The night time is the right time – Pittsburgh by sleeper

Part 3 of a 3 part series.

Last time, before I got sick, I explored the various ways in which state investment could improve rail service between Philadelphia and Pittsburgh. But even an optimistic projection of funding for new trainsets and track upgrades couldn’t bring PHL-PGH trip times much below six hours. Six hours is still a long time, even when used productively, and the necessary level of funding may be difficult to obtain, even though it’s relatively cheap by transportation infrastructure standards. Can intercity rail hope to compete if it’s restricted to the seven hour and twenty minute schedules of today?

It can, because it so happens that seven and a half hours is a nearly perfect timing for a direct sleeper train. Amtrak does not do much in the way of point-to-point sleepers today, so even veteran rail passengers may be unfamiliar with the business model. With sleeper cars no longer on Northeast Regional trains 66 and 67, the closest approximation in the United States is the Auto Train, which is something else entirely. But it’s a common practice for Deutsche Bahn, and First ScotRail runs a world-class exemplar of direct service on its London-Scotland Caledonian Sleeper, the lowland sections of which (Edinburgh and Glasgow sections) run on a similar seven-and-a-half hour Pittsburgh-Philadelphia schedule for a hypothetical Pittsburgher (to resurrect the PRR’s name for the all-sleeper train that ran the overnight New York-Pittsburgh schedule). First ScotRail is sandbagging their speed, of course; London-Edinburgh is only four hours and twenty-two minutes on the East Coast Main Line in the daylight. But the Caledonian Sleepers also spend half an hour every morning parked in the insanely busy Euston Station, taking up a track on the cusp of a frantic London rush hour, allowing arriving passengers a leisurely wake-up from the slightly short night. I’m not familiar enough with the layout of Pittsburgh Union Station to know if it’s feasible to replicate that mode of arrival there, but it’s most certainly possible at 30th Street Station, however much Amtrak Operations will inevitably grumble about it, and similarly impossible in the madhouse of Penn Station New York.

And what a schedule! Late evening departures from Pittsburgh and New York, around 21:30 or 22:30, give or take, mean that arrivals in Philadelphia, Pittsburgh, and New York can all occur before 8:30, enabling a full day of work or activity immediately on arrival; any arrival before 7:00 at 30th Street can’t be meaningfully beaten, even by the 5:20 (ow) flight out of PIT; the 6:31 PHL arrival still leaves the early-morning flyer a 20 minute train or cab ride from Center City, and in the other direction, airport connections from PIT are strictly worse. And unlike a predawn flight, you don’t experience any of indignities of flying, or anything else for that matter; the perceived travel time by train can be only minutes, depending on how quickly after boarding you can fall asleep.

An overnight train that matched the Pennsylvanian‘s current seven and a half hour runtime would be slightly short of the eight-to-nine hour ideal, which is much better than being slightly too long; there are infinitely more, and cheaper, ways to chew up extra time on the way than there are to save time. The next easiest after arriving early and just sitting, may be to run under electric power and change locomotives at Harrisburg; the current Pennsylvanian changes to diesel at 30th Street, and I’ve been assuming so far that increased Pittsburgh service would also run as diesel-under-wire across Pennsylvania. The process of changing locomotives is more time-intensive than the performance benefits of electric power, but it would mean being better neighbours to the towns along the Main Line. Other options include slowing the train down to present a more freight-like speed profile on Norfolk Southern, in exchange for better track access conditions (i.e. fees), or simply padding the schedule to guarantee hitting restricted slots on time, especially eastbound towards Philadelphia and New York.

The shortish schedule also helps Amtrak avoid one of the biggest money bleeds on its sleeper services: the diner car. Food and beverage losses dog all Amtrak routes, but especially the long-distance trains, and while Amtrak has pledged to balance the F&B books in the next five years, it will probably come from properly accounting for free meals in diner cars for sleeper passengers as loss-leaders for sleeper service. A service originating hours after dinner and terminating at dawn in a major city’s downtown should have no need of a full diner car; trains 66 and 67 never had them, in any of their Boston-Washington or Boston-Newport News incarnations. A café car alone should be more than sufficient for any insomnia-fuelled midnight snacking or pre-arrival coffee needs. The Auto Train, which feeds every passenger dinner and continental breakfast and thus has very predictable costs, has commissary costs of $52.89 per passenger (PDF), before even the cost of on-board labor, which runs about $40/hour total compensation. That level of additional cost would devastate the economics of Pittsburgh-Philadelphia or Pittsburgh-New York service, if that tab was picked up by passengers, and would sink the politics if it were the burden of the state.

