidly. This alone would serve to increase terminal station capacity, since moving trains in and out faster increases the maximum throughput.
Dedicated Terminal Tracks
The best industry practice at a terminal station serving multiple lines is to separate different lines to different, consistent platforms and minimize interaction between the lines. The reason for this is to isolate delays: if trains on one line are delayed, then they will delay other trains on the same line no matter what, but with perfect separation, delays will not cascade on any other lines.
Unfortunately, MBTA practice is light years away from full line separation. The T prefers being able to connect every line to every terminal track, in order to allow trains on one line to substitute for trains for other lines. The MBTA’s aging diesel locomotives today break down every 7,600 miles, so often that the T assumes breakdowns will happen in the terminal zone as a routine matter. Modern EMUs are far more reliable - the LIRR’s M-7 trains break down every 500,000 miles. With high equipment reliability, it is easy to separate trains between tracks; the T would not need a train on one line to substitute for a train on another. In short, the T is letting the unreliability of its aging diesel locomotive fleet have a negative impact on its ability to adopt best practice line separation at South Station.
Even with today’s rolling stock, it’s unclear that the MBTA’s current practice is necessary or optimal. However, what is certain is that given train reliability levels achieved in London, Paris, Zurich, and other cities with 21st-century practices, rail line separation is the best policy. London is spending considerable money and effort on disentangling different services in South London for this very reason. Switzerland has many small-scale track separation projects at junctions preventing delays on one train from cascading to other lines; the Netherlands is copying this policy at Utrecht, the country’s busiest train station, citing Japanese precedent.
Boston already has the infrastructure for separation, especially at South Station, as the four mainlines entering the station do not cross; unlike London, the MBTA need not spend much money to disentangle its operations. At South Station, it is very easy to separate the Providence and Worcester Lines from the Fairmount and Old Colony Lines; separating each of those lines further is possible, but requires more extensive modifications to Tower 1, which currently has a six-track pinch-point. At North Station, separation is more difficult, but there is less traffic, so that this is less critical.
The bottom line: with the high reliability of electric trains (or EMUs as we propose), and mostly separated tracks between lines, it is possible to run trains with high schedule discipline. High schedule discipline in turn permits running trains more regularly and more often, at least at the frequency levels we have outlined earlier in the Frequency section of this paper.
The Importance of Clockface Scheduling and Rapid Turn Times
A disciplined schedule must repeat on a clockface pattern. This means that if a train runs every 15 minutes and serves a station inbound at 9:05, it will serve it in that direction at :05, :20, :35, and :50 every hour all day. Passengers can memorize these schedules more easily than the complex schedules favored by American planners. Moreover, infrastructure planning is simplified when trains run at consistent intervals, since overtakes and meets on single track are at predictable locations. One Swiss planner humorously put it this way: “We Swiss are lazy, so we plan one hour and repeat it for the rest of the day.”
If trains enter and exit a station throat on a frequent, repeating timetable, and they only occupy the interlocking for a minute in each direction, the maximum capacity of the throat is much higher than current practices allow. Functions such as refueling, which currently require trains to reverse to the yard and not back onto the mainline at downtown terminals, can be handled at suburban terminals and layover yards (and with electrification, refueling ceases to be relevant.)
Turning trains more quickly increases terminal throughput and capacity. American commuter trains turn in 10 minutes at New Haven, and occasionally in Worcester when recovering from delays. Amtrak Keystone trains regularly turn in 10 minutes at Philadelphia’s 30th Street Station. By contrast, in Germany, trains routinely turn in less than 5 minutes. The MBTA can achieve these turn-around metrics as well if it adopts the global best practices we set forth in this supplement.
High frequency all day in both directions, proof-of-payment fare collection, and automatic door opening all combine to increase labor efficiency to the point that train crews can quickly disembark from the train they used to reach Boston, and operate another train ready for departure. This way crews can be perfectly positioned on standby (these are called dropback crews) for departure, reducing turnback times below 5 minutes.
Even accepting 10 minutes as a turnaround time, trains can be scheduled to occupy each track for 15 minutes: 10 minutes of turn time and 5 minutes of approach time and schedule contingency. While far from world class, even improving the MBTA’s turn times to this extent would allow a peak frequency of 4 tph per terminal track. With 4 tph, South Station’s 13 tracks could accommodate 52 trains per hour. Today, peak traffic into South Station is 20 trains per hour per direction, less than half of what is realistically possible, while still being at the lower end of best-in-class railroads.
