Phasing the Grand Junction viaduct

In a previous post, this page explored how the Allston Turnpike relocation could be built without a large road viaduct. Thanks to input from several sources, here is a very rough idea of how it could be phased, with no interruption to Turnpike traffic or Worcester Commuter Rail line (aside from quick weekend-long changeovers) and only minimal closures (in the months range, not years) to the Grand Junction railroad. Note that while these are largely to scale, they are by no means engineering drawings, and that (especially laterally) there may be changes, as I am estimating this off of measuring aerial photographs on Google Maps, not actually surveying the highway.

I am first going to show the elevations for each phase, crossing at a point in the middle of the current viaduct. There will obviously be differences at either end, where roads and railroads will have to ramp up or down to reach the construction phase. However, those will be shown in the aerial views, further down. In each of these, the Charles River is on the left, and Boston University is on the right, in other words, they are looking east. The scale in feet, with a grid showing every 20 feet horizontally and every 10 feet vertically.

Current conditions have the Turnpike on a viaduct over the railroad, Soldier’s Field Road (SFR) and the bike/ped facility (Paul Dudley White Path, or PDWP).

In Phase 1, several steps would occur:

  1. Soldier’s Field Road is relocated towards the river and potentially narrowed slightly (Memorial Drive is 10 feet narrower than SFR yet still has four lanes of traffic). The PDWP is moved slightly towards the river.
  2. The DCR “parkland” between SFR and the Turnpike—currently weeds and gravel—is excavated approximately four feet and three westbound Turnpike lanes are build there. Three lanes are viable during the project, as the current three westbound lanes (owing to the Commonwealth Avenue bridge project) are suitable for traffic volumes. This grade is four feet lower than the current Soldiers Field Road, level with the lowest point where the Turnpike crosses under Commonwealth Avenue.
  3. The Grand Junction is retained under the westbound Turnpike lanes. A new, permanent bridge is built for the railroad over SFR, with extra width to accommodate a bicycle/pedestrian path looping up to the BU Bridge (and eventually the “People’s Pike” along the Grand Junction to BU and Allston). The new bridge is also built for the Grand Junction under the existing Turnpike viaduct, while the Grand Junction’s current connection is kept to the north. A ramp down to grade is also built below the current viaduct.
  4. Once at grade, the Grand Junction alignment is moved slightly south to allow excavation and construction for foundations for supports for the new Grand Junction viaduct. These can be built during or after the Turnpike viaduct is removed.
  5. The PDWP is relocated under the new, wider Grand Junction bridge along SFR, eliminating the dangerous (narrow, bumpy, slippery when wet, poor sight lines) boardwalk under the BU Bridge. There is plenty of room under the BU Bridge for both a level path and a ramp up to parallel the Grand Junction (24 feet, enough for each path to be 12 feet wide).
  6. The Turnpike is moved under the new Grand Junction bridge and on to the temporary alignment. At this point the Grand Junction’s current alignment is removed and service is suspended until the viaduct can be removed above the easternmost section of the Grand Junction. Once this is done, the rest of the removal can proceed with the line in service, with care to be taken to allow for daily and as-necessary traffic on the Grand Junction. 
  7. The westbound Turnpike ramps back up to the viaduct near where the current exit ramp leaves the mainline. A spur track is built under this westernmost area to serve Houghton Chemical. 
Phase 1A (above) shows how temporary supports would be built on the south side of the highway (near BU) and the right of way slightly widened to allow for the Worcester Line to be relocated from underneath the viaduct structure.

Once Phase 1 is completed, half of the Turnpike has been removed, and the supports for the Grand Junction viaduct have been put in to place. In this phase, second half of the Turnpike comes down. For this to occur, the following has to happen:

  1. The three westbound Turnpike lanes are kept as-is, and four eastbound lanes are built between the current eastbound structure, above the location of the Grand Junction viaduct.
  2. The Worcester Line is moved to the outside of the Turnpike structure. This may require temporary supports and/or the temporary taking of some land from Buick Street behind Boston University. This will need to be explored further. Another ramp is built under the to-be-demolished Turnpike, again for the Grand Junction, and a temporary bridge across the location of the eastbound lanes. The Grand Junction will be placed out of service while the viaduct is demolished above this grade; again, this would be the first priority to minimize disruption.

At this point, the main highway work is mostly complete, with the ability to have four 12-foot lanes for each direction of highway in the main cross-section, as the eastbound Turnpike lanes are built at the same level as the westbound lanes where the viaduct previously stood. The Grand Junction is again put on a temporary ramp (with a potential short shutdown), this time alongside the Worcester Line railroad. Some extra room is left in the median of the Turnpike to allow for construction of the Grand Junction viaduct. This could be pre-cast off-site (perhaps in the Allston yards area) and assembled at off-peak times with temporary lane closures.

Once the Grand Junction viaduct is built, several punchlist items are completed:

  1. The bike path and Storrow Drive are moved four feet to the south, as the westbound Turnpike is narrowed slightly taking up space previously used for construction of the viaduct. 
  2. Both sides of the turnpike now reach their final alignment with four twelve-foot lanes with three-foot shoulders (wider than most of the Turnpike extension today; so perhaps a few feet of this leeway which could be better used for the PDWP), and with far better grades and sightlines than the current highway; important since with open road tolling vehicles will no longer be slowing down or speeding up at the toll plaza.
  3. The Worcester Line is double-tracked lowered to the same grade as the Turnpike (or perhaps a couple of feet lower) to allow future overbuild.
  4. Supports for a wide bicycle/pedestrian/park facility are built over the eastbound Turnpike and railroad tracks, connecting Allston to the BU Bridge and PDWP, and eventually along the Grand Junction bridge to Cambridge and beyond.

This is the final elevation of the project. Here are the corresponding aerial views:

The approximate location of the cross section is shown. A is on the right side of the cross sections, B on the left. Colors on the aerial views match those on the cross sections.

What if the Allston Viaduct was rebuilt … without a huge highway viaduct?

I’ve gone to my fair share of meetings and written quite a bit about the Allston viaduct project.

The long and short of it: $250 million for a replacement viaduct basically replicating the old one, and maybe some transit, walking and biking improvements if they can “find” the money. Everyone agrees the project has to happen: the viaduct is falling down. There are preliminary plans, and arguments over how big and wide of a viaduct to build.

But what if you didn’t build a big, wide viaduct at all? If you think outside the box (as this page has before), there’s the potential to save money. Considering how much cheaper it is to build other roads at grade, potentially lot of money. $50 million. Maybe more. The resulting highway would be safer, have better grades and sight lines, and integrate better in to potential development in Allston. It just requires a slightly change of the state of mind.

I won’t blame anyone for not realizing this earlier. I’ve been writing about this for years and it took a prompt to see if everything would fit at grade to figure it out. (The answer is you’d be about 30 feet short of lateral space to put everything at grade, and you still need to cross the Turnpike and Grand Junction.) But once you change you frame of reference, elevating the Grand Junction over the Turnpike, instead of the other way round, makes a whole lot of sense.

$250 million is a lot to spend for a mile of roadway; much of that cost is in building a high-and-wide viaduct, supports and steel and drainage, and said viaduct then costs more to maintain. This project has a lot of moving parts (bike/ped paths, Soldiers Field Road, Mass Pike, four railroad tracks going to two separate destinations) in a narrow corridor, including one (the Grand Junction branch) which has to cross another (the highway). But what if you could do it all without a hulking, expensive highway viaduct? An at-grade highway would save a lot of money. It would cost less to maintain. And it would remove the Great Wall of Allston between the river and the neighborhood.

With the state budget deficit and without increased gas taxes for infrastructure spending, we need to fully analyze projects so they are as fiscally responsible as possible. In this case, the project is necessary—the viaduct is the same age as the crumbling nearby Commonwealth Avenue bridge—but replacing a viaduct with another viaduct is far more costly than moving as much of the project as possible to grade. There is a way to do this—to build much less elevated structure—which would likely save tens of millions of dollars. We need to seriously consider it, although so far any project which does not include a large, wide road viaduct has been dismissed out of hand. That must change.

