So, several months later, I realized that I never posted my presentation from this year’s TransitCon. The talk’s about the (false dichotomy of) improving rail in the US incrementally or in big leaps, and more precisely how the risk of overspending and underdelivering that plagues our project delivery accrues to incremental projects as much as huge ones. You can find the slides here.
And when I went to post that video, I realized I never posted last year’s, either! That talk was about railroad electrification. You can find the slides here.
I should emphasize that in both cases I was speaking entirely in a personal capacity, and not representing any employer past, present or future. If you have questions or want copies of the slides, you can reach me on Twitter or through the Contact Me function of this blog. And of course, huge thanks go out to Hayden and the TransitCon team for putting this wonderful event on!
Anyone who follows me on Twitter surely saw that 10 days ago (Thursday night the 21st into Friday the 22nd) G and I rode the Lake Shore Limited (hereafter abbreviated LSL) from Albany to Chicago. When I was an undergrad I took the LSL between Chicago and Penn Station in coach several times; with a massive bag allowance (assuming a few hands willing to help at either end), it’s a great way to move a ton of stuff between home and college. This time, heading out to Chicago for my little brother’s high school graduation, we were lucky enough to be able to book a Viewliner roomette for around $360 for both of us, which is a) well cheaper than flying two people from Albany to Chicago and b) a great price for an overnight sleeper. The schedule, leaving Albany at 7 PM and getting into Union Station (theoretically) at 9:45, works nicely as well. Being me (and having engaged in several educational Twitter discussions about saving Amtrak’s long-distance service in recent weeks), I decided to take some notes on the experience and pass them on in blog form. I’m only now getting the chance to write this up, so forgive that please. Without further ado, Sandy’s notes and lessons from the Lake Shore Limited:
Unbundle the Freaking Food Service Already
It’s well-known (particularly, it seems, to Congressional Republicans) that Amtrak loses money hand-over-fist on food service on its trains. [Quick aside for those who might never have been on one: Amtrak’s long-distance trains (roughly over 700 miles, according to the Congressional definition) usually have both a full-service sitdown dining car and a more informal cafe/lounge car. Meals in the dining car come bundled with the sleeper fare, but coach passengers have to pay per dish if they want the food.] The railroad has cycled through phases of trying to cut costs and trying to make the meals luxurious enough to attract people to the dining car, with the result that the food is both pretty mediocre and very, very expensive. All long-distance trains now serve the same menu, so if you (like the nice older couple from Northern California who sat across from us at dinner) are taking a multi-day cross-country itinerary, the dishes are going to get really old, really fast.
As sleeper passengers, we got two meals in the dining car (except not really, see below). Dinner–served in a vintage-1950s dining car with low lighting and a dingy, Spartan, almost Soviet (I see you, Rep. Mica) feel–was the very definition of “meh.” It featured a green salad with a small amount of iceberg lettuce, and the vegetarian entree was an incredibly insubstantial “trio of stuffed pasta shells” that sells to non-sleeper passengers for a whopping $15.75. To be fair, the meat options did look better, and the best part of dinner was the upside-down strawberry cheesecake I had for dessert. Though the menu explicitly promises that several of the lunch options are also available for dinner, we were told that was no way, no how a possibility, and the waiter, a generally nice guy, refused to account for the kitchen’s refusal to serve the published menu. Meals are served on throwaway plastic plates and cups (though the silverware is real), which undermines Amtrak’s professed image as the “green” way to travel by generating an enormous volume of waste, as well as the nostalgia for the glamorous age of railroad china and dining car meals that Amtrak relies upon to sell its bundled food service. And remember, despite all the cost-cutting and enormously high prices that sleeper passengers have no choice but to (literally) eat, Amtrak is still losing money on food service!
And we didn’t even get breakfast on the train, because the crew shuts down the dining car at 9 Eastern time, a full 1:45 out of the scheduled time into Chicago, and typically even further. Not that that was communicated clearly to us; our attendant told us that it would “be best” if we got there at 8:00, but didn’t specify further. The train switches to Central Time at Waterloo, Indiana:
So we showed up to the diner at 8:00 AM Central Time west of Waterloo, just as it was being shut down–because apparently, the diner functions on Eastern Time for the entire ride. Bad communication and customer-unfriendly practices (do they REALLY need almost two hours to clean out the dining car?) for the win! (loss) Worse, the microwave-centric cafe car shuts down at the same time. The dining car is willing to serve a continental breakfast after 9, but these are unforced errors that just make the rider experience more hostile for no good reason.
So what’s the solution? As with any complex problem, solving the death spiral of food service on long-distance trains is clearly going to be a multifaceted project. The first step is almost certainly to unbundle food service from sleeper fares, as Amtrak is planning to trial on the Silver Star starting this summer. The trial will involve eliminating the train’s full-size diner and restricting food service to the cafe car. Many riders–particularly romantics and the older “land cruise” set who Amtrak seems to target–will surely mourn for the lost elegance of the dining car, but at this point there’s clearly little elegance left. Unbundling food service is expected to reduce sleeper fares by 25-28%–a not at all insignificant savings that should attract plenty of demand, certainly enough to make up for the loss of people who really want bundled sit-down meals (though demand isn’t really Amtrak’s problem when it comes to sleepers). Unbundling is just the first step though; ultimately Amtrak is going to have to open up the provisioning of food to the digital age.The cafe car on our Lakeshore was prominently posted with signs prohibiting non-Amtrak-provisioned food, which is just an obnoxious practice that should stop. Amtrak is also experimenting with at-seat food service, a common amenity in Europe. I also really like an idea that Alon Levy has brought up on Twitter:
@sandypsj Business proposal: use an app to match train riders to restaurants and cafes at stops with long dwells.
Building an app like that would allow Amtrak to replicate the functionality of the 19th-century Harvey Houses, getting riders fresh, variable meals without the bother and massive expense of actually preparing them on the train. Of course, it would require a level of digital literacy far beyond that currently sported on Amtrak trains, where diner attendants still require sleeper riders to fill out a complex paper form rather than using an iPhone to scan ticket barcodes like the conductors. Amtrak’s current food service paradigm may very well be in an irreversible death spiral. But there’s plenty of hope for the future, with some creative thinking.