Why Single-Level Trains
In order to achieve the quickest possible station departure and turn times, the MBTA should use single-level trains, not bilevels. The busiest urban rail lines in the world run single-level trains, and so should the T. Bilevel trains have higher seated capacity than single-level trains which is why they are favored by the old-fashioned peak-focused commuter rail model in service in Metro Boston and throughout North America. Unfortunately, bilevels require climbing or descending stairs to reach the egress doors, which requires trains to have longer dwell times at stations.
Single level trains can have several sets of evenly-spaced doors located along the length of the car. It will help the reader to think of the door placement on subway cars as opposed to the end-only door placement on typical commuter coaches (the older MBTA single level commuter coaches are exactly the wrong model to have in mind). Even the best-made bilevels (i.e. more and wider doors) have longer egress times, which lead to longer dwell times in city center at rush hour. Long dwell times result in reduced capacity per hour. Therefore, lower capacity single level trains actually translate to higher capacity per hour if the single level trains are paired with frequent service (all-day frequent service being the main thrust and assumption of Regional Rail).
In Tokyo, the crowding level is such that practically all equipment is single-level with many doors, usually four pairs per car. Bilevel trains would simply take too long to unload. Berlin and Munich use single-deckers as well on their S-Bahn networks, with three door pairs per car. The Munich S-Bahn does so in a context in which one line has 840,000 riders per weekday, almost as many as all MBTA rail lines combined.
Ultimately, theoretical capacity based on seats per train set is an insufficient metric against which to weigh the merits of single-level versus bilevel cars. At frequencies sufficient to achieve all-day frequent service (a minimum headway of 15 minutes at peak inside Route 128), the excessive dwell times and accessibility challenges imposed by bilevels cancel out their theoretical capacity gains. If passengers fill single-level trains to capacity, the solution is to buy more cars and run longer, more frequent trains. This serves to not only move more passengers, but increase flexibility of the service through more frequent trips. When trains are sufficiently frequent, passengers become relatively indifferent to which train they are on so long as they can get on the next one. This reduces the extent of peak crowding now seen on specific trains.
Though the MBTA is accustomed to ordering and maintaining bespoke equipment, modern trains are more like commodities. Vendors offer modular products, fabricating them at their existing plants with customization for local needs. Such trains have wide doors, weigh about 44 short tons per US-length car, and cost about $2.5 million. The contrast with today’s MBTA equipment is stark. The MBTA’s coaches do not all have automatic doors - conductors manually operate the doors. The aisles are narrow (and easily obstructed) and the doors are at the ends of the car rather than at the quarter points (four evenly spaced doors per car-side), slowing down the boarding and alighting process. Some trains take 5 minutes to fully unload at South Station at rush hour. It can feel like waiting to deplane from the rear of an aircraft. These dwell times completely undermine the speed and frequency required for regional rail to be a functional, competitive, and favorable transportation mode.
We are now well past the point of delaying the decision that must be correctly made in the public interest: Massachusetts must immediately commit to procuring single-level EMUs, starting with the Providence Line and continuing rollout to other lines while in the course of rapid and successive electrification of the entirety of the current and planned commuter rail system. Any further investment in bilevel coaches or diesel locomotives would be, in our view, not merely questionable – it would be irresponsible as it continues a system that is highly inefficient and that, because of its inherent inefficiency, serves as a constant drag on better frequencies and requires unnecessary costly initiatives like SSX. Our proposed Regional Rail operating model would move more people by optimizing frequency gains from single-level EMUs. The existing equipment, both locomotives and coaches, is not compatible with modern operations, and the write-down on its remaining useful life is less than the damage it causes through slow operations and limited capacity.
Nonetheless, assuming a staged adoption of electrification, current coaches with remaining useful life should be reallocated to the non-electrified lines to increase service frequency and capacity to the extent possible prior to electrification and completion of high-level platform construction. They may also be useful for new intercity service to western Massachusetts, or even as far as Albany, at least until such service is electrified.