I think it comes down to priorities. With so many moving parts, there needs to be some back and forth. Certain things are immutable: you can’t get rid of the Turnpike, or the railroad. Others—what goes where, construction impacts, and the like—can be changed. So here is a rundown of priorities, roughly ranked:

  • Retain Turnpike capacity at completion.
  • Retain Worcester Commuter Rail at completion.
  • Retain two-track right-of-way for Grand Junction at completion.
  • Retain Paul Dudley White path along river.
  • Provide connectivity for a future West Station
  • Minimize cost.
  • Maximize economic development opportunities by minimizing above-grade land use for transportation infrastructure in the Beacon Park Yards area.
  • Minimize disruption to rail and road traffic during construction.
  • Improve quality of life for surrounding neighborhoods. 
  • Add width to highway to improve emergency lanes and sight lines.
  • Build a “People’s Pike” connecting the river to Allston and Boston University
  • Improve parkland along the Charles River.

The first four items are the immutable ones. A project which fails to address them is dead on arrival; Turnpike is not going to become an Arborway-like boulevard with crosswalks and bike lanes (at least not in the next 50 years). But beyond that, it gets more interesting. The most complicated piece of the puzzle is getting the Grand Junction rail line from the south side of the Turnpike to the north side. Since 1962, it has gone underneath the roadway, and perhaps because “we’ve always done it that way,” all the plans for the future have the same scenario. Yet this requires a large, wide viaduct, and those don’t come cheap. And even setting cost aside, it turns out that it might not even be the best way to build the project anyway.

If you take a few steps back, the highway viaduct just doesn’t make sense. The Grand Junction is two railroad tracks wide, or 30 feet. The Turnpike is, at a minimum, 110 feet wide. Why not put the turnpike at ground level, and build a viaduct that is one quarter the width ? Does it make sense to put the wider, higher, larger structure above the lower, narrower one? No. Elevating the Grand Junction would be far less expensive.

What’s more, on the eastern end of the project area, the Turnpike passes underneath Commonwealth Avenue, while the Grand Junction passes over Soldiers Field Road. 1000 feet out from the narrowest part of the corridor, where one has to pass over the other, the Grand Junction line is 16 feet higher than the Turnpike. Yet the railroad slopes down and the Turnpike ramps up, as they zigzag in vertical space to attain the necessary grade separation. From a terrain standpoint, putting the Turnpike at ground level just makes sense.

In the current setup, coming from Cambridge, the Grand Junction ascends to cross Soldier’s Field Road, and then descends sharply at a 1.3% grade to move under the viaduct. The roadway, on the other hand, ascends at a 3.5% grade to the viaduct, and then nearly as sharply at the other end to reach the toll plaza. This results in limited sight lines on the Turnpike. In the proposed Grand Junction viaduct, the ruling grade of both the road and the rail line is reduced; the rail line requires a 1.1% grade towards West Station (although this could be mitigated by slightly raising the elevation of the Worcester Line tracks), still better than the current 1.3%. And the highway has no grade steeper than 0.5% west of Comm Ave, where before it was seven times as steep, improving sight lines and safety.

The charts below show the gradients for the current and proposed scenarios. The map below corresponds which each of the letters shown on the charts (data from MassGIS LiDAR data). Note how the Turnpike starts and ends at a lower elevation than the railroad. It makes no sense to have it resemble a rollercoaster.

What we have right now is a half-mile-long, eight-lane-wide viaduct to cross a double-track railroad right of way. It’s as if a highway were tunneled under a small stream instead of going over it on a small bridge. For whatever reason, a highway viaduct may have made sense in 1962. It doesn’t today.

MassDOT “considered” elevating the railroad, but they dismissed it as infeasible. Why? Because they assumed that all four tracks of the railroad would have to be elevated. I don’t think it’s nefarious; I think everyone’s mindset is that the highway has to go over the railroad, because that’s how it’s always been. But the Worcester tracks stay on the same side of the highway as the whole way: they can stay at ground level; there’s no need for grade separation. The Grand Junction is much easier to elevate, and requires a much narrower structure, too. Here are three slides (from here) dismissing the overhead rail line as impossible, and my annotation on each as to why it is not the case. (Click here for full size.)

One issue is that MassDOT “requires” 135 feet of highway width for full-width shoulders. While this is “Interstate standard,” waivers can be granted for constrained areas, and much of the Turnpike (including the existing viaduct) doesn’t have any such lanes. If the highway is kept at grade, it will have much gentler grades and better sight lines. This added safety will mitigate the narrow width—and such a compromise would allow for better use of the corridor as a whole.

So you should kee the Grand Junction line high, and the highway low. Westbound drivers would cross under Commonwealth Avenue and then stay level. From their right, the Grand Junction line would ascend slightly—about 5 feet at a 1˚ grade—and pass on to pedestals over the center of the highway. It would continue down the center until it curved off to the left towards West Station and descended to grade. This is not a novel concept: the JFK AirTrain in New York operates in a similar manner in median of the Van Wyck Expressway. And since initial foundation work could be completed under the existing overpass early in the project, the bottoms of the pedestals and median could be pre-built with minimal impacts to traffic; less of an impact than shown here. One note: the AirTrain uses lighter vehicles than the Grand Junction, so it may require more closely-spaced supports and a somewhat thicker deck akin to what is used for other “modern” T bridges like the Eastern Route crossing of the Mystic River.  (The photo on the right is of construction of the Van Wyck AirTrain, from this site.)

Which of the following is more inviting?

Since the AirTrain is elevated above cross streets as well, it’s quite a bit higher than a viaduct would be in Allston, what a cross-section would look like from the Paul Dudley White Bike Path is similar to what it looks like when it crosses a cross street. That’s a bit less obtrusive than this. Or see below:

The photos above are actually taken from approximately the same distance away. Note how much more of the sky you can see above the Van Wyck. Wouldn’t you rather have something like the image on the right?

Not only is it one third the width, but it appears (and, in fact, is) lower for two reasons. In the nearby Prudential Tunnel, roads require about 14’3″ feet of clearance, trains 16’9″, so a road viaduct over a rail line is higher than the other way around. Second, trains don’t require guard rails. Put together, these add up to the top of the structure being six or seven feet lower for a rail structure. Much like how the Somerville Community Path is being built as part of the Green Line Extension, the “People’s Pike” path, it could cantilevered off the side of the Grand Junction viaduct (with additional support extending to the higher ground in—and connections to—Allston, if necessary). For an observer standing on the Paul Dudley White bike path, a rail viaduct in the median of the highway would be visible up to 6˚ in the sky, and 15˚ when a train passed over. By matching the initial grades, the highway can be lower than the Grand Junction since they don’t have to swap positions. The nearer and higher highway alternative would be visible in 20˚ of the sky, and 25˚—nearly a third of the way to vertical—when traffic is taken in to account. The smaller rail viaduct would cast many fewer shadows across the bike path and river as well.

Here is a cross section of my plans and the viaduct plan. Colors denote uses, and are scaled to the height of the users/vehicles. Scale in feet:

By elevating the Grand Junction, there is enough room for the rest of the uses:

  • 14 feet for the Paul Dudley White path (an increase from 8 to 10 feet now, narrower in places)
  • 24 feet for each side of Soldier’s Field Road, with a two foot median and two feet separating the bike path. This is the current width, and the road would need little modification overall.
  • A four-foot barrier between Soldier’s Field Road and the Turnpike.
  • 54 feet of travel space for each side of the Turnpike, enough for four twelve-foot lanes, a two foot left shoulder and a four foot right shoulder, with a six foot median where the supports for the rail line sit. Including barriers, this is 11 feet more than the current 107-foot-wide viaduct, although it is not as wide as the proposed MassDOT viaduct with full shoulders. However, by mitigating the steep grades and sight line issues, the road could be designed in this section with these narrower shoulders. With 11 foot lanes, an 8 foot shoulder could be accommodated.
  • Six feet for the concrete pedestal supports for the box-girder structure for the Grand Junction rail line, which is 30 feet wide.
  • 30 feet for the Worcester Line right of way, which would be relocated only slightly once the viaduct abutments are removed.

From above, it would look something like this (with just a sketch of ramps and street grid in Allston, which would built above grade):

And a focus in on the narrow viaduct area:

It would also have benefits in the Allston area. In current plans, the Turnpike viaduct has to slope gradually down through the Beacon Park Yard area. If it were at grade, it could be easily decked over, and ramps could go up to a street grid on that deck. The Grand Junction would have to descend towards West Station, but it has a much smaller footprint and would start its descent from a lower elevation, thus it would much more easily integrate in to the surrounding area.

