Crossposted to Spacing’s Wire

TTC Chairman Howard Moscoe waxes on about the benefits of automatic train control (a.k.a. “driverless trains”), suggesting that they are “a bargain” at $750 million. The Toronto Star has more details. In his view, putting computers in charge of subway operations would allow the TTC to:

• reassign TTC personnel to other duties, like keeping the stations cleaner.
• run trains closer together, significantly increasing the capacity on the subway lines.
• opens up the prospect of night service, or increased service throughout the day, since you don’t have to pay for the train crews.

It all sounds well and good. Almost too good. So, what’s the catch?

There are a few.

First of all, if operations elsewhere are any indication, automatic train control will not mean an elimination of train crews — much to the relief of the TTC unions, and to passengers unwilling to put their trust in an on-board computer. Montreal’s Metro has been running on automatic for decades, but there is still somebody at the controls. The same is true for the Scarborough RT, where the computer drives the trains, while the human looks on, opening and closing the doors, and available to take over in case something goes wrong.

It’s possible that automatic train control could allow the TTC to reduce train crews from two individuals (driver and guard) to one (the driver, taking on the guard’s duties), but anybody picturing the TTC’s subway trains operating with nobody at the helm should adjust their picture.

But the biggest fallacy might be that automatic train control could dramatically increase the capacity of the subway lines by allowing the TTC to run trains closer together. Automatic train control would allow trains to operate as frequently as 90 seconds apart, and it would reduce the likelihood of human-generated delays, but the bottlenecks remains the terminals. The layout of the crossovers at Finch and Downsview stations in particular means that, even with computers at the wheel, the fastest practical turnaround time for trains to pull into the platform, drop off the passengers, switch ends, and pull out, is 135 seconds — very close to the current rush hour frequencies of 140 seconds.

There is no easy fix to this. The provincial government looked at the possibility of operating the Yonge line at 90 second frequencies back in 1989, and the solution they came up with was to extend the Yonge subway west along the Hydro right-of-way corridor north of Finch Avenue, and down Dufferin Street to join the Spadina subway at Wilson. By eliminating the terminal stations entirely and creating a gigantic belt line, the subway trains would never need to switch ends, and could operate more frequently. This plan morphed into a Steeles loop, to provide service to York University, before being abandoned when proposals to extend service into Vaughan materialized.

Another way of getting around the terminal bottleneck is to short turn service. During the morning rush hours, every second train short turns at St. Clair West station while the others take the full run to Downsview. This means that trains at Downsview and at St. Clair West have 280 seconds to turn their trains around. It’s easy to imagine bringing this time down, to 180 seconds each, producing a combined frequency south of St. Clair West station of 90 seconds, but the problem comes when you try to do this to the Yonge line. Ridership along the North Yonge extension demands full service to Finch, so short turning cars at Eglinton (the only suitable spot because the pocket track north of the station is best suited for such short turns) means reducing service between Eglinton and Finch from every 140 seconds to every three minutes.

This operation only becomes feasible if the Yonge subway is extended north, to Clark Boulevard as the TTC proposed in the Rapid Transit Expansion Study, allowing the TTC to short turn every second train at Finch, but this increases the cost of Moscoe’s “bargain” from $750 million to close to $2 billion, though it might still be a worthwhile investment if the rest of the system and the surface network didn’t desperately need attention.

Steve Munro goes further in taking Moscoe to task for his pie-in-the-sky prognostications. There are plenty of benefits to be had in automatic train control, but we should keep our expectations realistic.