State’s rail link study full of flaws

Costs way overstated, benefits minimized

METRO BOSTON’S disconnected commuter rail network is a problem. Without a direct connection between the networks feeding South Station and North Station, travelling between opposite sides of the city and region usually means foregoing rail and travelling by car over congested highways. This access barrier makes it harder for people to get to jobs, or to live and work in the places they desire. For example, the northern suburbs lack direct, efficient access to most of downtown Boston’s jobs, which cluster near South Station and Back Bay Station or in the burgeoning Seaport District. Commuters from south of Boston are held back from pursuing employment opportunities to the north of the city.

Late last year, the Massachusetts Department of Transportation agreed to reevaluate the North South Rail Link (NSRL), a short rail tunnel that would close this access gap and unify the two networks. NSRL would represent a major capacity improvement, lifting the cap on trains per hour into Boston, and providing frequent rail access to jobs and housing across regions. Done right, NSRL would complete a larger reimagining of the MBTA’s outdated and failing commuter rail operating model.

Unfortunately, the MassDOT study, conducted by the global consultancy Arup, reflects NSRL done wrong. The draft report’s numerous flaws stem from a faulty framework which views NSRL as a stand-alone infrastructure project (rather than the logical end point of a transition to a regional rail system), thereby ignoring the connection between infrastructure and service. The report evidences a peculiar disinterest in international best practice, and assumes the continuation of the same outdated, inefficient, and costly practices which shape current commuter rail service. As a result, the draft report overstates the expense of NSRL by a factor of about two, while woefully understating its benefits. We therefore reject claims that this study is, or should be, the final word on the project’s viability.

TransitMatters’s critique of the NSRL cost analysis is informed by our belief that the most useful and meaningful approach to projecting the cost of the NSRL is to assume it will be designed and built in the most cost-effective manner, making use of global best practices. With that framework as our guide, we offer this analysis of the Arup report in order to provide state policymakers and advocates with a more balanced assessment of NSRL’s feasibility.

Electrification benefits overlooked

Perhaps the most prominent failure of the report is its approach to electrification. In order to avoid “burdening” the study with added costs, the scope of study presumes only electrifying up to the first station on each line beyond the tunnel (with the exceptions of the Providence Line, future South Coast Rail, and possibly the Fairmount Line). In practice, this assumption of bare minimum additional improvements imposes higher operating and capital costs than an NSRL built after or concurrently with system-wide electrification. The dual-mode locomotives which would be required under this assumption are extraordinarily expensive (and built by few vendors), adding $2.5 billion to the project cost estimate. In order to accommodate these heavy locomotives and unpowered coaches, the proposed grades are capped at 3 percent or less. This constrains the alignment and limits the possible portal locations. The report’s definitive assumption of dual-mode operation strangely ignores that electrification of the system is squarely on the table for serious consideration in both the short and longer term.

As we noted in our regional rail white paper earlier this year, system-wide electrification would allow the MBTA to replace the entire fleet with electric multiple unit trains, or “EMUs,” where every car is powered and there is no separate locomotive (just as there are no locomotives on the subway). Just as cars with four-wheel drive outperform cars with two-wheel drive, EMUs outperform locomotives and can climb grades of up to 5 percent, potentially allowing for shorter, less expensive tunnels and new portal locations. Indeed, the report’s own cost comparison table notes that EMUs are the cheapest option available. Remarkably, while the Arup study fails to consider a fully-electrified system, it mentions the supposed promise of battery-operated locomotives, a technology that could charitably be described as in its infancy and, in contrast to EMUs, results in heavier trains requiring shallower (and costlier) grades.  We believe that assessing NSRL as part of an electrified system would reveal lower construction costs and enhanced flexibility during both construction and operation (perhaps, for instance, allowing Back Bay Station to remain open during construction).