Amtrak frequently, and generally justifiably, complains that the freight railroads and their dispatching are responsible for the chronic lateness of the long-distance trains. If our experience on the LSL was at all indicative, though, there is plenty of room for improvement in Amtrak’s own timekeeping practices. Station stops on the LSL can best be described as “leisurely,” and the crew often seem to have weird concerns–while trying to watch another train pull in on the neighboring track in Albany, I was told to stay well away from that entire side of the (wide) platform by an attendant who told me “You’re freaking the engineer out! He doesn’t know whether you might be suicidal!” For the record, I wasn’t even on the track-adjacent warning strip, and if the nerves of Amtrak engineers are that jumpy, let’s please make sure none of them get hired as New York City subway drivers any time soon!
Anyhow, about those leisurely stops, I timed a few. With the train leaving Albany already 27 minutes late, I figured there would be some attempt to make up time along the way. Not exactly:
Utica: 10-12 minutes
Syracuse: ~15 minutes (and Syracuse has high-level platforms! what could possibly take that long?)
Rochester: 17 minutes
Toledo (where I woke up the next morning): scheduled 30ish minute break, not at all shortened despite running an hour behind at that point.
It’s entirely possible that long stops at these stations were necessary to maintain the LSL’s slot in the parade of freight trains with which it shares tracks. It is, however, more likely that the long stops are because of the LSL crew’s bizarre (and against Amtrak standards) sometimes practice of checking every ticket on the platform, rather than on the train (scroll down to the comments for discussion). What’s clear is that there is plenty of room for improvement in crew attitudes about stop timeliness.
Cost Savings? Halve the sleeping car attendants.
Not literally, of course, but currently Amtrak assigns one attendant for every sleeper. In the roomettes at least, the beds really aren’t that hard to figure out, and we were able to get them down with no help and no trouble. Surely, some riders will need physical or organizational assistance, but it hardly seems necessary to have one attendant per sleeper. Cutting down to one attendant for every two sleepers or 2 for every 3 would not only cut costs but allow the sale of one more room for revenue (alternatively, it could be converted into a full-scale bathroom, which the Viewliners lack). There would be gaps to fill, particularly when it comes to things like signing up for dinner seatings, but the impact of the staff cuts could totally be mitigated by the fact that
Digital technology has the potential to transform the long-distance train experience
I’m not just talking about Amtrak’s ridiculously slow WiFi, which the LD trains don’t have but really should. My attitude about Amtrak’s WiFi problems is that of Bruce Springsteen’s character in “Rosalita”–“Someday we’ll look back and it will all seem funny.”
As I mentioned above, Amtrak conductors have recently started carrying iPhones that they use to check tickets. It would be truly transformational for the sleeper experience at least if Amtrak could get a touch screen interface into each room, which could take food orders, make seating reservations (if, indeed, the sit-down diner persists), call an attendant when necessary, and play videos and TV shows like those available on airplanes. WiFi should be the first priority, obviously, but getting a real, easy-to-use digital interface into sleeper rooms (and really, onto seatbacks in coach too) should be a major priority for the long-distance fleet, and might allow significant crew cost-cutting.
Managing Demand and Different Markets
Despite all of the problems I’ve listed here, our experience on the LSL was generally very positive. The roomette experience is cramped but manageable for two people, even when one of them is a jumbo like me. We were 57 minutes late into Chicago Union Station, which for a train that had 30% on-time performance in the month of April is not that bad. The food wasn’t worth either the extra expense or the 3-car hike to the dining car, but we brought our own, so that wasn’t too much of an issue. The staff were generally friendly and kind, which isn’t always the case.
And–importantly–our train was nearly 100% sold out. When I checked the day before we left there were zero coach seats available and only one bedroom. It’s important to understand that for all of its warts Amtrak’s problem is generally NOT a lack of demand for its services. And that applies to the long-distance trains just as much as the corridor services. Not to be cliched, but the challenge is transforming a pattern of service and operation that dates back to the 19th century to a digital age that offers both danger and promise.
This trip was my first opportunity to sample Amtrak’s sleeper offerings since my 16th birthday present, a trip on the Coast Starlight from Salem, OR to Los Angeles, and it was clear that the sleeper attendants and staff weren’t used to dealing with twentysomething travelers who could fend for themselves. And that’s a problem. The land-cruise market of older Americans who are afraid of or can’t fly, and who remember the golden age of passenger rail, isn’t all that big, and is arguably (because of the last condition) shrinking, but Amtrak’s sleeper services remain entirely focused on them, from high staffing levels to mandatory meals.
That doesn’t have to be the case. Sleepers will always be expensive, but Amtrak should be selling them as a budget option competitive with flying from midpoint cities. Take the case of the LSL. Flying between New York City and Chicago is fairly cheap and generally will remain that way. But there are plenty of smaller cities in between with medium-sized airports, and in the current hub-and-spoke configuration of the airline network, it can be very expensive to fly into and out of those places (believe me, we flew back to Albany). Taking coach from Albany or Rochester to Chicago, or from Toledo or Erie to New York City, might suck too much to be worth consideration–but a sleeper compartment might very well provide a happy medium between the expense of flying and the pain of an Amfleet or Greyhound coach seat.
Of course, as currently configured sleepers are a luxury experience that comes only at considerable expense–we got very lucky to book at the price we did. But that’s not necessarily inherent in the sleeper experience. Sleepers will always be more expensive than coach, simple because you can fit a lot fewer people in them. But by reducing staffing levels to a budget-friendly level, streamlining the experience with digital technology, and marketing the train as a budget-friendly, yet comfortable, experience, I think Amtrak could capture a large slice of the college student and young adult market. Remember, indications are that Amtrak might be able to slash sleeper prices by as much as 25% just by allowing people to bring their own food on board. And, assuming your stop has checked-baggage service, you can bring a hell of a lot more stuff on a train than a plane.
Obviously as a fairly tech-savvy, and train-inclined, twentysomething I have a self-centered bias in what I think will work for Amtrak. But the way they’re doing things right now isn’t working for much of anyone, including the political overlords who hold Amtrak’s fate in their hands. Why not try something fresh and new?