Now, it’s not currently feasible for Amtrak to run an additional sleeper service anywhere east of the Mississippi, because Amtrak is already using all 50 of its Viewliner sleeper cars. Fortunately, they’re buying more; 25 as part of a more comprehensive 130-car Viewliner II order currently under construction by CAF. While Amtrak presumptively has plans for all 25 sleepers of that first wave (at first, to sub in for the Viewliner I fleet as they go through midlife rebuilds), they have options to extend the order that they have not yet exercised. PennDOT can purchase and exercise options for its own Viewliners, which would operate in Amtrak’s equipment pool but would be accounted for as an in-kind contribution to the new train. Purchasing Viewliners would cost on the order of $20 million up front, for two trains each of three sleepers and one bag/dorm car; Amtrak may insist for operational reasons on PennDOT purchasing additional “protection” cars for the Pittsburgh base, which may add up to $10 million on top. That purchase cost would be amortized over the expected 30-year life cycle of the cars (and probably longer), and would also earn Pennsylvania a significant discount on the contract with Amtrak to operate the service. It’s not a coincidence that, now that the PRIIA 2008 rules for Amtrak state-sponsored services are fully in place, sponsoring states including California, Illinois and Michigan are suddenly very keen on buying their own passenger rail equipment. Since the Viewliner IIs are still just getting started at the factory in Elmira, NY, and Amtrak’s 130 are going to go first before any option orders, it means that PennDOT can’t rush this plan into fruition, but between PennDOT and Amtrak that wasn’t likely to begin with.

How much ongoing expense would the Commonwealth be taking on? Well, the current estimate is that Amtrak sleeper trains cost $20 per train mile more than their all-coach counterparts. The Pennsylvanian has expenses of $16.3M per year, so an overnight NYP-PGH Pittsburgher would have expenses of approximately $22.8M/year, with today’s Amtrak standards. However, every single one of those Amtrak sleeper trains has a diner car, and we are trying to absolutely avoid the necessity of the expense. Exactly how much of a train’s expenses are from the dining car is variable, but I’m comfortable calling an estimate of 15% conservative, which would reduce our estimate to $19.4M/year. In order to maintain the same operating ratio as the Pennsylvanian, the train would need to sell $12.4M per year in tickets, and receive $4.5M/year in state support; in the likely scenario that total ticket revenue can be higher than that, state support can be reduced. Revenues will be helped by the ticket premium that sleepers command; they will be hurt by the low numbers of “seats” per car. Setting the exact balance will be a business and political decision between Amtrak and the Commonwealth; the three-sleeper-plus-three-coach configuration Amtrak uses for its current long distance trains is not very ambitious, and I would hope that the success of an overnight train in attracting passengers would lead to further Viewliner purchases and the lengthening of the train, which will help with fixed costs like the capital cost of the locomotives and the labor cost of the engineers operating them. But using the $166 round-trip cost of a Amtrak Business Class tickets on the Pennsylvanian PGH-PHL as a floor, and the $292 round-trip cost of a connecting PIT-BOS-PHL flight (currently the cheapest, for a Wednesday 10 weeks from now) as a soft ceiling, I don’t see ticket revenue being a major problem; if Amtrak and PennDOT decide their only real competition are direct flights on US Airways, and the service is popular, it’s within the realm of possibility to cover the entire cost of the train through ticket revenue. If I had to bet serious money, and assuming breaking even on operational costs was higher priority than access to the service, I would put the probability of breaking even happening in the first five years of the service at no more than 25%. But that’s just a gut call without a market study or math behind it, and needs to be tested against reality.

Even without the financial albatross of a diner, a Pittsburgher has nontrivial handicaps between it and success. A Viewliner car simply does not carry many people; high-density American sleepers went out with the retirement of the last Philadelphia-manufactured Slumbercoaches in 1995, and American railroads have been traditionally afraid of the non-private couchette car layout that keeps sleeper service accessible to budget travellers across Europe, which is a travesty but not one for the Commonwealth to fix all by itself. Direct service to New York will probably be necessary to strengthen the finances of the train, but the eastbound’s access to Penn Station in the morning will be impossible unless the slot for Keystone Service Train 640 (the 5:00a departure from Harrisburg) can be stolen and repurposed. (The feasible alternative is to keep the train parked at 30th Street until the mad rush is over in New York, and deadhead the train empty/coach-passengers-only after 10:00 when tunnel slots open up.) Politically, it may be a much harder sell for the interior of the state to make such a clearly Pittsburgh-and-Philadelphia-centric transportation investment, versus more daylight trains better-serving the intermediate cities, but the capital costs of sleepers are lower, even in the worst case, than what Norfolk Southern may demand for additional daylight slots.

Still, I think the point has now been thoroughly made, that it’s fully within the power and budget of the Commonwealth of Pennsylvania to make the investments that will tie its two largest metropoles together by rail. It’s now a question of choosing priorities, both in the Legislature and at PennDOT. With statewide elections coming up, now might be an interesting time to start whipping up support, both from gubernatorial and legislative candidates, with special consideration for the Allegheny and Philadelphia legislative delegations. With the Viewliner production bottleneck squarely between us and any potential beginning of service, there’s plenty of time to line up support. But the journey of 353 miles needs to take its first step.

This post has been crossposted to this old city.