The iron road to the Steel City: Keystone West

Part 2 of a 3 part series.

So, the question I posed yesterday was, “How can intercity rail help fix the mess that is travel between Pittsburgh and Philadelphia today?” The key to implementing any program of improvement to the Keystone West line (the term of art for trains that continue west of Harrisburg), will of course be Jarrett Walker‘s old axiom: Frequency is Freedom.

When Southwest made its doomed run at the PHL-PIT market, it flew six trips a day. That’s a typical schedule for any airline flying a mid-market origin/destination route like this one. Greyhound, an early and enthusiastic adopter of the Pennsylvania Turnpike, has about ten scheduled runs each way between the Steel City and the 10th/Filbert bus terminal. Megabus, making a strong play for the Turnpike market with its 81-seat double-decker buses, runs three daily round trips from curbside at 30th and JFK. If ordinary business and leisure travellers are going to take Amtrak seriously as an option, the schedule is going to have to look more like those, and less like the once-daily schedule of today.

The present Amtrak timetable lists a 12:42 departure from 30th Street Station and a 20:05 arrival at Pittsburgh Union Station. That is mostly convenient for connections to the west on the Capitol Limited, which calls at Pittsburgh at midnight on its way from Washington to Chicago. Eastbound, the Pennsylvanian is again timed for Capitol Limited connections as much anything else, leaving Pittsburgh bright and early at 7:30, and arriving Philadelphia 14:55. That schedule might be convenient for a Pittsburgher with an evening meeting in Philadelphia, or for a Philadelphian able to charge an extra night or two in a downtown Pittsburgh hotel to an expense account. But it’s pretty useless for anyone else, apart from aforementioned transfer passengers. Even if you think that’s an acceptable schedule, what’s good for the goose should be sauce for the gander; it cries out for a midday departure from Pittsburgh, arriving Philadelphia in the evening, and an early train westbound from Philadelphia arriving in Pittsburgh early enough to check in to a hotel before dinner. And as much as anyone gets twitchy about the timekeeping of any Amtrak train, it’s barely humane to the people changing from the Pennsylvanian to the Capitol Limited to keep them sitting in an underused Pittsburgh station for nearly four hours; eastbound connections only seem to require two and a half hours, from a route where timekeeping is much more out of Amtrak’s hands.

I’ll note briefly, this being a Philadelphia-focused blog, that there are more markets served by a PHL-PGH train than those endpoints. Intermediate cities like Johnstown and Altoona have lost all direct air service with both Philadelphia and Pittsburgh. And the Turnpike serves a different alignment; if you want to serve all those intermediate markets by bus, your schedules and timekeeping are going to go pear-shaped very quickly.

The current Pennsylvanian service provides one-seat rides to and from New York City, which is nice to have but theoretically not essential for every run; passengers dislike transferring, but Pennsylvanian (and Keystone) trains are shorter than Northeast Regional trains, making them a less-efficient use of valuable tunnel slots at Penn Station. Any one-seat ride between New York and Pittsburgh necessitates a change of locomotive between electric and diesel, probably at 30th Street where the train would also have to “change ends” (i.e. reverse direction). That process takes 30 minutes, which in my opinion falls squarely in the realm of “might as well transfer”.

If one takes one train departure every three hours as a service guideline (sparse but plausible as a minimum), and reasonable waking hours at Pittsburgh and Philadelphia during which all service must happen as 07:30-21:30, that allows for three trains a day each way. It’s not a coincidence that that’s how many trips Megabus is running, although Megabus is much more lax about early departures and midnight arrivals. It seems to be a reasonable floor for overland service frequencies in daylight. More might be better, but it’s a good place to start.

The next step is to consider how to reduce the time it takes to get from A to B. Now, I put this as a much lower priority than increasing frequency, because I have confidence that many people will find the experience of riding the train to be better than the experience of driving, or of riding the bus. On the train, you don’t have to devote all your attention to driving, nor do you have to carefully time stops for meals or bathroom breaks. The nearest bathroom is never more than an 80 foot walk away, and while the Amtrak café car can’t be mistaken for five star cuisine, neither can anything you can get at Sideling Hill Plaza.