Construction would also be much simpler, because instead of a multi-stage process where the viaduct was unbuilt and rebuilt at the same time, it would simply be unbuilt in stages, with the new roadway built on the ground below. The current staging plans call for a new, wider viaduct to be built in stages, two lanes at a time, while the old bridge is being torn down, including many temporary supports and other engineering issues. By putting roadways at the surface, you could accomplish the entire task without building any new bridging, in fewer stages (dismantling four lanes at a time instead of two), and only shoring up a few supports temporarily. It’s even possible that pedestals for the Grand Junction—which would likely be built under the current viaduct before other construction—could be used during construction to shore up the viaduct as it is dismantled.

As for the final product, rather than a 110-foot-wide, 30-foot-high viaduct, you’d have a 30-foot-wide, 24-foot-high one. The rail structure would have just 22% of the mass of the road viaduct. Road noise would be lower to the ground (with the potential for an overbuild park to cap noise all together), and the viaduct would be further from the river and from buildings at BU. And ongoing maintenance costs would be far lower. The main loss is the inability to tuck some of Soldier’s Field Road under the viaduct and create slightly more parkland in the vicinity of the viaduct. But for a cheaper project, you get a much smaller viaduct which casts fewer shadows over this area, so the bike path is a more pleasant experience.

The only major issue would be a relatively long-term closure of the Grand Junction branch, so MBTA operations and a small amount of Everett-bound freight would have to move via Ayer and Worcester. This happened recently when the Grand Junction bridge was undergoing emergency rehabilitation for several months; the biggest issue was the 10 mph speed limit on Pan Am trackage between Ayer and Worcester. (The railroad out to Worcester is Class 3 and permits 60 mph speeds, the rebuilt railroad inbound from Ayer will mostly be Class 4 with speeds of 80 mph.) With a small portion of the money saved from not building a road viaduct, that track could be upgraded to Class 2 (30 mph) trackage to save considerable time with equipment swaps and improve a freight rail trunk line as well. It would also be an opportunity to fully rebuild the Grand Junction bridge, and other portions of the corridor to Kendall Square and beyond. (One other minor issue is serving Houghton Chemical, which could be accomplished, by relocating Houghton Chemical to the other side of the highway as depicted above. In the “armpit” of the Turnpike and railroad, it would be a perfectly good site for such a use, sustain a family business in the area, and free up developable land near the river.)

Once you take a step back from the current situation and think about this, it’s obvious. It makes no sense to build a wider, higher viaduct for a roadway when you can build a narrower and lower one for the railroad. All it requires is some grading underneath the viaduct, construction that is certainly no more complex that what is being proposed. A surface roadway has a much longer lifespan and lower maintenance costs than an elevated one, and it provides a much less obtrusive—and more future-proof, as it fits in better with potential development in Allston—structure as well. It’s a better project. Best of all, given the cost differentials for projects like the Casey Overpass elimination (and McGrath and Bowker), it could probably be built for $50 to $100 million less than the viaduct, with future savings from lower maintenance costs.

That money would go a long way towards full completion of this project, mitigation steps, and ensuring the safety of other roads and bridges across the Commonwealth.

Beyond Better Buses: Build a Better Network

This is the final in a series of articles about the place and practicality of bus rapid transit in Boston. Previous installments can be found at this link, or below:

and finally, this conclusion discussing how to build a better transportation network.

The Boston BRT report’s heart is in the right place. They want better transit. But we don’t just need a better bus rapid transit network. We need a better transportation network. Let’s not fight mode wars, let’s leverage the resources we have, and work towards a better network and better transit all over.

A recent article in NextCity put it well: the US can’t afford “nice” transit, so people fawn over BRT. And it works, in places. But this is a false dichotomy; it sells ourselves short. If we push BRT when another option would make more sense, it’s a square-hole-round-peg solution and we’re bound to have a system that is either underutilized or over capacity. No one mode is always the answer. If someone publishes a study positing that, we should ask if it is actually proposing a solution, or an agenda.

What Boston does need, wholesale, is a better transportation network, with improved hub-and-spoke routes and, especially, better connectivity to growing urban activity centers.  In the early 20th century, transportation infrastructure focused on the center of the city, with high capacity transit converging downtown. Later in the 20th century, most transportation infrastructure focused growth on the outskirts, accessible only by road. (Every so often someone proposes bus rapid transit or a monorail or something along 128, but that’s a lost cause. The employment density, with offices scattered amongst sprawling parking lots a mile from the roadway, is way too car-oriented for effective transit of any type. Last mile shuttles can provide decent connections for some employees; most others will have to brave traffic; any other transit is massively subsidizing car-oriented development.) Many other cities are continuing down the downtown-suburban split: even in San Francisco where tech companies either hole up in downtown towers or sprawling suburban office parks all but inaccessible by transit and reliant on highly subsidized corporate shuttles which are still at the whims of traffic.

Boston has managed to establish growing, dense and urban recent development focused in Longwood, Kendall, the Seaport and similar areas. While not as accessible by transit as downtown, they’re close enough that the just last mile needs to be solved or enhanced to leverage the existing transit network. (Apple, for instance, is building a “transit center” in to their huge new campus. That’s the last mile. Unless they can build a bus lane on 280, they just have to figure out the first 40; their goal is to have just 2/3 of their employees driving alone to work; neither Kendall, the LMA or even the Seaport is nearly that high.) Boston is lucky: many of fastest growing large employment centers are dense, transit-oriented and close together.

They just need more and better transit. Boston needs to go from hub-and-spoke to a network. It’s a hard choice to make, and system expansion needs to dovetail with system maintenance; one can’t cannibalize another. But while Boston’s hub-and-spoke network doesn’t serve the next century particularly well, that can be remedied, and improve transit for all. And better buses are certainly part of this solution.

But only part. In the past two posts, I’ve gone through many of the routes proposed by the ITDP and myself, and examined which mode would be best, how they would interact with the current transportation system, and how they would form a better network. And, as I am wont to do, created a map:

There’s a lot of BRT on that map. It acts as a feeder service to transit lines, as crosstown routes, and to speed transit through major chokepoints. Of course, none of these would likely qualify for “gold standard” BRT. All would be cheaper, and are in corridors where they are the right solution, not something that would work well in Bogotá or Mexico City transplanted in to Boston. The rest of the network builds much of the Urban Ring—not with a zigzagging overbuilt bus route—and adds significant capacity to the system. It solves huge last mile issues to the commercial nodes in the city, and good circumferential routes will take a lot of connections and take pressure off the downtown routes. It leverages huge portions of the existing network—especially the Commuter Rail lines—and brings them closer to the quickly-growing areas. By doing so, it brings much more housing, both in the cities and the suburbs, within a reasonable transit commute of most major downtown employment nodes.

It’s a network. It connects people to jobs. It encourages mode shift. It provides system redundancy, so if one line has issues, there is another way around. It brings good service to underserved neighborhoods, it puts many more people’s jobs within the reach of commuter rail termini, and it doesn’t force everyone to transfer through downtown, or take a slow bus through rush hour traffic. It is not focused on one mode over another: some work well with buses, others with light rail, others with heavy rail, and still another (the Grand Junction and its extensions) with an RER-style commuter line operating at high frequency. It provides the kind of system the city needs to grow without overstressing the infrastructure we have now. But it doesn’t put all the eggs in one modal basket.

If we are going to have dreamy, long-term proposals about transit in Boston, let’s at least have some that fit in with the system we have, not the system on another continent.

How to improve transit (including buses) in Boston, part II

In the first installment of this post, we looked at four corridors highlighted by the ITDP for “gold standard” bus rapid transit. I agreed that those corridors need investment, but questioned whether BRT was the right technology for each. However, I think there are many, many parts of Boston where better bus facilities could dramatically improve service. It just might not conform to the “gold standard” the ITDP requires (and, uh, made up).


Here again are the non-ITDP bus corridors I will examine.

Bus improvements High
Frequency
Straight Street
Width
Bypass
Traffic
Connectivity Last
Mile
Right
Mode
White
Elephant
Arlington to Harvard N/A
Central to South End via Mass Ave N/A
Arsenal-Western (70 Bus)
Mount Auburn St (71/73 bus)
111 Bus, N. Wash Bridge & Tobin
Northbound I-93 express buses N/A
Huntington Ave (39/66/E Line)
Ruggles-Jackson Sq-Seaver-Ashmont N/A
Washington St, Quincy N/A
Union Sq-Kenmore
O’Brien Hwy during GLX Busing



And here they are on a map (orange: ITDP; green: others). Details below.