As part of my research project, I’ve been doing some looking into Boston’s North-South Rail Link concept–the idea of a tunnel linking, finally, North Station and South Station. NSRL is a complicated project whose merits are certainly up for debate (though I tend towards thinking it’s a good idea), but that doesn’t appear to be on the horizon anytime soon. But did you know that there was a point at which Boston could have had the tunnel built at minimal cost to taxpayers?
A March 8th, 1910 headline in the Boston Globe blares :
The gist of the accompanying article was that Timothy E. Byrnes, Vice President of the New York, New Haven, and Hartford Railroad, had conveyed to the state’s legislative committee on metropolitan affairs that the New Haven and the Boston & Maine–then united under the ownership of J.P. Morgan protege Charles Mellen–would be perfectly willing to pay most of the costs of a tunnel between Boston’s two stub-end stations, if the city would let them. The catch? They wanted the city to create a “boulevard” between the two stations, presumably cutting a broad swath through downtown Boston to enable cheap cut-and-cover tunneling. The railroads estimated the costs to the city at $10 million–approximately $240 million today–of which they were offering to cover one-fifth, plus the costs of the tunnel itself, and of electrification of the Boston suburban rail network. Byrnes claimed the tunnel–and accompanying electrification–could be finished within three years, hauling both passengers and freight between northern and southern New England.
Surely, $8 million was a lot of money in 1910. But with the railroads offering to cover 2/3 of the cost of a tunnel, how did a project that would have proved so advantageous to the city that the idea is still being brought up a century later die?
Just four days after Byrnes–who had been on his way out of the hearing chamber at the end of the day when asked to speak–dropped his bombshell on the committee, the city responded. The Globe headline on March 12th read:
Babson was the city’s Corporation Counsel, who made sure to tell the committee that “under present conditions the tunnel would not be worth to the city anything like an outlay of $8,000,000” and that “the city should not own the tunnel because it does not own the railroads at each end.” A legislator suggested that the railroads might lease the tunnel from the city, paying 5.5-6% interest and with ownership reverting to the city after 40 years, but neither Babson nor the railroads were amenable to that suggestion.
The tunnel idea did not die immediately–according to one article I have found, it was still kicking around two years later, as part of contentious negotiations between Mellen and the Massachusetts legislature over the fact that his control of both the B&M and the NYNH&H gave him a near-monopoly on rail traffic in New England. The idea of joint government-railroad control of the tunnel, based on a lease with relatively steep interest, had resurfaced, and government investment in the tunnel was on the table in return for Massachusetts allowing Mellen to exercise full control of the B&M (which the legislature had previously intervened to stop). But the Mellen monopoly was, by 1912, in its last days; besieged by legislative action, lawsuits, and public pressure, the New Haven would soon divest itself of the B&M, though not before 21 of its directors (though not Mellen) were indicted for conspiracy to monopoly. I haven’t delved deep enough to know exactly when the tunnel idea last saw the light, but it seems to have died a quiet death.
From a public policy perspective, it’s hard to look back at this episode and resist the impulse to punch a wall. The short-sightedness of the public officials who could have allowed the tunnel to be built–and to high standards, with four tracks and electrification of the Boston suburban network!–is remarkable. But 1910-12 was a very different time in America, a time when railroads were very much still an immediate enemy in public life and a chief target of progressive reformers. And indeed, the 1910 tunnel proposal was very much a product of monopoly–it was integral to Mellen’s vision of a unified NYNH&H and B&M (which, according to the 1912 article, would have been re-branded as “New England Lines”), and the public benefits, while substantial, were entirely secondary. And as quick as we are to condemn brutal transportation-oriented urban renewal in Boston and other cities, can we really look back and say that the idea of clearing a boulevard between North and South Stations was moral?
Though we know that constructing the tunnel would have had horrible outcomes for the residents in its way, and it might have strengthened the Mellen monopoly, it’s hard to feel any other way than that Boston’s objection to the project was short-sighted. A full century later, the proposal to build a similar project is still percolating, and it’s hard not to wonder–if there is still an argument to be made for a tunnel, now that the region has sprawled and developed away from the rail-oriented suburban mini-cores that predominated then, how different might the suburban development of the Boston region have looked if city officials had been more willing to work with the railroads 105 years ago?
The utility or lack thereof of those two stations is up for debate, but I want to highlight another aspect of their addition to the New Haven Line. With the other two recent infill stations, Fairfield Metro Center and West Haven, there are now 21 stations (including the endpoints) in the 46 route-miles between Port Chester and New Haven, or on average one station every 2.19 miles. With the additions of Barnum and Orange, there will be 23, or exactly one station for every two miles. Though few trains make all local stops, that stop spacing is rapidly dropping toward that of a spaced-out urban rapid transit line, rather than a “commuter rail” system. And that, I think, offers a great opportunity to transform the New Haven Line into something unique in the US, an interurban regional rail system with a focus on moving people not just to and from New York City, but within Connecticut as well.
Here’s the thing about the Connecticut Shoreline region: while its economy and culture are intricately and inextricably linked to New York City, it is not a pure bedroom region. The region’s older cities–New Haven, Bridgeport, and Stamford–all have an economic identity and gravity of their own (though Bridgeport’s can be kinda hard to figure out), and Stamford has evolved into something of an Edge City, attracting corporations fleeing the costs of Manhattan and pulling in commuters from its own hinterlands. In fact, only a relatively small proportion of people living in the Shoreline region commute to New York City. 61.60% of employees in New Haven County work in the same county; Fairfield County takes second place with 16.10%, and only 1.10% of New Haven County employees, or 4,150, work in Manhattan; the other NYC boroughs and Westchester add about 3,000 people. Fairfield County follows the same pattern; 67.7% of employed residents work within the county, and another 9.2% work in New Haven County, with only 6.2%, or 24,086, work in Manhattan. 4.5% (17,514) work just across the border in Westchester County, and the other NYC boroughs have 0.4% each, or around 4,500 workers total. Full numbers (from Census LEHD data) are included here:
So, the Shoreline towns share closer economic links to each other than they do to New York City, but Metro-North’s rail operations are, somewhat inevitably, still quite NYC-centric. This is to some extent justifiable; people commuting to Manhattan from Connecticut are obviously way more likely to travel by transit than people commuting within the highly suburbanized Shoreline region. Yet, the Shoreline does have a core of transit-friendly, somewhat dense cities and towns linked along the New Haven Line, and it’s not just Stamford, Bridgeport and New Haven; Greenwich, Norwalk, Darien, Fairfield, Stratford, and Milford all have downtowns that are close to the train station and walkable or potentially walkable. The region also suffers from an insufficient transit system, aside from Metro-North (or including it, depending on who you ask!). Jobs, especially those in the service sector, are heavily suburbanized, making job access by transit for disadvantaged populations especially difficult. Given the close intra-regional economic links, and the close station spacing along the New Haven Line, using the trackage for frequent transit-like service makes a hell of a lot of sense.