The Role of NSRL
In our Regional Rail report, we said that while the North-South Rail Link was not critical to implementing a robust Regional Rail system, it would be a “highly useful enhancement providing the flexibility and connectivity to which many riders and potential riders would be drawn.” If NSRL is constructed, frequencies are likely to rise because of an increase in passenger traffic demand (especially on the North Side, as North Station is not in the CBD and South Station is) and the more useful service would induce much greater demand. Thankfully, through-stations do not have problems with terminal interlockings and turn access capacity to which much of this paper is devoted. The following frequencies will become viable upon completion of the NSRL:
Worcester Line: 8 tph on the inner segment to Newton, 4 continuing farther out
Providence Line: 4 tph
Stoughton Line: 4 tph
Franklin Line: 4 tph
Fairmount Line: 12 tph if Franklin trains operate via Fairmount, or 8 if Franklin trains remain on the Southwest Corridor
Old Colony Lines (Kingston/Plymouth, Greenbush, and Middleborough/Lakeville): 12 tph, 4 per branch
Eastern Lines (Newburyport/Rockport): 12 tph on the inner segment to Salem, 4 per branch
Haverhill Line: 4 tph
Lowell Line: 4 tph if Haverhill Line trains continue to operate as today, 8 if they go via the Wildcat Branch
Fitchburg Line: 12 tph on the inner segment to Brandeis/Roberts, 4 continuing farther
NSRL would also allow a further increase in speed, since the tracks would continue through downtown rather than terminating at stub end terminals where trains must slow to approach. The reason to enter South and North Stations at 30 mph (as advised in this report) is that the consequences of overrunning the bumpers are catastrophic. At through-stations, entering at 50-60 mph even in city centers is feasible.
With NSRL in place, only a small subset of trains would still need to navigate the surface terminal interlockings at North and South Stations. Within the tunnel, tracks should be dedicated similar to the track separation we propose for terminal stations, in the sense that one portal only pairs with the Providence and Worcester Lines and another only with Fairmount and Old Colony. This would permit about 24 tph in each direction per tunnel pair, or 48 for the four-track system. The remaining additional trains not traversing the tunnels would use surface terminal platforms.
Schedules would continue to run clockface, except at higher frequency. The S-Bahns in Berlin and Munich have high frequency and almost total through-running and maintain their clockface patterns, as does the Paris RER off-peak.
Single-level trains become even more crucial with NSRL. The minimum headway in the NSRL tunnel is determined by the sum of station dwell time and the time it takes the train to stop decelerating from full speed. Bi-level train dwell times will never support the necessary headways for running a regional rail system through NSRL tunnels.
Ultimately, NSRL is a major booster for Regional Rail. It is not necessary for the basic Regional Rail system, nor for more efficient use of current South Station platforms, which require good operations and electrification. But as a non-trivial investment in concrete infrastructure, NSRL is the logical extension building upon the modernization of organization and electronics as prescribed above because it greatly improves access to Boston and the entire metro area. As such, NSRL engineering must be based upon optimized Regional Rail operations, specifically the use of single-level EMUs. Unfortunately, as we’ve pointed out, the 2018 NSRL feasibility reassessment commissioned by MassDOT grossly understated the benefits of the NSRL by building in assumptions of antiquated operations and equipment in the tunnels, which had the added effect of artificially driving up the projected cost. Moving forward with NSRL without first committing to achieving Regional Rail operations as a predicate would result in a NSRL tunnel that could never live up to its true potential.
Regional Rail is not an unattainable vision; it is an achievable standard for intercity rail in Massachusetts. This new business model for the provision of intercity rail service is an essential component of a regional approach to linking Gateway Cities to Greater Boston, reducing traffic congestion, and providing access to jobs, healthcare, school and affordable housing to Metro Boston residents. In short, Regional Rail is essential to our quality of life.
This supplement to our initial Report has focused on one essential aspect of our approach to Regional Rail: improving capacity at North and South Stations. We have demonstrated how a short list of relatively simple and low-cost actions can significantly improve capacity at the downtown terminals and systemwide, thereby increasing throughput without the need to implement a costly expansion of platforms and tracks. We state clearly: the proposed $2-3 billion South Station expansion is neither necessary nor advisable. The financial resources for transit and rail needs are too scarce to waste on expansion of South Station, and are better spent on measures which will have substantially more impact improving the mobility of Massachusetts and Rhode Island residents. Throughput and capacity can be increased at relatively low cost simply by following the advice we have offered in this supplemental report.
At TransitMatters, we are duty-bound to offer specific ideas and best practice recommendations that will get the job done at the lowest cost possible. Our Regional Rail report, and this supplement do exactly that. We hope that the MBTA will seriously consider delaying South Station expansion and exploring the efficacy of our recommendations in a serious and measurable way. We look forward to working with them and other stakeholders as we advance the effort to bring 21st century Regional Rail to the people of Metro Boston.