BU Central to Harvard is pretty much exactly what's proposed as an Urban Ring spur in all the official Phase II concepts, so that isn't far-fetched at all:
Take a light rail Ring where you assume that the Phase III cross-Brookline tunnel is unbuildable and that a Comm Ave. subway extension from Kenmore to BU Bridge is necessary for giving the Grand Junction a hook-in. That's pretty much what you've got here, except it's a Harvard AND Grand Junction split not an either/or. You'd have a flying junction past BU Central station separating the B from the Ring before they both ascend to the surface, and the subway portal for the Ring would pop out on the grassy hillside next to BU Bridge right by the Storrow rail bridge.
Easy at-grade junction on this hillside:
- Grand Junction ring traffic bangs a right and heads over the bridges.
- Harvard Branch traffic bangs a left onto the old Grand Junction incline under the Pike viaduct to Beacon Park.
- Wye track allows for thru running from the Grand Junction to Harvard, skipping BU Central and the Comm Ave. subway.
The trek across Allston from West to N. Harvard can be completely above-ground and grade-separated so long as Harvard's grand redev of Beacon Park preserves an unbroken transit reservation on the plazas in front of the new buildings (Harvard's own Allston Master Plan says the transit reservation is planned and sacrosanct, but there's no legal obligation keeping Harvard from changing it's mind). If so, the only thing the build entails from BU Bridge portal to N. Harvard is a trajectory that crosses the street grid at 90-degree angles to enable minimalist overpasses. e.g. Tweak the map to conform to property lines, recycle the freight spur underpass of Cambridge St. by the tolls if any of that grade separation gets preserved as post- Pike realignment residue, curve around the Ohiri Field perimeter instead of splitting the middle, etc.
The only tunneling then required is the big-deal Charles crossing. To do that you've got to go about 2 blocks further west than where your map currently draws, because the structural insertion points into Harvard Station all cling to a Brattle Sq. alignment and not JFK St. Here's how it would go:
- Start at "Harvard Biz." station @ Ohiri Field. Align the track to curve along the south border of the field and cross N. Harvard on an overpass into the Stadium parking lot. May want your Harvard Biz. above-ground station to be right here on the south end of Ohiri all along to tee up that eventual tunnel trajectory, or to relocate it across the street in the Stadium parking lot when time comes to add the tunnel.
- Curve out of the parking lot to behind the Stadium, splitting between the Stadium and O'Donnell Field. Start inclining down and hit a portal somewhere behind the 1st-base side bullpens @ O'Donnell.
- Start descending into a deep bore tunnel.
- Trace a straight line across the river from the Stadium side path to the leafy ped plaza splitting the JFK School complex. This is the trajectory you must hit for reaching Harvard Station. Keep it straight except for a slight S-curve shimmy around building foundations on Soldiers Field Rd. (you're deep enough that passing under building cornerstones won't cause problems, so don't swerve too severely).
- Incline up to cut-and-cover depth by the JFK School plaza out to the corner of Bennett & Eliot St.'s.
Final hook-up to Harvard Station is then:
- The bus tunnel...swerve cut-and-cover from plaza to Bennett Alley, merge into bus tunnel mixed-traffic. Some street-running improvements around Cambridge Common for more efficient bus/trolley loop back into the bus tunnel.
- The abandoned Red Line yard tunnel, ending at the Bennett/Eliot intersection literally under the sidewalk. Punch through wall, go to stub-end platform with egress into station lobby behind the closed MBTA ticket windows (the only available access point from the new Red station to the old Red alignment).
The abandoned tunnel is conceptually nicer for service, but hasn't been studied on engineering feasibility so we don't know if there are blockers. Therefore, must provision for the bus tunnel as the safer fallback option.