Of course, I’m far from the first person to advocate this approach. Just to highlight one take, Alon Levy included the New Haven Line in his series on a regional rail system for New York on The Transport Politic. Rather than preserving an NYC-centric approach, though, I want to focus on the New Haven Line’s potential for intra-Connecticut journeys. The true difficulties of such a transition lie in operating practices and labor costs; the only capital investment I truly see as necessary is the replacement of the previous fourth track between Devon (the junction where the Waterbury Branch joins the main) and Woodmont interlocking, the only stretch that doesn’t currently have all four tracks active and electrified. That would also require the replacement of the NYC-bound platform at Milford, which sits on the former fourth trackway, but that platform is short anyhow and should probably be replaced. I’d also argue that, long-term, the New Haven maintenance facilities should be shifted to the old Cedar Hill freight yard north of New Haven, allowing all trains to run through both New Haven Union Station and the downtown State Street Station rather than turning, and opening the current land up for development.
RPA/Getting Back on Track Report
In any case, running trains every 15 or 20 minutes shouldn’t be a problem even given current infrastructure; trains already run that frequently during rush hour. Though the half-hourly trains between New Haven and NYC are a start, I’d like to see those trains become what Metro-North calls “super expresses,” stopping only at Bridgeport, South Norwalk, and Stamford, with another 2 trains every hour running local between New Haven and Stamford and between Stamford and Grand Central. That would give the major stations service every 15 minutes all day, and considerably shorten running times between the major cities. Obviously, that’s just a start; I see no reason that once the four-track gap is closed and the signaling improved (and perhaps the power system improved some) the line shouldn’t be able to handle four locals and two expresses per hour.
There is one more element to my plan for improving transit along the Shoreline. Here in Albany, the local transit agency operates a successful limited-stop bus service, BusPlus, linking Albany and Schenectady along Route 5, the region’s major commercial axis. The buses, soon to be running every 12 minutes during weekday hours, have improved ridership in the corridor, cover the 15.5 or so miles in around an hour, carrying about 4,000 riders per day.
CDTA BusPlus (Route 905)
The concept of a limited-stop bus service linking close-together major cities along a major suburban arterial would, I think, be a perfect approach for transit supplementary to the New Haven Line along the Shoreline. The Shoreline even has its own counterpart to Route 5, the famous Boston Post Road (US 1). A tertiary through-route (since it parallels both I-95 and the Merritt Parkway) along the Shoreline, the Post Road has instead taken on an identity as the region’s main commercial drag. It’s a (fairly depressing) 50-mile-long cluster of mini-malls, full-scale malls, big-box stores, and the other stuff that accumulates along a suburban commercial arterial, but it does hit all of the region’s major towns, and intersects with the New Haven Line at multiple points.
CDTA’s plan for an extended BusPlus network (Red Line currently in operation)
Obviously the Shoreline is too long to be served by one bus service along the Post Road (and such a service would be redundant with the rail service anyhow), but I’d argue that splitting the stretch between Port Chester and New Haven into three segments makes sense. These links exist, to some extent, already, though they’re quite fractured; CTTransit operates (different) infrequent buses in both directions out of Stamford, Norwalk Transit operates a relatively frequent service (somewhat randomly) through Bridgeport to Milford, and CTTransit picks up again there. I’d like to see these services upgraded to an Arterial Rapid Transit standard, with signal preemption, turn and intersection pockets, nice stations, and fare integration with Metro-North. Split into three sections, each service wouldn’t be much longer than the BusPlus service, and though Post Road traffic is heavier than anything in the Capital District, the buses should be able to stay on time with decent scheduling and some signal help. The point of the buses wouldn’t be to carry people between the cities, but to distribute riders to local destinations from train stations, and to improve access to jobs along the Post Road from the urban cores. In essence, an ART system along the Post Road would be the local to the New Haven Line’s express service. Here’s a map:
The exact routings and boundaries of the bus services are only approximate. You could make the case that stopping a (long) block away from the train station in Stamford, rather than going right into it is good enough. Norwalk is also an awkward situation; should buses terminating there serve both the South Norwalk train station and downtown Norwalk? Nor have I accounted for the presence of trains coming off of the branches to New Canaan, Danbury, and Waterbury; and I’ve included a speculative re-routing of trains through downtown New Haven to turn at Cedar Hill. There are other infrastructure aspects of the line that need significant attention, especially the numerous movable bridges. But the overall structure, I think, is solid. The Shore Line Metro would be a relatively low-investment, non-capital-intensive way to create regionwide non-automotive mobility. It would involve service and fare integration between agencies, regional planning cooperation, and various other things American planning tends not to be too good at, but the payoff could be tremendous.
Park-and-Rides are one of the most controversial topics in planning and transit circles. Some contend that such facilities encourage unnecessary car use, while others believe that in the right circumstances they can reduce car usage. One example of a notoriously unsuccessful Park-and-Ride facility that came up on Twitter last night is the massive garage at the extreme southern end of MBTA’s Providence/Stoughton line, Wickford Junction, Rhode Island. Opened in 2012 at a cost of $44 million, the station includes a massive, 1100-space garage that is supposedly aesthetically modeled (on the outside, presumably) on the historic Lafayette Mills building nearby. Suburban American kitsch is the best kitsch.
Via RIDOT on Flickr, here’s the garage as completed….
The payoff for all of that investment? A massive, staggering number of riders–a full 159 inbound boardings per day in 2013, as you can see in the handy-dandy ridership map provided by the MBTA in their 2014 Blue Book.