Mass Ave. Arlington to Harvard. The stretch of Mass Ave north of Harvard Square is wide, traversed by the often-crammed 77 bus (and others), and almost entirely given over to cars. The street had safety islands for boarding streetcars until 1955; at the time, the City of Cambridge was the main antagonist against the use of such islands. (Safety islands or “safety zones” were constructed in the middle of wider streets with a large concrete pylon at the upstream end to provide a safe waiting and boarding area for streetcar passengers while letting the streetcar stay in the center of the street; a sort of proto-BRT in the early 20th century. An impediment to cars, they were removed with the streetcars so that traffic could flow better as buses pulled to the curb, slowing transit further. Examples here and here.) Since then, Mass Ave has had a median added and been given four wide travel lanes and two parking lanes; most of the street north of Harvard is 72 feet wide.

And sometime in the next few years, the street may be up for full reconstruction. While “gold standard” BRT is unlikely, a center-running bus lane would be possible (if traffic lanes were reduced to one in each direction). This would allow the 77 bus—and the 96 from Davis—to run without delay through all of Cambridge. Many passengers don’t transfer at Porter since the long escalator ride makes it less convenient than the indoors transfer at Harvard, and such a lane would allow reliable operation from Alewife Brook Parkway to Harvard.

In Arlington, a huge opportunity for better transit was missed with the reconstruction of Mass Ave. The current streetscape, which dates from the end of streetcar service in the late 1950s, is being rebuilt for cars first, then bikes, and then transit. The current single, wide lane is being rebuilt with two lanes, to provide storage for the traffic light at Alewife. However, a single bus lane could be repurposed from one of these lanes, allowing inbound buses on Mass Ave to bypass the constant congestion at ABP and feed in to the lanes to Harvard.

Before 7:00, this inbound trip is scheduled as 10 to 12 minutes. By 8:00, it is scheduled for 22 to 24 minutes. All-door boarding, bus lanes and signal priority would reduce travel times by up to 50% in this corridor. It might not fit gold standard BRT, but it would provide the same benefits.

Mass Ave. Central to South End/Columbia. Between Harvard and Central, bus rapid transit would be difficult: the path through Harvard is narrow and congested, as is Central Square west of Central. (It is also duplicated by the Red Line below.) East of there, the street gets wider, gaining a third lane, which becomes a fourth across the Harvard Bridge and in to Boston. With the 1 and CT1 buses, this is the busiest bus route in the system—apart from the Silver Line—despite heavy traffic, frequent bunching, slow travel and overcrowding. (I’ve had full #1 buses pass me midday and midnight.) The MASCO shuttle also runs frequently at rush hours. There is huge latent demand on this corridor which could be dealt with by providing bus rapid transit facilities.

I wrote recently about putting bus lanes on the Harvard Bridge. While this bridge is narrower than most any BRT corridor, because it is devoid of stops, lanes could be installed while keeping room for a vehicle lane and bike lane in the 53 foot width. On the Cambridge side of the bridge, the road is wide enough for BRT, although other modes need to be accommodated. From Vassar to the river, a busway through MIT’s campus would certainly be possible (left turns might need to be accommodated at Vassar, but street parking is oversupplied here and could be shifted to other roads). Towards Central, one lane might work. For instance, from Lafayette Square to Central Square, buses frequently sit in a slow-moving conga line of traffic. If they could, instead, use the left-turn-to-Pearl-Street-lane and then have a signal allowing them to queue jump at Pearl Street, they could save considerable time. A much larger-scale option would be to rebuild Central Square for transit, diverting traffic off of Main Street and on to Green and Bishop Allen, allow for better bus-rail transfers, and keeping buses out of traffic. This would also improve the currently piss-poor conditions for bus riders in Central, many of whom wait for the 70 and other lines along an exposed brick wall on a narrow sidewalk.

South of the river, the street is very congested and constrained, but there are enough lanes for bus lanes, sort of. Stations are again an issue. At Beacon Street, buses could have offset stops on either side to best utilize street width. Left turns would be an issue at Beacon, Marlborough and Comm Ave, but they are low volume movements with high congestion prices: they could be eliminated. San Francisco has banned left turns on Market Street. The same could be done on Mass Ave. A slightly longer trip for a few drivers (taking a series of rights to reach their destination) would speed trips for tens of thousands of bus riders. Another option is bus and left-turn only lanes. These would require signals to trip when buses approach, allowing cars to turn left and buses to go straight, which might be more difficult to design, build and enforce.

There’s plenty of room for a bus stop between Newbury Street and Boylston street at Hynes, and along Mass Ave to Symphony, the Orange Line station and beyond, to Andrew or JFK/UMass. Most of it would require a reduction in traffic lanes. But this would be a price to pay for better transit service. Perhaps the most intriguing part of the Mass Ave corridor is that it connects many subway lines, providing excellent crosstown connections. In addition, most stops on the route are existing transit stops or major destinations (MIT, Symphony, Berklee), meaning that using longer buses with larger doors and all door boarding would speed boarding significantly.

Some of this overlaps with the ITDP’s Dudley-Longwood corridor, but the #1 bus has proven demand and would likely carry many more passengers if it was more reliable (overcrowding, bunching, traffic). The uncoordinated 1 and CT1 could be combined in to a single, more frequent route with some consolidated stops (some extraneous stops—particularly at Comm Ave—have been removed) and capacity would be increased with the same number of operation hours if buses weren’t subject to the daily whims of gridlock. Bus times currently range from 15 minutes during low-congestion times to 30 minutes—or higher—at peak hours. Again, time savings of 50% would be possible.

Arsenal-Western (70 bus). This bus, and this route, likely wouldn’t be a candidate for full “gold standard” lane separation. However, there are several congested areas of the route where bypass and queue jump lanes could dramatically improve the overall performance of the route. More coherent scheduling would help as well. For instance, a queue jump lane from the Arsenal Mall to the Arsenal-Western bridge is a site of frequent delays with plenty of room. A bus lane could also be built through the Soldiers Field Road/Western/Cambridge/River joggle. While the new Western Avenue separated bike lane is a boon to cyclists, there is probably not enough room on the roadway for a bus lane. However, floating bus stops that don’t require the bus to pull out of and then attempt to reenter traffic will help bus traffic along Western Avenue. (The neighbors are up in arms, so they must have done something right.)

Mount Auburn Street, especially where it is traversed by the 71 and 73 buses is much like Western Ave: it is too narrow for a full BRT treatment, but would benefit from queue jumps in a few locations. The most glaring one is at Fresh Pond Parkway, where there is plenty of room for an inbound bus lane to bypass the worst of the traffic, and a retimed priority signal could allow bus movements in between the various signal phases in the area. Other signal priority measures—especially at minor cross streets near Harvard Square—could help to speed these buses which, at peak times, operate every two to three minutes. To quote the linked article: “Saving 10 or 20 minutes a day for thousands of people is one of those little things that is actually a very big thing.”

North Washington Street / Charlestown Bridge. This is one of the busiest stretches of bus route in the city, carrying the 111, 92, 93, 426 and 428 buses. The 111 is a lifeline to Chelsea, running every three to four minutes at rush hours, and the combined buses have ridership of 20,000 daily—nearly as much as traffic on the bridge. The bridge is scheduled for reconstruction in the next few years, and while it’s being sold as “multimodal” it includes no transit priority provisions. There is room for transit lanes in addition to the two travel lanes, which would then allow for queue jumping at the end on to North Washington Street to Haymarket (or, perhaps, even a bus lane on that narrower stretch of street) which would save several minutes at rush hours, when these buses can sit in gridlock. Not building transit lanes here—where there is enough room and there is no parking to be lost—would be a huge missed opportunity.

The 111 could be further enhanced by bus lanes on the Tobin Bridge. Going inbound, the rightmost of the three lanes could be opened to bus traffic (and perhaps HOVs) from the entrance in Chelsea to the exit to Charlestown. Outbound is tougher, since buses have to merge from the left side of the road to the right. An entrance off of Chelsea Street might be possible, but could be quite difficult, involving short merge lanes and moved bridge supports. Still, a priority lane up Rutherford to the entrance, signal priority there and perhaps a bus lane on the entrance ram would all help buses in the outbound direction.