Questionable cost estimates

The Arup report’s assignment of fleet costs to NSRL further paints a distorted picture. The study implies that new trains are a burden which would be imposed by NSRL and, as such, the project should carry those costs. In reality, as regular commuter rail riders know too well, the T’s existing fleet is unreliable, with much of it already requiring replacement. If anything, full electrification should reduce the cost of rolling stock: it’s cheaper to buy EMUs than to buy locomotives (either diesel or electric) and unpowered cars. Moreover, connecting the north and south parts of the commuter rail system would require fewer trains because allowing trains to continue to run in revenue service increases the possible throughput of available rolling stock.

As our regional rail report observed, full electrification of the mainline rail system would provide such advantages as faster travel times, reduced vehicle and maintenance costs, less particulate pollution, and more reliable equipment. If the Commonwealth is committed to getting the most out of its rail system, electrification and fleet replacement should happen first, as its own project. In contrast, the partial electrification scheme assumed by the report is penny wise and pound foolish, needlessly increasing projected costs by assuming expensive and complicated construction, vehicles, and an operating paradigm that will deliver poor service. A meaningful and useful study must consider how an electrified system would affect the costs and the resulting service, without burdening the cost of NSRL with equipment that would need to be purchased even if it were not built.

The report’s headline infrastructure cost projections are also wildly out of sync with the costs of any actual rail project; we do not believe their figures stand up to informed scrutiny. The cleanest example of this is again electrification, to which Arup assigns a per-track-mile cost of $13 million, using the Caltrain electrification project in the San Francisco Bay Area as a benchmark. The San Francisco project carries one of the highest electrification costs in the world, thanks to unique design flaws. Arup appears to make the major error of conflating the Caltrain project’s route-mile cost with its track-mile cost, increasing the projection by a factor of two. Other projects carry far lower costs: Arup’s projected cost is more than a factor of six higher, for instance, than AECOM’s estimate for Indiana’s South Shore Line.  Amtrak’s electrification of the Northeast Corridor to Boston in the 1990s cost $5.5 million per mile in today’s money, and recent electrification projects in Israel, New Zealand, Portugal, and France cost around $2 million per route-mile.

This incongruity with comparable projects also underlies the report’s tunneling projections, owing largely to steep contingency estimates. The report pegs the cost of a four-track tunnel at $12 billion in today’s money, or $17 billion in 2028 dollars. At about $1 billion per single-track mile, this would be the highest cost in the world, taking both the stations and the tunnels connecting them into account. This projection is contradicted by the findings of last year’s independent NSRL study conducted by Harvard Kennedy School researchers. That study estimated the total project cost of the four-track option to be about $6 billion in 2025 dollars, or $5 billion in today’s dollars. The Harvard researchers relied on three separate methods, using both real-world costs of urban rail tunnels around the world and standard cost items used by the federal government to estimate infrastructure costs.

The study unpersuasively challenged the Harvard analysis, asserting that it did not measure the appropriate length of tunneling. That is not the case. There is nothing on the face of what the Harvard Kennedy School did – measuring the individual length of each of the four tunnel tracks at three miles – that is improper or inaccurate.  It is true that the Harvard study approached this in a different manner than Arup (which measured the entire route length), but that does not mean the Harvard approach is wrong. MassDOT also claims that Harvard did not account for contingencies for risk of cost escalation. But the Harvard study did make allowances for risk, at a 50 percent contingency on top of real-world cost comparison items. This approach looked at actual expenses, after any cost overruns, rather than hypothetical costs. Thus, the Harvard cost projection reflects the reality of what comparable projects cost, as well as a robust contingency. Based on what we have seen, the study’s criticism of the Harvard analysis is either misplaced or insufficient to diminish its contribution to the discussion.