There are few words for the amount of fail that the Wickford Junction project represents. To (perhaps) justify the investment, the station would need to attract nearly seven times the number of boardings it currently sees every day. And that’s to say nothing of the in track, trains, and crew costs that were necessary to extend commuter rail operations that far south in the first place. The overly rosy projections for Wickford Junction ridership date back as far as the 2003 Environmental Impact Study RIDOT conducted for the commuter rail extension, which contains the following helpful predictions:
Given the dire situation in Wickford Junction, then, I was intrigued by the situation along the rest of the Providence/Stoughton Line (and by the system as a whole, but hey, let’s start with one line, that’s easier to get a handle on). Here’s a spreadsheet with ridership numbers and parking capacities at all of the stations between Wickford Junction and Hyde Park (I decided it would be silly to look at Forest Hills and Ruggles, where things are complicated by the presence of the Orange Line). Ridership data is from the map above, parking data is from the MBTA website.
Since the sheet doesn’t display well on WordPress, full link is here. In graphic format, the patterns look like this:
Thanks to my partner G for Excel help!
A few things jump out.
1. The Rhode Island stations south of Providence are obviously a hot mess. If money had to be spent on extending MBTA service down there, the way to do it wasn’t to build fancy stations and then only run a few trains per day. Frequency matters. That should have been obvious from the beginning, but it’s exceptionally clear now.
2. Providence is a nice example of an urban station that doesn’t feel the need to provide huge amounts of parking to attract riders. New Haven could learn a lesson.
3. At the highest-ridership suburban stations on the line (South Attleboro, Attleboro, Mansfield, and Sharon) parking capacity is only about half of boardings, suggesting that many people do indeed walk or get dropped off (these station don’t have much in the way of feeder bus service–Sharon I know for sure has zero).
4. Stoughton and the Canton stations are an interesting case. Parking capacity doesn’t come close to matching the number of boardings–but according to the MBTA’s data, each station has 40% or more of its spaces available on a given day. Stoughton and Canton are relatively walkable despite a dearth of feeder bus service; presumably, many people walk or are dropped off at these stations, and others park for free on town streets to avoid MBTA parking charges.
5. Despite its entirely car-dependent location, Route 128’s parking remains underutilized by several hundred spaces, even taking combined MBTA and Amtrak ridership into account.
A few lessons seem apparent. First, MBTA and the agencies it works with have a problem with overestimation of parking demand–and in that, they’re certainly not alone among American commuter rail operators. Not every lot along the Providence/Stoughton line is underused, but the more monumental ones (Wickford Junction, T.F. Green, Route 128) certainly are. Operators and planners should let the demand for parking come to them, rather than trying to anticipate how it will develop. By all means, reserve space for a garage if necessary, but don’t built it until demand makes itself known. And for god’s sake, tell the locals to charge for parking too, or they’ll take away all of your paying parkers! And remember, even at the best-utilized park-and-ride stations on this line, it certainly appears that around half of riders don’t park to ride the train.
Second (and this may seem obvious, but MBTA and friends don’t seem to have grasped it) the quality of the built environment around the station matters. One of the CityLab articles I linked to at the top argued that distance from the city center should be the determining factor in the decision to build a Park and Ride facility; I’d substitute characteristics of the local built environment. Providence/Stoughton stations that are embedded in relatively walkable areas–Providence, Stoughton, Canton Center, Canton Junction, Hyde Park–exhibit very weak parking demand. Old New England towns don’t exhibit a linear progression from urban to suburban, and planners should pay attention to that. I don’t have much of a problem with Park and Rides in very suburbanized areas, but we shouldn’t be expecting people to drive half a mile from home to the station if they can walk it.
A similar experiment, is now being attempted on commuter lines serving Philadelphia–so far with a similar result. Philadelphia appropriated $160,000 on the promise of the Pennsylvania and the Reading to step up commuter and off-hour service to the northern residential suburbs. The Philadelphia, city fathers explained that they did not regard the $160,000 as an actual subsidy, but merely as an underwriting of increased operational costs, to see if improved rail service would take some of the clutter of cars off city highways. If this could be done, the pressure for ever larger highways, ever more bridges and tunnels, ever increasing parking facilities would be eased, and a subsidy to the railroads, even from the taxpayers’ standpoint, would justify itself.
The New York Herald Tribune reported last month on the Philadelphia experiment at the end of the first month of a six-month test period. The Pennsylvania had increased the number of trains on its Chestnut Hill run by 33 per cent to thirty-six daily; service was stepped up to every fifteen minutes in rush hours, every half-hour in off-hours. The Reading boosted the number of its trains from thirty-three to thirty six daily, Saturday service on both lines was almost doubled, and cheaper fares were tied in with bus-line transfers. The result: In the fourth week of operation, the Pennsylvania carried 4,133 more passengers than it had in the test week of October 6, before the plan went into effect, a gain of 14.8 per cent; and the Reading picked up 2,422 passengers in the same week, an improvement of 7.6 per cent over its test week in May. For the entire four weeks, the Pennsylvania gained 11,128 additional riders; the Reading 7,099. The effect on city traffic already was observable; 600 fewer automobiles a day were coming into the city from the suburbs.
Philadelphia hopefully assessed the advantages of the plan this way: cheaper fares mean a saving, for the individual commuter, of 90 cents a day over automobile operation (including parking fees, insurance and fuel costs), or a total of $100 in the six months of the test period. This saving to the individual driver is projected into a much greater saving to the city. It means, Philadelphia estimates, that about $81 million annually can be saved on the cost of maintaining existing roads and providing police protection. And this is apart from the merry-go-round cost of building ever more and wider highways.
This was a truly different world for transit. SEPTA wouldn’t be formed until 1963, and it wouldn’t take over any responsibility for the commuter lines until 1966, or direct operation of them until 1983. And yet, the truth that frequency translates to ridership is apparent in this report. Indeed, the actions taken on the Pennsylvania’s Chestnut Hill West Line (the two Chestnut Hill lines run very close and parallel to each other, entirely within the boundaries of the city of Philadelphia), mirror closely the idealized set of recommendations that Jarrett Walker or any frequency-minded planner would make: clockface schedules, decent off-hours service, reduced fares, and schedule and fare coordination with buses. There are, quite frankly, any number of transit agencies that can’t get their act together to make these things work in 2014. Perhaps we don’t give the people running our public transit systems during their decline phase in the 1950s enough credit?