Rutherford Avenue north of the North Washington Bridge is another candidate for bus lanes. In theory, it parallels the Orange Line and has no current bus service. In practice, it could provide an bus lane for northbound buses to complement the southbound HOV lane on 93. It helps that the roadway is well over 100 feet wide and underutilized. As Sullivan Square is rebuilt, both local and express buses could use these lanes to access the Square, with express buses—and perhaps even intercity buses, vanpools and other HOV vehicles—could use these lanes to leapfrog some of the worst gridlock on 93, where a variety of merges cause traffic that sometimes slows to less than 10 miles per hour. Lanes could even extend along Mystic Avenue to the next exit, allowing buses to bypass even more of the worst traffic, and perhaps, for some Medford buses, to Medford Square, eliminating congestion delays on I-93 all together.

Huntington Ave. Between Brigham Circle and the Riverway, Huntington Avenue is one of the most heavily used transit corridors in the region. It is traversed by two “key” bus routes—the 39 and 66—and by the Green Line’s E branch. Delays here affect the commutes of 50,000 daily passengers, whether they’re in the vehicles or waiting for delayed buses or trains in that traffic jam. Peter Furth of Northeastern pointed out that by providing a center median and consolidated stops, these routes could increase speed and reliability dramatically by bypassing the chronic congestion on this stretch of roadway. Like many other bus lanes, it would also provide emergency access for ambulances which frequently access the hospitals there from Route 9 and further west. This would require removing a lane of parking or a travel lane from each side of the roadway, but like Mass Ave, the roadway is not constrained by width but by signals at either end, and thus is used mostly for storage, not throughput. With a bus or train traversing this roadway 50 times per hour, transit operations should come before slightly faster car travel.

Ruggles-Jackson Square-Seaver Street-Ashmont. This is the longest corridor that I’ve identified, and it dovetails well with the aforementioned Blue Hills-Washington corridor, which I argue should, in the long term, be a light rail line. This line, however, runs diagonally across it (and, for a short stretch of Blue Hill Ave, concurrent with it) from Ruggles and the LMA to Ashmont station. From Ruggles to Jackson Square through Egleston to Blue Hill Ave, there is ample room for a bus lane; in fact, Seaver Street cars ran on a separate right of way until, like many streetcars on wider streets, it was removed in the 1950s for more cars. At the time, Seaver was a major commuting street—it is less-so today. Once on Talbot Street southeast of Blue Hill Avenue, buses might run in mixed traffic, but most of the route would be in separate lanes with better stations, speeding this crosstown trip, and making a good connection with the BRT or LRT line on Blue Hill Avenue.

Washington Street, Quincy. I threw this in partly for geographic representation, but partly to show that there are some busy bus corridors in the suburbs. The 220, 221 and 222 combine for 10 trips per hour in the morning peak. There might not be the street width of political will for full BRT here, but a faster connection to the Red Line would encourage more transit use.

Union Square (Allston) to Kenmore. This stretch of Brighton Avenue, like Seaver Street once had streetcars in a median, which was removed for more traffic. There’s plenty of room for such a treatment, and I’ve argued that with signal priority, 57 buses on Commonwealth Avenue could share the reservation with the B Line, and speed through congestion on Comm Ave, especially at the BU Bridge to the Kenmore transfer. By removing bus stops along Comm Ave, it’s possible that this could actually increase street parking.

O’Brien Highway, especially during Green Line Extension construction. When the new Lechmere Station is built, the Green Line will shut down for a period of months (or even years) while the end of the viaduct is rebuilt and connected to the extension. During this time, the line will be replaced with buses running along O’Brien Highway, which devolves in to gridlock on a daily basis. (Disclosure: the EZRide Shuttle, which is managed by the Charles River TMA and for whom I work, also uses this stretch of roadway.) There are three inbound lanes and three outbound lanes west of the Craigie draw, and they are regularly stopped at rush hours, so it often takes 10 to 15 minutes to travel this half-mile corridor. Customers riding the Green Line will be subjected to this traffic if nothing is done, and with the Longfellow closed, the backups have increased.

As with many other examples, the issue is not that the road isn’t wide enough, it is the throughput of the lights at the ends of the segments. So the extra lanes act just to store cars, not to move them through. Repurposing one of the lanes for buses would mean that during the Lechmere Viaduct closure, Green Line passengers would not have trips 15 minutes longer—or more—than their regular commute. There is no parking to worry about, and even without signal priority, it would allow better transit times.

Are these corridors as sexy as “gold standard” bus rapid transit? No. But they’re more realistic: unlike the Silver Line, they don’t over-promise and under-deliver. CityLab had an article about the ITDP report (they didn’t debunk it’s half-truths) in which MBTA planner Scott Hamwey (living in reality) had the following to say:

Hamwey says shorter stretches of exclusive bus lanes can still lead to better service—“even if it doesn’t add up to a brand new, shiny, four-mile corridor.”

That’s exactly the point. Fixing Boston’s buses will not begin an end with a couple of unrealistic corridors. There are places—the North Washington Street Bridge, or all-door boarding on the 1 bus, for example—where a few minor changes would allow much better service for the whole corridor. Making major changes—long stretches of bus lanes and the removal of significant parking—is not going to happen overnight. It will require incremental changes. A bus lane and queue jump here, a prepay bus stop there, done well. These will pave the way for better buses in other corridors. The Silver Line has been a step backwards for better buses; it’s helped to convince people that Boston can’t do buses well. More expensive projects with lofty promises and cloudy outlooks will only do the same. The goal should not be infrastructure that doesn’t fit with the geography of the city, but instead to start picking the low hanging fruit, and to use existing infrastructure to leverage better buses in to the future.

The Silver Line—and the Boston BRT study—overpromise and underdeliver. They sell you the moon, and you get, well, a slightly faster bus. Think big, but think realistic. Build a better network, not just a better bus.

How to improve transit (including buses) in Boston, part I

McKayla Maroney is not
impressed with the ITDP
Boston BRT study.

Up to this point, I’ve been critiquing the ITDP’s Boston bus rapid transit report. That:

  • Boston doesn’t have wide enough streets for the kind of BRT the ITDP reports on.
  • The ITDP’s reported BRT capacity is mostly exaggerated.
  • BRT costs far more than the ITDP suggests and their report is intellectually dishonest suggesting BRT is an order of magnitude less expensive than light rail.
  • Rather than looking to Central and South American cities, Canada might be a better analog.
  • The ITDP’s agenda may be less about better transit and more about funding their studies.

I think that about sums it up. If you need a tl;dr, see the picture to the right. What this and the subsequent post will focus on is where and how Boston can improve it’s buses, and where bus rapid transit is, and is not, the best technology to use. We’ll look at street widths, capacity constraints, growth potential and system integration, several factors the ITDP glosses over in their “lines on a map” “analysis.” Here are, more specifically, the factors I’m using to analyze these routes:

  • High frequency service. Corridors or bottlenecks with frequent service, whether on one line or many, give a better “bang for your buck” than bus lanes that don’t see much use.
  • Relatively straight routes. One criticism of the Boston BRT study is that it tries to reargue the urban ring, a project which uses buses for what should be a rail project, with significant looping and zigzagging rather than trying to build a faster corridor. (Oh, and a billion dollar tunnel under the Longwood area.) Buses should go in straight lines. Time spent curving back and forth to serve destinations’ front doors or wide-enough streets is time wasted. The urban ring, and the ITDP’s reprisal, do just this.  
  • Wide-enough rights of way, especially in bottlenecks. One of the major failings of this report is that it assumes that if most of a route can be BRT-ified and a few sections are too narrow, it’s fine. But the capacity issues often fall in those narrow sections. Buses may save a bit of time with better lanes in the areas where it’s easier to put those better lanes, but still lose a lot of time in mixed traffic. (See Line, Silver.) The low hanging fruit should not be “areas where it’s easy to build facilities” but rather “areas where we can save the most time” even if you don’t get a long, shiny new busway.
  • High potential to bypass traffic. This ties in with the previous note, but if you build a two mile BRT line which parallels an uncrowded road, you’ve spent a lot of money on useless infrastructure. (An extreme example; we have few such roads in Boston.) If you build a 0.2 mile lane which moves buses out of constant gridlock and saves each several minutes, it’s a much wiser decision.
  • Build connectivity. Use new transit to enhance the utility of existing transit, whether by building crosstown connections or connecting major transit nodes to housing and employment areas.
  • Be wary of BRT for major last-mile connections. One of the failings of the Silver Line waterfront tunnel is that it is over capacity. It serves as the last mile connection between the Red Line and South Station to the waterfront. It currently runs at better two minute headways, and while it could conceivably run somewhat more frequently (say, loading multiple buses at South Station at the same time) it has a finite capacity of maybe 4500 per hour (it operates below 3000 right now). But that’s assuming balanced loading. If two busy commuter trains and two crowded Red Line trains come in to South Station at the same time, it can mean 5000 passengers arriving together. If 10% of them are traveling to the waterfront, it takes 5 packed and bunched buses to carry that many people. Or one three-car train. BRT is better for corridor systems with more balanced loads, but are hard to scale for these sorts of spikes.
  • Choose the right mode (especially to tie in to existing transit). Bus rapid transit was studied and rejected as an alternative for the Green Line Extension to Somerville and Medford. While it would have been cheaper to build in the corridor, it would have needed a destination downtown, and required bus lanes along O’Brien Highway and then in to downtown Boston. With existing Green Line infrastructure, light rail trains can run in to the central subway tunnel across the Lechmere viaduct. If a corridor has high ridership potential and can connect with existing grade-separated infrastructure with spare capacity, it’s wise to use that infrastructure. And if you’re building a long tunnel, BRT is probably not the way to go.
  • Don’t build a white elephant! If you spend $500 million on a BRT busway and the buses are over capacity and delay-prone, you’ve thrown that money away. Instead, look at investing that in a higher-capacity system with better grade separation. If you can’t build that on day one, look for lower-cost things you can accomplish, and plan for a longer-term goal of higher-capacity, better speed transit. In other words, don’t build a high cost but constrained if it’s not a good long-term choice.
With that said, let’s look at some areas that could benefit from better transit in Boston. I am going to break this in to two posts to keep it from getting too long: one examining the ITDP study corridors, and another other corridors which I think might have a better return on investment for bus improvements.