Arup’s contingency assumptions appear to take as a given that MassDOT and the MBTA will be unable to manage costs. Estimating (and being prepared for) contingencies is an essential component of any major project. However, Arup’s estimates are distorted by outlier project failures, and an assumption that Massachusetts cannot bring construction costs in line with global standards, which are lower than the American average for rail projects. The costs of Phase 1 of New York’s Second Avenue Subway, a trouble-plagued project with myriad delays and cost overruns, and the failure of a tunnel boring machine in Seattle, impacted Arup’s formula. While it is important to learn lessons from other projects, a failure of the magnitude of the Seattle disaster is quite rare, and that project was undertaken by an inexperienced contractor. We do not see these as examples that ought to carry inordinate weight in assessing NSRL contingencies.

More to the point, this approach ignores how Massachusetts dealt with serious cost pressures emerging in the GLX project.  If we were able to resolve a nearly billion-dollar projected cost increase in GLX, what makes anyone think the same approach and determination cannot be put to bear on NSRL, or any other project?  The MBTA and FMCB must insist that projects be designed, engineered, and constructed based on global best practices, and the most reliable cost estimates are those that make reasonable assumptions, and do not rely largely on worst case scenarios.

Downplaying NSRL’s benefits

The study is further seriously compromised by multiple errors of analysis and faulty messaging, especially regarding the NSRL tunnel’s capacity. The most glaring of these is the claim that service frequencies of 16-17 trains per hour on each track have not been operated on lines with positive train control (PTC). This is false: the North River Tunnels under the Hudson River accommodate 24-25 trains or more per hour per track in the morning peak direction. This example employs Amtrak’s ACSES PTC system, the same system that is operational on the Providence Line and with which the MBTA’s PTC installation is intended to be interoperable. Moreover, the emerging global standard for rail signaling, the European Train Control System, which meets the American regulatory definition of PTC, has enabled some services in Switzerland to run less than 120 seconds apart.

The study appears to portray expansion of South Station as more economical. However, it does so by grossly downplaying the benefits of through-running, even though this practice is NSRL’s central operating benefit. The report assumes a maximum frequency of 17-21 trains per hour per direction (not including Amtrak slots). At a June public meeting, Arup and MassDOT clarified that this service level was not reflective of the maximum possible throughput, but instead reflected limitations on individual line throughput and the imbalance in possible pairings between the northern and southern lines (there are seven lines to the south versus four to the north). These issues are resolvable, through incremental improvements and scheduling of tunnel slots based on demand for through service. Yet the initial release of information did not make this clear, instead only appearing to detail the minimum downstream capacity improvements to enable the 16-17 trains per hour frequency. In other words, MassDOT took only the worst-case scenario and failed to consider the approach any competent operator would take.

In another example of artificially bolstering the case for surface terminal expansion, the report also presumes a need for added layover capacity downtown. However, an alternative allowing Old Colony and Fairmount trains to use the tunnel would eliminate this need; indeed, it would render such a layover facility wasteful. By ignoring this elementary consideration, the report undersells NSRL’s benefits, doubles down on old approaches to operating intercity rail, and artificially maximizes its costs.

Rejection of best practices

The world’s best and most efficient regional rail networks, whose practices inform our regional rail vision, take full advantage of their infrastructure to provide service at rapid transit levels, using modern electrified trains, frequencies akin to those on downtown subway lines, and scheduling and fare integration with the rest of the transit system. Through-running makes such a vision even more beneficial and transformative: continuing to terminals outside of the central city both increases the capacity of lines and increases the available destinations. Yet the report assumes that the MBTA will take advantage of none of these benefits; that NSRL would have to accommodate the slow, antiquated, and unreliable operational paradigm of shuttling 9-to-5 commuters on expensive equipment. How can we rely on a report that assumes the worst outdated practices will continue into the 21st Century? Does MassDOT believe in a future where only the bare minimum improvements will be made to the intercity rail network?