Note: The numbers reported work out to a daily 1958 ridership of a little under 4,000 on Chestnut Hill West and about 4,400 on Chestnut Hill East before the experiment. Today, both lines carry about 5,500 riders daily, despite service that maxes out at half-hourly at peak and hourly off.
There’s a long-running dispute in the transit and planning world about the relationship of commuter rail to land use. Does commuter rail to suburban and exurban areas damage the environment by enabling sprawl, or help preserve it by taking long car trips off the road?
It can be hard to tease out correlation from causation in these circumstances. Over the last several decades, commuter rail systems in the nation’s major metropolitan areas have spread ever-deeper into the suburbs and exurbs, propelled by powerful suburban politicians who crave the glory of being seen as “relieving congestion” (one thing that almost anyone can agree American-style commuter rail can’t do). But does commuter rail being built to a sprawly area (like, say, Elburn, IL) make that area’s growth possible, or would the growth occur anyhow, with the potential rail riders simply driving to work?
Commuter rail to sprawlsville: the western terminus of Metra’s UP-W line.
Commuter rail has, for better or for worse, been an increasingly popular mode over the last several decades, so figuring out which way the dependency goes has been hard; there are plenty of areas that are sprawly without the benefit of commuter rail, but few that have actually lost service since the modern era of commuter rail (defined roughly as the takeover of bankrupt private services by government corporations in the ’70s and ’80s) began. Identifying such an area would allow us to determine whether the loss of service arrested growth, forcing it into a more compact area, or whether growth continued unabated, with commuters switching to cars.
There is one rather infamous example of such a loss of service–SEPTA’s former diesel operations. When SEPTA took over responsibility for the Philadelphia-area regional rail system (first through subsidies paid to the operating freight railroads, then directly) it inherited not only the core electrified services of the former Pennsylvania and Reading networks but several diesel-operated semi-intercity services, extending to Newark, NJ, Bethlehem and Allentown, and Pottsville/Reading. The two all-Pennsylvania branches, in particular, essentially served as extended commuter services for riders to downtown Philadelphia. By 1981, amidst a funding crisis and apparent apathy from SEPTA, service had ended on all three extended routes (diesel service remained for two more years on the shorter Newtown-Fox Chase branch).
And while (among other things) the end of diesel service caused SEPTA Regional Rail ridership to crater (it had been around 118,000 in 1975 and fell to around 85,000 in the mid-’80s after the opening of the Center City Commuter Connection), it also gives us an opportunity to examine suburban growth in the sudden absence of commuter rail. In the interests of seeing what happened, I examined population growth data from each of the towns along the Bethlehem/Allentown line and graphed them against growth trends in Montgomery and Bucks Counties and the Philadelphia MSA as a whole. I included data from towns along the line in those two counties, but not from Centre Valley or Hellertown, the two Lehigh County towns on the line aside from Bethlehem and Allentown, on the logic that those towns were much more tied to the economies of the Lehigh Valley than that of Philadelphia. Town- and county- level population data is from Wikipedia (because why dig into census sheets when someone else already did it?); MSA data from here. You can access the full sheet here: (Allentown Branch), but this is what’s important:
If the elimination of SEPTA’s diesel service had impacted suburban growth along the line, we would expect growth in those towns to fall during the ’80s. Instead, the towns as a total grew by 14%–11% more than the Philadelphia MSA as a whole, and 5% more than their containing counties. Growth fell a little below regional trends in the ’90s, but almost indistinguishably. Over the 60 years I examined, growth in the station towns either matched regional trends or was actually slower. It’s hard to argue from this data that there’s any correlation between the presence/absence of commuter rail service in a particular town and its growth.
There is both good and bad news for transit advocates in this (admittedly unsurprising) conclusion. On the one hand, that suburban sprawl can continue without transit means that advocates and agencies should feel free to resist the loud calls for expensive (in terms of both capital and operations) outward extensions from exurban politicians and commuters. On the other, growth will probably continue regardless of transit, so why not try to get people out of their cars? I haven’t seen extensive data from SEPTA’s Regional Rail operations in the wake of the diesel service eliminations, but the overall fall in ridership suggests that commuters from the areas that lost service chose not to drive to closer-in termini, but to drive all the way to Center City.
Should we build commuter rail to sprawl? On the whole, I think there are (as always) much higher priorities for transportation funding (and government funds in general). Commuter rail as currently conceived in the US is really expensive to operate, and sending it out to the far reaches of a metropolitan area is essentially a favor to exurban commuters and a subsidy for bad regional planning. But if the funds are available (and can’t be spent on anything better) and if suburban towns are willing to shape their land-use decisions (at least in the immediate station area) around transit, I suppose some service is better than none. Either way, if the Allentown Branch case is even remotely representative, it’s pretty clear that while commuter rail might be a subsidy for sprawlers, it isn’t actually a cause of new sprawl.
Personal note: it’s been quite a while since I’ve posted here. Beginning of the semester sucks. But hopefully 2600+ words makes up for it 😉
Expansion of east-west passenger rail service in Massachusetts has been a topic of discussion for quite a long time. Politicians from decaying industrial cities like Worcester, Springfield, and Pittsfield want a reliable connection to Boston’s vibrant economy; travelers want an alternative to the fast-if-there’s-no-traffic-but-there’s-always-traffic Mass Pike, and the state’s liberal voters tend to be more supportive than average of infrastructure projects. There’s also a good bit of nostalgia for Massachusetts’ days as the technological and political haven of American railroading.
Today, there seems to be a good bit of momentum for extension of passenger service west of its current terminus at Worcester. All of the Democratic candidates for governor agree on the necessity of such service, and it given the state’s recent spree of line acquisitions for passenger service, seems likely to happen one of these decades. That seeming momentum got me thinking about the possibilities for a more thorough east-west service along CSX’ Boston Line, the former Boston & Albany division of the New York Central. Service to Springfield is one thing; getting up and over the sparsely inhabited, hilly, and curvy line across the Berkshires to reach Pittsfield and Albany is another entirely.
Of course, I have a personal stake in exploring this possibility; I live in Albany, many of my friends are in Boston, and I would love to have convenient rail service. But is it feasible? The situation I face as a consumer is thus:
Google Maps estimates a driving time of 2:39 from my apartment in Albany to South Station. Realistically, you have to leave 3-3.5 hours, because while the Mass Pike is fast and free-flowing from Albany well past Springfield, once you hit the interchange with 84 in Sturbridge, all bets are off.