ITDP study corridors High
Frequency
Straight Street
Width
Bypass
Traffic
Connectivity Last
Mile
Right Mode White
Elephant
Silver Line to Mattapan
Hyde Park to Forest Hills (32 Bus) N/A
Harvard to Dudley via Longwood
Sullivan to Longwood

Other bus improvements High
Frequency
Straight Street
Width
Bypass
Traffic
Connectivity Last
Mile
Right
Mode
White
Elephant
Arlington to Harvard N/A
Central to South End via Mass Ave N/A
Arsenal-Western (70 Bus)
Mount Auburn St (71/73 bus)
111 Bus, N. Wash Bridge & Tobin
Northbound I-93 express buses N/A
Huntington Ave (39/66/E Line)
Ruggles-Jackson Sq-Seaver-Ashmont N/A
Washington St, Quincy N/A
Union Sq-Kenmore
O’Brien Hwy during GLX Busing

Let’s examine these in a bit more detail. In the interest of brevity (ha!) we’ll start with the ITDP’s corridors and cover the others in a subsequent post.

Silver Line, downtown to Mattapan; Washington Street and Blue Hill Avenue. This is broken in to two routes in the ITDP study, but really they combine as one. And this is an incredibly important transit corridor, linking a huge swath of lower income neighborhoods in Boston in with the downtown core. Better transit would serve these communities well, and increase transit-accessible housing in the region dramatically. This is important in curbing gentrification by dramatically increasing supply so that demand, latent and induced, can keep pace.

And this is a good candidate for bus rapid transit! South of Warren Street, Blue Hill Avenue is more than wide enough for a high-quality BRT system. North of there, there are constrained sections, but only a couple of major bottlenecks (Warren Street, just north of Dudley Square). But there are two major issues with bus rapid transit on the corridor. First, it might hit capacity, like the Orange Line in LA has. Without grade separation or passing lanes (neither of which are feasible, especially in the more heavily-used northern portion of the line) better-than-three minute headways are difficult without without bunching and load balancing issues. And with narrow roads downtown, even 60 foot buses are a stretch, longer buses would be difficult even if they were used in the US.

Second, downtown. Boston had a problem with surface transit clogging downtown 120 years ago. We solved it by putting transit underground. I’m very on the record as saying that buses are a good thing in downtown Boston. But with narrow streets and limited terminal space, downtown Boston doesn’t need more buses, especially buses which need to maintain even headways in both directions; most of the current routes are express buses which carry few people on their outbound trips; others mostly use the already-busy Haymarket busway. (I’ve written ad nauseum about the perils of scheduling a busy route without recovery time at both ends.) We certainly shouldn’t banish the buses we have, but we also should be wary of too many more.

What makes sense for this corridor is light rail. It can also deal with a few short street-running segments (although the ITDP would believe otherwise: see page 22 here where they sell BRT because it can run in mixed traffic while light rail cannot; this is patently untrue) and has higher capacity. But the major advantage is that instead of threading through narrow downtown streets, light rail can use existing infrastructure and use the subway along Tremont Street to reach Government Center and beyond. As someone commented to me, we solved the issue of running transit on downtown streets 120 years ago. Well, more like 118.

Is there capacity in the subway? West of Boylston, there is not. But from Boylston to Park Street, there are four tracks, and the inner two tracks which come in from the west are separate from the outer two tracks, with a turning loop to send cars back. So by terminating some cars at peak hour at Park Street, it would open capacity to have cars from the now-abandoned Pleasant Street incline in their slots. This might mean, at rush hour, turning B cars at Park Street. Some B Line riders might have to make a cross-platform connection to high-frequency through cars for the trip to Government Center (where they terminate today), but it’s a small price to pay for proper underground service.

And if you do that, you don’t have to try to engineer BRT on to narrow downtown streets where it doesn’t belong. In Mexico City, Line 4 is the only section on similarly narrow streets, and those streets have no vehicular access to businesses, something that would be difficult to implement as most streets in Boston require deliveries made by vehicle. Even so, that line in Mexico can only run 40 foot buses, platforms have few of the amenities associated with BRT (there’s nowhere to put them) and capacity is constrained. The report’s “technical analysis” spends a lot of time examining how to route BRT downtown by completely taking over streets, which is basically a non-starter. I’m not sure how businesses in the Centro Histórico are serviced (parallel streets, perhaps) but for most areas of downtown Boston, there needs to be at least some vehicular access to the streets, especially since the Centro Histórico is more akin to the North End than the Financial District. In any case, this is probably a nonstarter on downtown streets in Boston.

Yet the cost of rebuilding the Pleasant Street incline is certainly not inexpensive. Both downtown solution are costly, yet one only integrates with existing light rail capacity and another only with BRT. Perhaps the solution downtown, right now, is to do the minimum. Run buses on the Silver Line route, perhaps with better lane priority, but don’t try to carve out BRT through the Financial District for a high cost and low return.

And south of that point? Do everything. Build a true light rail-ready BRT corridor. In name, San Francisco is doing this on Geary, but conversion would still require the removal of concrete and the installation of electric capacity, both of which would be disruptive to transit operations for some time. If you are going to completely rebuild the street (which either mode would require) put in the disruptive infrastructure for the future. The marginal cost of setting rails in concrete is minimal (and some portions would be used by buses even if light rail trains ran as well), and if you’re already digging down, building the necessary electrical conduit system costs a lot less (substations can be added later). Then, if and when money becomes available for the Pleasant Street incline (which may dovetail with capacity constraints), you can easily convert the corridor to light rail operations without digging up your investment.

It’s also more of a commitment to the community. One of the reasons (I believe) the 28X proposal was sunk is that it was an inferior product in the neighborhood without any promise that it would ever be comparable to rail service. The current study does little to quell those fears, especially since the Silver Line is often full and the bus lanes it uses disappear where they’re needed most downtown. The 2012 study of modes that made the curious assumption that a light rail line with one-seat rides downtown would actually decrease transit use was further “proof” to community activists that the deck was stacked against rail transit. (This may be due to fewer transferring passengers, so there would be more overall journeys, but fewer unlinked trips. Of course, this is coming from an agency which defines highway capacity by the number of vehicles moved per hour, not the number of people (!) so take it with a grain of salt.) And proposals for limited-stop buses along the 28 (from the same study) still don’t promise a one-seat ride downtown, although at rush hour, considering the chance of getting a seat on the Orange Line at Ruggles, it might be a good descriptor.

Building a light-rail ready line, and actually putting some track in the ground, would be a commitment that, yes, the goal is a system which integrates with the rest of the transit network. The fact that there is spare capacity in to Park Street is icing on the cake. For much of the T, getting more capacity would require major track and signal upgrades. But the tracks at Boylston are there, and the quad-track running in to Park could allow for additional service.