This indifference to modernization creates serious environmental justice and regional equity concerns. Take the Fairmount Line as an example. This line serves dense, lower-income, high-pollution portions of Dorchester, yet only the most expensive option studied includes a second southern portal which would allow Fairmount Line trains to access the tunnel. The fast acceleration of EMUs is most advantageous on lines with closely-spaced stops like Fairmount; dual-modes have similarly poor acceleration to current rolling stock. Electrification would also eliminate diesel emissions, a major source of pollution along this corridor. We have ample evidence that modernization provides real, quantifiable, immediate benefits: when the MBTA increased off-peak frequency on the Fairmount Line to hourly and expanded the 1A fare zone to cover the line’s entirety, ridership tripled from its previous historically dismal point. NSRL would achieve the goals of increasing possible frequency and providing access to jobs north of South Station (the study’s own projections predict success in running Fairmount trains through to the Fitchburg Line.) Given the report’s low projected price point of $50 million for a second portal, it is curious that Fairmount was not included on any of the three other alternatives. The Fairmount Line must be a priority for a through-running regime, not rendered a secondary option.

To put all this simply: this reassessment assumes that the MBTA will, in perpetuity, operate the intercity rail system in the worst possible way, ignoring modern best practices. These failures of analysis serve to raise costs, decrease NSRL’s benefits, and depress projected ridership. In fact, MassDOT’s narrow parameters are divorced from the MBTA’s own aspirations of making full use of its existing mainline rail assets.  MassDOT and the MBTA are currently undertaking a comprehensive commuter rail vision study, which will include an assessment of the viability of a regional rail business model. If NSRL is analyzed against a regional rail model, the costs would almost definitely be lower, and the outcomes would be far more efficient, viable, and appealing than the bleak outcomes projected by this constrained analysis.

Reliance on faulty modeling

Finally, the study’s ridership model is plagued by status quo bias and has been discredited by many experts, including Transportation Secretary Stephanie Pollack and several FMCB members. The model’s treatment of commuter rail as an “independent mode,” divorced from the standard methodology used to evaluate rapid transit, renders it entirely inappropriate to assess the benefits of a modernized and unified regional rail system. For example, most central business district jobs in Boston are near South Station, Back Bay, and Ruggles, and yet the model predicts that two-thirds of riders from the north will disembark at North Station. Nor does it account for the changes to commuting patterns which would be enabled by connecting the disconnected housing and job markets to the north and south of Boston. With such a glaring error, it’s hard to trust the model’s overall ridership projections.

The way forward

The NSRL feasibility reassessment is a seriously missed opportunity. It ignores best practices and, in so doing, projects outrageously high costs. Caution and diligence are necessities for delivery of any project, but that diligence must be informed by the on-the-ground reality of comparable undertakings. The way forward is emphatically not for policymakers to throw up their hands at jaw-dropping cost projections designed to scare the public away from support of NSRL, but to critically question the assumptions underlying them, and demand a comprehensive assessment of the ultimate deliverable service from infrastructure investments.

We believe that the following course of action is necessary: MassDOT should acknowledge the limitations of Arup’s study and its scope, and, crucially, those of the ridership model employed, with a commitment to identify a more suitable analytical tool. Once this tool is identified, it can be applied to all analyses of the commuter rail system, including the current commuter rail vision study. When that informed analysis is completed, a reassessment of NSRL must occur, guided by knowledge of best operating practices and drawing from conclusions arising from a high-leverage regional rail operating model.

The economic and environmental future of metro Boston depends in large part on a transition from the current failing and outdated commuter rail model to a modern business and service delivery model, as outlined in our regional rail report. NSRL done right would leverage our intercity rail system to provide the best possible frequencies and regional connectivity. We’ve got to take the steps to get these analyses right.

Meet the Author

Transit Matters

Transit advocacy organization, TransitMatters
Advancing change is never easy. That’s why it’s unacceptable to erect artificial barriers through reports relying on faulty assumptions. Our public servants and the consultants they employ have an obligation to explore the most cost-effective and operationally efficient practices and put them into use. The public interest – indeed, our very future as a region – depends on it.

This article was written by TransitMatters members in a collaborative effort led by Ethan Finlan, with contributions from Ted Pyne and Alon Levy. It is excerpted from a forthcoming report, to be released in 2019 as part of TransitMatters’ Regional Rail initiative.