Greyhound offers direct schedules in the 3:30 range, with a stop in Worcester, but there are only a few buses per day in each direction. There are also local Greyhound buses that stop in the Berkshires towns, but they require a transfer in Springfield to get to Boston, and the trip is over 4 hours. All Greyhound buses are subject to Mass Pike delays.
The less said about Amtrak’s lone train on the route, the Boston section of the Lake Shore Limited, the better; this post is about the future. But: it’s currently scheduled for 5:45 eastbound and 5:40 westbound. So there’s that.
My hypothesis is that if a train could get between Boston and Albany in 3:30, it would attract high enough levels of ridership to keep it going; I’d probably ride at that time point. And of course anything faster would be a bonus. But can we get the trains going that fast? Albany to South Station is exactly 200 track miles (compared to 170 on the freeway, a major reason trains have had trouble competing in the corridor), so a 3:30 trip time corresponds to an average speed of 57 mph. On the one hand, 57 mph isn’t a particularly ambitious speed goal. On the other hand, Amtrak’s Lincoln Service, which uses predominantly flat, straight lines with stretches of 110 mph running, is scheduled for a 53 mph average speed between Chicago and St. Louis (over 284 miles), and Empire Service trains between New York and Albany are around 60 mph on average. So to achieve competitive travel times, Boston-Albany passenger trains must achieve average speeds comparable to, or even higher than, those on many of Amtrak’s higher-speed corridorservices, many of which face fewer geographic obstacles. Is that doable? Let’s delve in.
As mentioned above, the Boston & Albany corridor is notoriously difficult for high (ish)-speed trains. The route opened in 1841 as one of America’s first long-distance railroads; its climb over the Berkshires also claimed the title of the world’s highest railroad at the time. The routing is tortuous and twisting, following river valleys to find an acceptable grade. That being said, unlike most American railroads (at that time and for about 50 years thereafter) the line was designed to an extremely high standard. Supervising engineer George Washington Whistler (the less-famous parent of the painter) insisted on curves as gentle as possible under the circumstances and clearance of the right-of-way for double-tracking from the very beginning. In essence, Whistler and the owners of the B&A traded more severe grades for gentler curves–the rival Fitchburg Railroad/Hoosac Tunnel route 40 miles to the north made essentially the opposite choice, with sharper curves but less severe grades. Those choices have made the B&A an operating nightmare for freight over the years, but they make it not totally hostile to passenger service, unlike the Hoosac Tunnel route.
For our purposes, though, the Hoosac Tunnel isn’t the competitor; the Mass Pike is. And as I already noted, the freeway’s route, built with the advantage of mid 2oth-century technology, is 30 miles shorter than the B&A. Here, too, though, the B&A has at least one advantage. Unlike the Mass Pike, the railroad serves the downtowns of the three major Massachusetts cities along the route–Worcester, Springfield, and Pittsfield–directly. In Worcester and Springfield, freeway spurs lead to downtown, so the distance of the Pike isn’t a big deal, but Pittsfield has no direct freeway access and is a good 20-minute drive off the Pike. The lack of immediate freeway access also means that buses cannot serve the Boston-Albany corridor in a linear manner. That’s why Greyhound doesn’t run buses between Boston and Albany with stops in Worcester, Springfield, and Pittsfield: the repeated backtracking to the Pike would make it an unacceptably long trip. Additionally, these cities are essentially the only feasible stops on a Boston-Albany service, and they fall nicely into an every-50-miles pattern: Pittsfield is 49 track-miles from Albany, Pittsfield-Springfield is 53 miles, Springfield-Worcester 54, and Worcester-South Station 44. The Lake Shore Limited makes an additional stop at Framingham, halfway between Worcester and Boston; that stop could probably eliminated with a timed transfer to/from a local commuter rail train at Worcester. The only other possible stops that I can imagine are Palmer, MA and Chatham, NY, but neither really warrants a stop on an intercity train. This is abnormally few intermediate stops for an Amtrak corridor service, which typically stop every 20-30 miles. The less-frequent stops might–might–help trains maintain a higher average speed, even if top speeds aren’t all that great.
But just how fast can we get the trains going? I know I promised not to speak much of the Boston section of the Lake Shore Limited, but its current schedule is the place to start.
One thing is immediately clear: this schedule is massively padded in both directions. If, following my division of the line into four segments (Albany-Pittsfield, Pittsfield-Springfield, Springfield-Worcester, and Worcester-South Station), we look at the two terminal segments, we can see the insertion of the padding. Boston–Worcester is scheduled for 1:03 outbound (westbound)–and 2:13 inbound (eastbound). Likewise, Albany-Pittsfield is scheduled for 1:04 eastbound, but 1:59 westbound. If we eliminate the massive padding, we can immediately cut a little over an hour off of the Lake Shore‘s scheduled time, cutting it to a still-uninspiring (and non-competitive) 4:45 or so in each direction. Of course, the padding in the current schedule exists for a reason; the Lake Shore‘s on-time performance is notoriously horrific, earning it the nickname Late Shore Limited. Any scenario that envisions increased passenger traffic will certainly involve re-installing double track along the entire B&A corridor (not a problem in terms of ROW), with the state paying in return for absolute passenger dispatching priority. Planned track improvements now that the state owns the Worcester Line between South Station and Worcester should cut another 15 minutes or so off of travel time, leaving us with a nice, round time of 4:30–still an hour slower than might be considered competitive.
One way to improve travel times is by increasing track maintenance to levels that will allow higher speeds. Currently, MBTA is struggling to boost its portion of the Worcester Line from FRA Class III (6o mph for passenger) to Class IV (80 mph for passenger) standards; but more can certainly be done. Most of the rest of the line seems to be maintained to Class III standards, but the ingredients exist for converting it to allow for higher passenger speeds: the entire line west of Framingham is signalled with (antiquated, but upgradeable) cab signals, and there are relatively few grade crossings due to the age of the line. That being said, upgrading absolute train speeds will have relatively little effect because of the line’s severe curvature; with the exception of the more-or-less tangent 20-mile Palmer-Springfield segment, the limiting factor on train speeds is generally curvature, not track or ballast structure. Certainly, building the theorized second track to Class IV or V (V requires cab signals, but luckily the Boston Line has them) would help, but is there a better way to boost average, rather than absolute, top train speeds?