Unfortunately for the cause of higher speeds, we’re caught up by the structure of the American railroad system. While Amtrak owns the tracks from Philadelphia to Harrisburg, Harrisburg to Pittsburgh is owned by freight carrier Norfolk Southern (NS). NS is the more-passenger-friendly of the two major eastern roads, but it’s not going to undertake upgrades on Amtrak’s behalf for charity. They’ll upgrade their line, but they’ll want a check from PennDOT to cover their costs. And that will be a very big check.

The good news is that the major improvements to the Keystone East (Philadelphia to Harrisburg) have already been accomplished. The story of how then-Amtrak CEO David Gunn was able to bring the Keystone tracks back up to 110 mph electric operations is a fascinating one, which you should read in his own words. While there might be time left to squeeze out of the line east of Harrisburg, not much of it is relevant; Amtrak’s current diesel locomotives have a top speed of 110 mph. The next generation of diesels being ordered on Amtrak’s behalf by a coalition of state sponsors will be capable of 125 mph, but that’s still not very much time savings left on the table.

Leading off the big-ticket items will be adding stretches of third track to the line. (The PRR had four tracks Pittsburgh-Harrisburg; Conrail ripped out two over most of the distance.) A 2005 study commissioned by NS said that adding frequencies would require $110 million in improvements, featuring 66 miles of new track along the 250 rail miles west of Harrisburg. Conditions have changed since then, but that’s a good order-of-magnitude estimate for the work that would have to be done, just as a baseline. Still, compared to expanding road or airport capacity, that’s still in the realm of dirt-cheap.

As it so happens, there are a lot of small, incremental improvements to the Keystone West line possible, that the state government could facilitate, and would mostly accrue benefits to NS freight ops. Most involve land swaps to straighten the line out more than was possible when the railroad was being built with picks and shovels in the 1840s and 50s. It’s easier and more important to optimize a railroad to be flat than to be straight, but we have industrial earth-moving equipment lying around that isn’t needed to build new roads, now that we’ve hit peak VMT (never mind that we seem to be building them anyway). We might as well use them to take out some of the sharper curves, since they force freight and passenger trains alike to slow down, which wastes time and fuel. How much time you could save from the schedule would depend heavily on how much you were willing to spend; straightening a 40-mph curve in a rural area to allow for 79- or 110-mph speeds should be a cheap operation, but also only saves a minute or so from the schedule. One major savings available would be to flatten the ruling grade east of Johnstown, but that would involve bypassing Horseshoe Curve and possibly the entire city of Altoona. Engineering marvel that it is, I’m not attached at all to the curve (there goes my railfan cred for all time), but serving the city of Altoona is important enough to me that I’m skeptical of the 15-18 minutes savings available from a bypass; NS might be more interested in the ability to forego helper locomotives for the steep climb up the Allegheny Frontal.

Electrification is a popular proposal, but sadly it’s still slightly too expensive for now. I’m a rare fan of freight electrification, but the freight railroads disagree with me for now, and it’s their money, not mine. And electrifying ~100 daily freight trains on behalf of between two and eight daily passenger trains is absurd on its face. I think PennDOT’s ideal role here, is to keep a plan updated for a rapid buildout of electrification, for use at some future point by NS, probably under some flavor of energy crisis conditions (various sources in the freight railroads have put the break-even price of diesel fuel vs. electrification at around $5.00/gallon; it’s slightly lower in mountainous terrain like central Pennsylvania, where regenerative braking is a bigger deal). The estimated cost of electrification is between $550M (Conrail, 1971) and $650M (Penn School of Design, 2010).