You can even, as a Phase I, grin and bear it with a short 1000 ft. street-running section across the bridge and into Brattle Sq. to buy a few years' placeholder of Heath St.-level frequencies into Cambridge while you save money for the much bigger tunneling job across the river. Maybe even limit the street-running to just crossing the bridge and then ride through JFK Park on the surface and stub-out with platforms on the plaza walkway for a temporary surface stop. Not all that pretty, but with the Allston surface construction being relatively easy and the Charles tunnel costing 2-3x more than all else combined there's no reason to wait or build it all as a monolith if you can get a temporary-terminus surface branch built ahead of time. Phase it accordingly, but do front-load that surface construction without waiting on the tunnel.
--------------------
Unfortunately there's not a lot of motivation to do some Big Cross-Cambridge Dig to link up the rest when building all the (mostly) surface projects does the job as well or better than trying to go for conceptual perfection with a tunnel down Kirkland.
- Grand Junction surface linking BU-Lechmere with run-thru to Sullivan-Airport.
- Harvard Branch linking BU-Harvard with run-thru from either Kenmore OR GC-Lechmere OR Airport-Sullivan-Lechmere.
- Porter Sq. extension of GLX Union Branch with run-thru from either GC-Lechmere or Airport-Sullivan-Lechmere.
- Red between Harvard and Porter for pinging between the Union Branch and Harvard Branch sets of service patterns...usually via the faster and less-crowded reverse-commute direction on Red.
Tally up all the options and work all the gears around flushed-up full frequencies on these network options and is there anything lost here by just grabbing all the available surface routes and working the transfers as robust as they can get? Is something
very much lost if a cross-Cambridge trunk is done in lieu of the Ring along the Grand Junction and GLX to Porter precluding a lot of transfer flexibility? Quite possibly yes.
Keep in mind what the whole value proposition of the UR is in the first place: transfers, transfers, transfers, and cranking up the network effects of efficient transfers. It is not a route geared towards linear, end-to-end, one-seat commute trips through downtown with transfers only in the CBD between one linear trunk and another. Attach GPS trackers to every B Line rider door-to-door on their whole transit trip: buses, B, other subway lines, everything. Now see what the plot looks like after a service day of tracking trips. These facts will immediately scream out:
- Regardless of what buses or other lines funnel in you're going to see a single BC-Gov't Ctr. trunk traced in ultra-fat magic marker on the GPS plot.
- The plot will escalate to maximum thickness between Harvard Ave., Kenmore, and Park St. More or less peaking around the CBD's peak density.
- Everything pulls so overwhelmingly along one linear corridor that the traces of bus transfers into the B look like little more than static noise on the plot. Even at big Kenmore bus transfer because the number of people going straight through to downtown so dwarfs the number transferring from buses.
Now do the same GPS plot for the UR mainline. It's going to look completely different.
- The UR mainline will still be the boldest thing on the map, but it won't trace out anywhere near as thick as a prototypical subway trunk. No screaming-obvious outskirts<-->CBD, "inbound-outbound", increasing<-->decreasing density like that B plot at its fattest.
- Unlike the B plot which sustains its fattest patronage closest to downtown, the thickness of the Ring mainline's plot continuously waxes and wanes between certain high-demand stop pairs. There's no one center of attention because network nodes from multi-seat trips are driving the travel patterns, not linear "inbound-outbound" trips driven by the density high-point on the map. Most people are not going to be traveling on the Ring for particularly long stretches...just between dominant transfer stops. So there will be multiple demand peaks spread across the route map.
- Because of the more muted prominence of mainline trips on a circumferential route, those transfer stops that look like static noise on the B Line plot are going to proportionately look like big starbursts on the Ring plot shooting flying sparks in all directions. Chelsea, Sullivan, Kendall, Mass Ave., Harvard stops all look like lit sparklers temporarily outshining the Ring mainline itself because these two-or-more seat transfer trips are the dominant pattern on a circumferential route, and the combinations of highish-demand 2+ seat transfers gets thrown in a blender at these stops. This contrasts with the B Line plot where straight-to-CBD one-seats are so orders-of-magnitude dominant a travel pattern that great big Kenmore bus terminal gets lost in the glare.