For the answer to that question, we can turn to the opposite coast, where for the last decade and a half Amtrak has been happily operating tilting Talgo trainsets on behalf of the states of Washington and Oregon on the Cascades. These Spanish-designed trainsets are lightweight (though not as lightweight as they could be, thanks to FRA regulations) and their tilt mechanism allows them to navigate curves faster than conventional trains. Various factors–expense, Talgo’s insistence on doing maintenance itself, mechanical discontinuity with other fleets, lower capacity–have kept the Talgos from being adopted more widely in this country, but they’re a very, very strong fit for a curvy, hilly route like the Boston & Albany. The criteria for their ability to save time are complex, but as this Trains Magazine explainer puts it: “Tilting reduces trip time only when the route has a reasonable concentration of curves with curve speeds between 50 and 80 mph. In this speed range, a Talgo-type train will be able to negotiate a curve at speeds 5-10 mph faster than conventional cars. Generally, tilting does not generate significant time savings unless the curve density on a route is 30 percent or higher.” This description could be written for the B&A. There are virtually no tangents of any significant length, but relatively few of the curves are so sharp that they necessarily drop the speed of the train below 50 mph. Equipping corridor trains on a Boston-Albany route with Talgo trainsets could do a lot to boost average speeds–but how much?
The current Cascades schedule shows Talgo-equipped trains saving only about 10 minutes over the Superliner-equipped Pacific Starlight, but that’s a product of ongoing summer trackwork. Historically, Talgo schedules have saved 35 to45 minutes, or about 15%, on the 187-mile Seattle–Portland segment, which is actually less curvy than the B&A (Talgos save little to no time on the very straight segments between Portland and Eugene, and only some north of Seattle). Knocking 15% off of the theorized 4:30 unpadded Lake Shore Limited time would give us a time of 3:50 or so, getting closer to our goal but not quite there yet, and still over an hour longer than a direct Boston-Albany bus. However, as mentioned the Portland-Seattle segment isn’t actually that comparable to the B&A, being less curvy and with lower potential maximum speeds because of the lack of cab signals. And sure, there is no other modern experience with Talgo operations in the US. A theoretical application of Talgo equipment, though, is perhaps the next-best thing, and that’s what we find in Pennsylvania. Found via this Sic Transit Philadelphia post, Samuel Walker of Test Plant managed to get a Talgo engineer’s estimate of time savings from using their equipment on the old Pennsylvania Railroad mainline between Harrisburg and Pittsburgh, a route that in age and alignment is very comparable to the B&A. By the engineer’s calculation, Talgo equipment could cut the 254-mile Harrisburg-Pittsburgh run from 5:30 to 4:10, or from 4:56 to 3:36 if the 34-minute schedule pad is eliminated. That’s a savings of 25% before padding and 28% after. If we cut 25% off of the paddingless 4:30 Boston-Albany running time…we get a time of 202.5 minutes, or 3:22.5–just below the magical (to me!) 3:30 time cutoff. Again, that’s with nothing assumed as to track quality other than the already planned upgrades inside Worcester and the installation of a second track built to Class IV speeds west of Worcester.
Of course, those upgrades are far from nothing–probably on the order of hundreds of millions, if not multiple billions, of dollars. But if Massachusetts can find the money for a second track and signal upgrades along the B&A and if state politicians are willing to negotiate hard with CSX over dispatching priority and if Amtrak or the state are willing to take a risk on Talgo equipment and if the Talgos prove able to do for the B&A what they could do for Pennsylvania…I see no particular reason that a functional Boston-Albany service couldn’t be established in relatively short order. A time of 3:22 end-to-end isn’t magical, but given that a train would be able to hit Pittsfield, Springfield, and Worcester within that time frame. And while a full 3:22 might be at the high end of the time savings that Talgo can offer, even if the time savings are more in the Cascades range of 15% a combination of new equipment and more extensive track upgrades should be able to get travel times down into the 3:30 range. That’s certainly better than any bus can do while stop at all three intermediate cities.
So I do think a renewed, relatively fast Boston-Albany service is possible. It would require significant investment, but it seems to be doable. The main advantage of a train over buses is that one service will be able to stop at all of the major cities in the corridor. Potentially, such a service could become the backbone of a frequent intercity rail network serving the entire state, with the Boston-Albany trains making connections at Springfield and Pittsfield to DMU services in the Pioneer Valley and Berkshires. That’s far, far in the future, but it would be an enormous mobility “win” for the entire state.
A couple of notes:
1. One particular challenge for the introduction of Talgo equipment to the line might be the presence of high-level platforms. There’s no question that the next-generation trains on the line will be built for high-levels; South Station, Back Bay, Worcester, and Albany have full high-level platforms and Springfield is getting them as part of the NHHS project, not to mention that Amtrak is going to an all-high-levels policy in the Northeast. Of current intermediate stops, that leaves only Framingham and Pittsfield. In Pittsfield, building a side track for a high-level platform so as to maintain freight clearances shouldn’t be too hard. Framingham is a little more of a challenge; it currently has mini-highs and will still be on a freight clearance route, which perhaps further militates for not stopping there. That being said, Talgos are low-slung and there are no examples of high-level-platform-equipped ones operating in the US, so that might increase costs some.
2. One of the problems with the current setup on the Worcester Line is that, while there are three tracks in segments, much of the ROW was cut down to two tracks from Newton in to accommodate the Mass Pike. There are, further, very few sets of crossovers. One of these things can be remedied; the other realistically cannot. More crossovers it is (this will help MBTA trains more than intercity service!).
3. CSX may not love the idea of ceding half of their ROW for a second track to be committed mostly to passenger trains, but it’s not like Massachusetts doesn’t have leverage. The state has already paid for full double-stack clearance, and along with that carrot can hold out the stick of capital investment in helping the Pan Am Southern Alliance clear the Hoosac Tunnel route for higher speeds and double-stacks. CSX doesn’t want to lose its huge advantage in the Boston market; the state shouldn’t be afraid to play hardball, perhaps even asking CSX to pick up some of the tab for the second track.