The cheapest and fastest way to shave time off the schedule also solves an entirely different problem: rolling stock. I’ve been assuming so far that any service would be operated with Amtrak’s P42 locomotives and Amfleet I coaches, and also that Amtrak has enough of both to cover any schedule. That is no longer, strictly speaking, true. The locomotives are not a problem for now, but after years of skyrocketing demand on the Northeast Corridor and its feeder routes, Amfleets would have to come from somewhere else, and PennDOT would be on the hook for making up that lost revenue. The next generation of single-level Amtrak coaches, meant to replace and expand upon the Amfleets, is still in the design phase, and won’t be available for years.

However, there are two high-speed Talgo trainsets sitting in Wisconsin that are capable of making much better time over the railroad than conventional Amtrak; Talgo engineers estimate that they could cut trip times west of Harrisburg from 5 hours 30 minutes to 4 hours 10 minutes, i.e. Philadelphia-Pittsburgh trip times could be reduced to 6 hours flat. Those trainsets were originally built for Chicago-Milwaukee service, and are available because the anarchist Governor of Wisconsin, Scott Walker, engaged in shenanigans to stop payment on them after his election in 2010; Talgo is suing the state of Wisconsin for $66M (i.e. more than the trains were worth to build) for breach of contract, costs, defamation, and other claims. The Talgos, sadly, are not a good long-term fit for Pennsylvania; they’re short and built for low platforms, which restricts them to a single track in 30th Street Station and will soon see them completely incompatible with key stations like Paoli and Exton. They are also incompatible with Penn Station New York, which would settle the question of one-seat rides PGH-NYP in the negative. But as a short-term stopgap, they’re probably the best solution available for Pennsylvania. We’d need to lease them from Talgo America and somehow boost them out of the factory in Milwaukee where they’re currently mothballed, but it’s certainly doable. To achieve maximum performance out of the trainsets, PennDOT would have to fund some amount of adjustment to the signal system on NS; that’s a completely acceptable expense. They may also require additional third track in addition to the baseline.

After all this, what’s the final bill? In annual operations, the controversy will be between those who believe that additional frequencies will achieve diminishing returns in ridership, and those who believe that additional frequencies will attract greater ridership through network effects. I am in the latter camp, but in the interests of budgetary conservatism, I will say that the effects will cancel each other out completely, and the necessary subsidy of three round-trips is three times the cost of one: by current estimates, a total of $11.4M per year. I will pessimistically estimate the total cost of track and signal upgrades to the Commonwealth at $250M. I think the total will be less than that, at least in any initial phase, and that NS may be convinced to contribute, since the upgrades will also benefit the expansion of their freight business, which is growing very fast as the direct rail link between Bakken oil and Philadelphia-area refineries. Still, again, I will be conservative. And if PennDOT pursued the Wisconsin Talgos, the purchase price would be around the $36M that Oregon paid for its two trainsets, although it might be wiser in our case to lease.

Still, the Pennsylvania Turnpike has an annual capital plan of between $500M and $750M for the next ten years, on revenues of more than $900M. The Turnpike Commission’s fiscal health is not what it was pre-Act 44, but Act 89 has pulled it out of the fire. While Act 44 was a poorly-designed piece of financial engineering, putting Pennsylvanian service in the budget of the Pennsylvania Turnpike Commission has an elegance that Act 44 did not, and also a strong element of fiscal realism; the PTC can afford it.

So now that we’ve dropped travel time from 7 hours 32 minutes to 6 hours, and are running three or four times a day, and identified a potential source of funding, are we done? Well, no. That 6-hour endpoint-endpoint time is still a big, nasty constraint; even if you can work from your laptop on the way, that’s still a lot of time to spend in transit, and time-sensitive people are still likely to fly, if expensive airfare is in their budget. Even if you’re fine with arriving at 11:00 at night, and not everyone is, you’re still catching the last train of the day at 5:00. We can still do better by travelers. Tomorrow in Part 3, I’ll show how the effective time on the train can be dropped to nearly zero, with far less in terms of investment or lead time required, and only a little upending of conventional wisdom.

This post has been crossposted to this old city.