Yes, this will cost somewhat more. But light rail doesn’t cost that much more than bus rapid transit, and if you don’t string wire, add substations or buy vehicles, it pulls the cost down further. Steel rails only cost so much. But you won’t be stuck in a Orange Line-esque sunk cost conundrum like in LA. There, the initial cost to build BRT plus the cost to upgrade it is nearly the cost of tearing it out and putting LRT in it’s place. It’s a sunk cost white elephant. Don’t build that here.

Neighborhood street with no parking
or bicycling facility. Also, no room for
a bus station. Yup, this makes sense.

Hyde Park to Forest Hills (32 bus). Here’s a line appropriate for BRT … if the street was wider. As it is, there would be enough room on much of the corridor for BRT, travel lanes and, well, that’s it. If there were a parallel street for, say, bike lanes, that might work, but there’s not. Hyde Park Ave is the main artery through the neighborhood, and it needs to be a complete street. The ITDP’s proposals leave enough room for one travel lane, or require one-way splits which are poor transit planning, especially if there’s a barrier (the Providence Line) in between. There is also no plan showing how to fit a bus stop in to to these narrow areas. Presumably everyone will walk a mile to wider parts of the street as empty buses roll through?

Then there are other users. There is, apparently, no need to have provisions for cyclists in Hyde Park Ave. We often talk about making streets more comfortable for cyclists, and a single travel lane shared with cars is certainly not a high level of comfort. This is also not somewhere where bike/bus lanes make sense, as buses would be slowed by passing cyclists more than they are in traffic (more on this later; in some short segments they do make sense). So to speed buses, all cyclists will either have to poach bus lanes, risk narrow travel lanes with no shoulders, or take a circuitous route. More likely, they just won’t bike. And let’s talk about parking. This is a neighborhood street. It would be great to give people a BRT option, but you can’t build a street with zero parking. Not just politically. At all.

Is what is proposed even gold standard? Not even close. By the ITDP’s own standards, I doubt this corridor would score higher than the bronze level. Would the 32 benefit from stop consolidation, proof-of-payment fare collection, better stop amenities and bus lanes on the wider part of the corridor? Certainly. Again, to make an analogy to the Twin Cities, this is exactly what they’re doing with the 84 bus on Snelling. Fewer stops, better amenities, off-board fare collection. Costs a lot less, gets most of the benefits. Perhaps those are the steps to take, not pushing through a politically unpalatable busway which will anger drivers, cyclists and pedestrians, and which won’t have much marginal benefit.

Dudley to Longwood and Harvard/Sullivan. Like the Downtown-Dudley-Mattapan corridor, this is one of the most important corridors in the city. It links several transit lines with major employment centers, both current (Longwood, Kendall, Harvard) and upcoming or potential (Sullivan, Dudley, and nearly Assembly). But there are no wide roads between them, and the roads that do exist have numerous pinchpoints and bottlenecks. Constructing bus rapid transit would be a massive undertaking (The urban ring proposal was for a mile-long tunnel below the Longwood area, perceived as “the rich get their tunnel and the poor get pollution.” If you’re going to build a tunnel, as we’ve seen with the Silver Line, don’t hamstring it with low capacity bus service.) and still would be serving as a major crosstown and last-mile route with high and imbalanced peak transfer loads, resulting in a system unable to cope with capacity.

It has two potential outcomes which the Silver Line illustrates well:

  • The SL4/SL5 scenario. A watered down BRT that will fail to deliver the promised time savings. This has happened on the SL4/5 from Dudley to Boston. It doesn’t have enough signal priority, street width, or enough priority in congested areas (where it really needs it) and is barely faster than the bus lines it replaces (the CT2 and others). It hobbles along with relatively strong ridership, but never really attracts new riders or takes pressure off the downtown subways or achieves the time savings to attract new riders who currently drive. This is the far more likely scenario. It is notable that even the ITDP’s rosy predictions for this route offer only modest time savings.
  • The SL1/SL2 scenario. There is money and political will to build a proper BRT system: full signal preemption, elevated platforms, 60 foot buses, significant grade separations, the whole nine yards. We don’t have the street width for passing lanes, so the system is run as a single line. There are quickly diminishing returns once you start running more than 20 buses per hour per direction and it’s very unlikely that you can get beyond 30 buses (one every two minutes): minor fluctuations in passenger loading will quickly cause bunching. Considering that the corridor acts as a last mile transit line for several major subway and commuter rail lines, this is all but inevitable.

    So you get a white elephant: spending a lot of money (the SL1/2 cost $625m in 2004, equivalent to $800m today, or about half the cost of the GLX, with much lower throughput) and quickly go over capacity with no way to add more. The 32 buses hourly in the Silver Line waterfront tunnel have a theoretical crush capacity of 3,200 passengers (less, actually, given the luggage racks in the SL1, and bunching and delays as the buses can’t handle influxes of transfers at South Station). This is significantly less capacity than a single branch of the Green Line (which could be more efficient with fare pre-payment and signal priority), and less than a quarter of the combined capacity of the Green Line central subway, which, with 37 trains per hour carries more than 12,000 passengers. If the Green Line ran all three-car trains, it would have six to eight times the capacity of a BRT line.

This is particularly worrisome for these corridors because they have the potential to be a major last-mile collector system between Commuter Rail lines and employment centers, driving mode shift beyond the corridor itself. If you build an urban corridor and move 1000 people from driving 5 miles to riding a bus five miles, you save 5000 vehicle miles traveled each day. If that same corridor also serves as a good last mile connection for 1000 employees who each live 20 miles away, you quintuple the congestion and air quality mitigation, not to mention getting more people on commuter rail corridors, most of which have some excess capacity. But it has to be fast.

Right now, passengers from any Southwest Corridor line (Needham, Franklin, Providence) or the Worcester line have to go downtown and transfer to the Red Line to get to Harvard or Kendall; passengers from the north have to change to the Orange or Green lines at North Station to get to the LMA, all of them crowding the downtown subway system. Worcester passengers pass within a mile of Harvard Square and Kendall, but have to travel all the way downtown and then backtrack on the Red Line through downtown to get to work. Even LMA workers have to get off at Yawkey; at least it’s a short bus ride—or walk, even—from there. Thus, many drive, because these last mile connections take a lot of time and make transit less time-competitive. 

Even if there were more frequent, direct buses, they still would sit in gridlock on the BU Bridge (for the CT2) or Mass Ave to make these connections (and the Boston BRT report points out that there’s really no room on the BU Bridge for BRT; the Harvard Bridge might be a different story). For a worker coming to Kendall from Framingham, the fastest option might be to get off the train at Yawkey and walk. But this is not a great solution, nor attainable for everyone (especially given fickle Boston weather, we’re not all #weatherproof).
What these areas need is better grade-separated, high-capacity transit service. Yes, rail lines. The first is obvious: from the proposed West Station in Allston along the Grand Junction. MassDOT proposes DMUs, although for a corridor of this length it would make more sense to put up wires and run EMUs or integrate with the existing light rail infrastructure (and, perhaps, tunnel through Cambridge, which would increase the cost, capacity, utility and land values above-ground). Commuting from west of Boston to Kendall Square is a black hole for transit: taking transit requires 25 to 30 minutes just to get back to where you started from; surveys frequently show much lower transit commute percentages from western suburbs than other areas in the region. A good connection from Cambridge could shave 30-45 minutes off the transit commute daily, and there’s plenty of capacity on the Worcester Line for those commuters. It could also catalyze development with a station in Cambridgeport, which is currently a compendium of empty lots, parking lots and low-slung buildings. 
But the real meat is in where it could be extended. First, an extension of some service to North Station would provide better service between north side commuter lines and Kendall (full disclosure: I work for the Charles River TMA, which operates the EZRide Shuttle between North Station and Kendall and is too often hamstrung by traffic delays; these opinions are my own), the Worcester line and the North Station area, but would also provide a better connection to potential development in Allston. Another branch could circle around the back of the Boston Engine Terminal and serve the third and fourth iron tracks (to the west of the Orange Line) at a major new transfer station at Sullivan before extending to Assembly or perhaps even towards Lynn along the Eastern Route. With these tracks, and the overbuilt (in anticipate on lengthening) Orange Line station, there’s plenty of room for such a facility, being a major transfer point and serving a high-potential TOD node. (See page 97 of this urban ring PDF for engineering drawings.) On the west end, service could be extended west along the Turnpike to the new “Boston Landing” New Balance station, and perhaps further as local service to upgraded Newton stations and a 128 park-and-ride, while current commuter rail service could provide an express trip, local trains serve the inner stations, with cross-platform transfers in Allston. 
Kendall has a huge head start over pretty much anywhere else in the world for biotech (and some other tech as well) and a lot of that is because it has good transportation access, as a commenter in this Universal Hub thread said a lot of the reason it’s grown is that “none of the new hires and hot recruits wanted to move to [exurban] New Jersey.” It’s well-located. But to truly cement that lead, the community needs to embrace more transit capacity, and an east-west line to complement the north-south Red Line is, I think, the ticket. A winding bus rapid transit line? Probably not.
Then there’s the Harvard to Longwood section. In the short term, bus lanes on Mass Ave may be able to help the M2 to better provide this connection. And if West Station is built, a shuttle bus from Harvard to Allston will move commuters along that leg. But in the longer run, this is also a corridor which would benefit greatly from a new, underground, high-capacity transit service. If you’re going to tunnel under the LMA (and getting high throughput through the LMA is easier said than done on the surface), it should have rails in it, which allow for higher speed and higher capacity service. Buses in tunnels need to be guided, and are still hamstrung by lower speeds and wide loading profiles. If they’re not electric or dual-mode, ventilation is more of an issue than with electric rail. And the cost of a tunnel is the digging itself; it’s no cheaper to construct a concrete road in a tunnel than steel rails.