It's useful to visualize it like this, because you can see a little more concretely how a circumferential route behaves way different from a linear route. It's all about the network effects, and the only thing that matters when building a circumferential route is hitting those transfer points of peak network effects. There's a natural tendency when looking at a 2D map in isolation to try too hard at tidying things up into fewer pieces and more linear/run-thru lines. Because it's what we're used to with the way the current system is heavily skewed to one-seat "inbound-outbound" and CBD as center of the universe. We forget that the subway was originally built around local transfer nodes and facilitating efficient
two-or-more seat trips. We forget that post-BERy the subway got way
worse at that job over time by too much MTA/MBTA target fixation on the CBD to exclusion of local transfers. And given that familiar first instinct is always "inbound-outbound to the CBD" it's easy to forget that more linear builds are not necessarily going to relieve that. Yeah, Red-Blue is a linear build that fills a big hole. But it's an oft-repeated misnomer that the pure linear North-South Rail Link is somehow going to take a big load permanently off Red and Orange. Quite the opposite: it's going to create 100,000+ all-new suburban transit riders and by sheer overwhelming volume throw 10x as many all-new riders on the NS, SS, and BBY subway platforms than it ever takes off.
The only way to meaningfully address this is by re-emphasizing some of that BERY-era transfer flexibility at local/non-CBD transfer nodes. Make
two-or-more seat trips nearly as fast and frequent as the one-seaters by roping in more of those outside-CBD transfers that fell through the cracks Postwar. Complement the modern CBD-anchored trunks with circumferential service patterns that behave more like that GPS plot exercise with the "starburst" effect away from the CBD...not the "fat magic marker" effect of plowing another trunk right through the CBD. It is less about what you build in steel and concrete than about frequency and variety of service patterns spreading the field.
This Kirkland St. subway may look like an elegant map-based solution to it all, but it's not a UR replacement. It's attempting to be another flavor of conventional "inbound-outbound" one-seater that tries to do its job on a single killshot of a primary service pattern (even if there is wiggle room for some alt patterns), behaving little different than a CBD trunk with a singular "inbound-outbound" density peak. Only because there is no CBD-like singular density peak on a line built almost entirely radial, ridership's not going to behave as efficiently on a GPS plot as the line--and its ops plan--look on the map.
By shooting too hard for conceptual integrity on its routing this ends up leaving too many "starburst" transfer stops on the cutting room floor and voluntarily forsakes too much in the way of
two-seat ride network effects. A unified Kirkland routing is not going to pull riders like the as-designed Ring route if Mass Ave./MIT and Kendall are substituted by the crossing of the just the 83/86 in Ward Two. Mass Ave./MIT, Kendall, Union, and Porter are all of relatively even heft as transfer destinations in/around Cambridge. As deputy alpha dog to Harvard none of those transfers are individually head-and-shoulders above the rest because the multi-seat trips served are so diverse. So something gets lost in translation if zeal for mapmaking integrity starts trimming too many destinations that behave like "starbursts" on that GPS plot.
It ends up more valuable to de-centralize away from a single route and start spreading more frequencies and service patterns around the map so all those Top 5 transfers in Cambridge get served on
some trolley's destination sign...even if basic Cantabrigian geography dictates they all can't feasibly be served on the
same trolley's schedule. It actually makes for a better-performing network to forget about games of picking winners and losers amongst transfer stops and simply work the scheduling variety for all it's worth. Seem counterintuitive? Well...think back to the GPS plot. The defining trait of the circumferential line are these "starburst" patterns where few
two-seat trip combos at a transfer are ever going to stand out head-and-shoulders above the other
two-seat trip combo in demand. And the defining difference from a conventional trunkline build is that the dominance of
one-seat trips is so very much muted by contrast. Not counterintuitive at all, because you can see the demand's behavior with your own eyes on the GPS plot. So shouldn't that defining behavior inform how hard one squeezes the map for sake of one single routing vs. spreading the map for sake of schedules to all the major transfers?
Eyes didn't deceive the official UR planners when they traced out that northern route with its separate Harvard spur. Spreading the field with high frequencies and schedule flexibility to all the major "starburst" transfer stops nets the biggest overall payoff for what a circumferential route is supposed to do.