Above: major and emerging employment centers outside of downtown (green, blue), other employment/activity nodes (pink circles), Commuter Rail stations (purple circles). On the left, current transit routes and demand, on the right, how two new rail lines would satisfy most of this demand.


The advantage of buses is that they can run on the surface outside of a tunnel, so you only have to tunnel in the most necessary locations. The disadvantage is that once you build a tunnel for buses, you lose the capacity you might get with rail, so it’s a big investment for minimal capacity. The issue with this tunnel is that the most expensive portion to tunnel is also the most necessary: if you build that portion of the tunnel, it makes sense to build logical extensions on either end.

In this case, the line extends in two obvious directions from the ends of this tunnel: towards Harvard Square and towards JFK/UMass station, where it can connect, on both ends, to the Red Line. In both cases, in fact, there are already yard leads in place for this connection. Such a loop would connect all of the South Side rail lines (and the Fitchburg Line at Porter) with Harvard Square, Allston and the LMA. It would provide much better service than any amount of bus rapid transit could for an urban ring. It would cost more to construct, but provide a much better network enhancement by connecting the Commuter Rail lines—all of which have potential to add more capacity—with major employment centers with which they are currently disconnected, and provide far speedier service than any at-grade option.

The easier Red Line connection is on the south end. North of JFK/UMass Station, there Cabot Yard leads extend, grade-separated, north towards South Station and the Red Line yard. This is basically a full two-track subway line for a mile from Columbia to Widett Circle (for what its worth, this line would basically connect all of the Olympic venues Boston2024 has proposed, as well). The next mile and a half to Ruggles could follow the Melnea Cass right of way (cut and cover: cheap, fast and it doesn’t impact parking or businesses), which was cleared for the unbuilt Inner Belt highway. Using this right of way would mean that a tunnel could be built below grade as a cut-and-cover project, far cheaper than a bored tunnel, with minimal land takings. It would also serve Boston Medical Center with a subway station.
From there, a tunnel boring machine would be needed under Longwood and Brookline. The launch box for the machine could be built in the Melnea Cass easement on one end and the Allston yards at the other; much of the cost of using such technology comes from the land needed to put the machinery in and out of the ground (see San Francisco, for example), but in this case, there is open land available. This section would provide an easy connection between the LMA and Ruggles, JFK and West Station, making the area far easier to access from the Commuter Rail system. This tunnel, with stations, would likely cost in the range of $2 to $3 billion. But it would transform the transportation options in the area far more than any other project for decades to come.
From West Station, another cut-and-cover tunnel could be built up towards Harvard Stadium and the under the river (probably a trench-and-box procedure since the river is quite shallow here) towards the JFK School. The state actually owns the pedestrian corridor through the JFK School campus to the old yard leads which still exist behind a cinderblock wall. (Don’t believe me? Check out these pictures. You could lay track in there tomorrow. More info in this foamernet thread.) Since Harvard has bi-level platforms, you could run trains on and off of the current trackage there without any crossing traffic since it’s a flying junction. Of course, this dramatically limits utility, because you would miss the Harvard station itself which is outbound of the wye.

Obviously, just a sketch, but it’s likely something
could be fit in to the existing Harvard station without
completely rebuilding it a la 1978-84.

So you’d have to do some construction in the station. But this might actually be quite easy. The main issue is that to build a full wye, you’d need to have some trains switch levels, which doesn’t exactly work. If you didn’t have a Central-Allston leg of the wye (which you don’t really need, but which you could retain for non-revenue moves), trains could operate through the existing platform and then have a switch just east of the platform end and a curved tunnel to the existing tunnels (with a less-severe curve than the current line). All of this could be built within the existing station cavern. The major change necessary would be to relocate the stairwell from “The Pit” in Harvard Square, but that could use an update after 30 years anyway, and it could potentially be built in between the bi-level tracks (see sketch). Beyond there the tracks would be left-running (think Britain) but the line could fly over itself somewhere (cut and cover in Allston, or perhaps in the bored section) to regain right-side running for the connection to JFK/UMass.


(Update: Note that this is all conceptual, and there are obviously operational and engineering hurdles to overcome. Just like squeezing a four lane busway in to a 35-foot-wide right of way.)

All of this costs a lot of money, but look at the utility that you would gain:

  • Coolidge Corner to Harvard: 8 minutes (currently 25-30)
  • Ruggles to Allston: 8 minutes (currently 15-25)
  • LMA to Harvard: 11 minutes (currently 25-30)
  • Ruggles to Harvard: 14 minutes (currently 30-35)
  • Dudley to Harvard: 16 minutes (currently 35-40)
  • Porter to LMA: 14 minutes (currently 30-35)
You basically kill off the whole hub-and-spoke notion of Boston by putting Harvard Square, Allston and the LMA the same amount of travel time from most subways and commuter lines as downtown. A trip from Natick to Harvard goes from 55 minutes to 30, a trip from Canton to Harvard from 47 to 36, and a trip from Concord to Longwood from 71 to 47. For many commuters, you’re saving nearly an hour a day in transit times, and making transit commuting time-competitive with driving, something which, in many cases, it isn’t today.
But isn’t BRT just as fast? Boston BRT claims a 42% time savings between Dudley and Harvard, going from 57 minutes to 34. This is impressive! Of course, it’s not true. The actual scheduled travel time from Dudley to Harvard ranges from 22 minutes off-peak (on the 1 bus) to 36 minutes at heavy traffic times (via several modes, the Silver Line or Orange Line to Red Line or even the 1). The 66 bus does take this long (including schedule padding coming in to Harvard), but no one would suggest that it is the fastest way to travel between Dudley and Harvard. The loop described above would go from Melnea Cass and Washington Street to Harvard in about 16 minutes, less than half what the current travel time—or promised BRT travel time—affords.
In any case, the four corridors proposed by the Boston BRT study are certainly some of the most important in Boston. And in the short term, they are certainly candidates for better buses. However, trying to install “gold standard” BRT treatments is not necessarily appropriate: in some it would mean kicking cyclists and, yes, vehicles to the curb (literally!) and in others it would mean using the wrong technology for the necessary demand. If Boston had a Bogotá-style network of wide highways—and if it wasn’t bisected by a wide river with only a few crossings—then BRT might make more sense for these heavily-traveled corridors. But given the geography of the city, this is not the case.

Do we have the money for this? We should. Obviously, we need to invest in moving the system to a state of good repair. But we also need to keep up with our “competition”: many cities are building multi-billion dollar projects to enhance their transit networks. San Francisco is building the Central Subway and rebuilding the Transbay Terminal to better connect people and jobs. LA is building several transit lines and connections to enhance transit in the city. New York is building the Second Avenue line, the 7 extension and East Side Access, among others. Chicago, having rebuilt many of its elevated lines in recent years, is seriously talking about the Brown Line flyover. Boston is installing a few third rail heaters. (The GLX project is a good one, but it is more an extension than it adds capacity to the network.)

We do need to look towards the future. And while other cities are all embracing BRT as part of the solution, they are investing in their rail systems as well. The ITDP’s agenda of buses-only is a detriment to that approach, and thus we may run in place while other cities move forward.

In the next post, I’ll discuss places where buses could benefit from elements of bus rapid transit, even though it might not fit in to the ITDP’s arbitrary standard system.