This installment completes Chapter 3 of the study with the evaluation of alternative signalling strategies. The recommented alternative is Automatic Train Control, no surprise there, based on the premise that it provides the maximum benefit versus the expenditure. Underlying this, however, is the goal of a 90-second headway and the increasing challenges to subway operations as the headway drops. ATC is treated as a means to achieve this dubious goal rather than a worthwhile move in its own right.
Chapter 3.4 Evaluation and Recommendations starts with a set of charts attempting to show a “cost effectiveness” value for the various options.
Exhibits 3.3.8 and 3.3.9 show the minimum headway and increased capacity for each option. These are, of course, an inverse relationship because shorter headways mean more capacity.
Exhibits 3.3.10 and 3.3.11 show the estimated capital cost and implementation periods for the four options. Note that the costs include neither include either the additional vehicles nor the station modifications needed at Bloor-Yonge or Finch to support the schemes that depend on reduced dwell time at this location. (See footnote 8 in the text.)
Exhibits 3.3.12 and 3.3.13 purport to show the “cost effectiveness” and “productivity” of the options.
Exhibit 3.3.14 summarizes the information for the four options. Note that this is a $708-million project in 1988 dollars before we even start on the infrastructure changes needed to make short headways possible.
Exhibits 3.3.15 details the operating implications of each option both while under construction and after implementation.
Cost effectiveness is measured as a ratio between the capacity increase and capital cost, but the absence of infrastructure and vehicle costs renders 3.3.12 questionable at best. The dip in the line for option 1C, extensive changes to the existing signal system, is cited in the text as showing that it should not be pursued because the “cost effectiveness” value does not go up as much as with the other options. Alas, we do not have a chart showing the same ratios with missing components included.
Productivity is measured as a ratio between total capacity and fleet size. Again option 1C fares relatively badly because it requires slower line operation and a larger fleet to provide a given capacity. In effect, this chart is a measure of line speed because speed determines the fleet size needed for a given capacity. Common to all options is a drop in line speed because trains are so close together.
The “cost effectiveness” ratios, were they recalculated with the fleets included, would be:
- 1A: $15.4-million per thousand increase in capacity
- 1B: $27.7-million
- 1C: $53.3-million
- 2: $49.2-million
In the text, we also learn that the increased revenues through riding may not cover the increase operating costs, although a change to one-man train operation could offset some of this. No specifics are given, and I caution readers that the cost of maintaining vehicles is a considerable part of total operating costs, and for subways in general the train crews account for less than one quarter of that total.
The text also cautions about the impact on customer satisfaction both during and after completion of the project. A vital paragraph warns:
The ability to consistently sustain the desired headway must also be considered during the evaluation process. For example, in order to guarantee that 112 second headway can be consistently sustained under all operating conditions, Bloor Station should be reconstructed to achieve 30 second dwells, and the Finch terminal should be modified as indicated under Option IB herein. (However, apart from the high capital cost of reconstructing Finch terminal and the small reduction in headway gained, productivity would be reduced due to train turnaround behind the terminal stations, and a fleet increase would be required. This would suggest that subway extensions to new terminals should be considered as an alternative.) With regard to 90 second headway operation, the fact that several worldwide transit systems which are designed for 90 second headway operation are not actually operated at 90 second headways, suggests that the headway objective under Option 2 may not be sustained under all operating conditions.
This should have been a red flag to everyone. If there is even the possibility that the 90 second headway could not actually be sustained, this calls out for a review of the basic assumption. Notwithstanding the concern, the report goes on to recommend approval in principle of a 90 second design guideline.
Hi Steve, thanks for posting these articles! If Finch Station turn around time is a constraint as pointed out in the report, what is a reasonable forecast for turn around time once ATC is in place on YUS (assuming without Yonge Subway extension)? The train guards on the YUS line are now at the end control cab of the trains, is a major reason for this new placement of train crew to reduce the time needed for turn around? Thirdly, what are some potentially cost effective way(s) to incorporate turn around tail track(s) (and perhaps even some overnight storage tracks) near Finch Station that would also take into account and best accommodate the possibility for future Yonge Subway extension to Richmond Hill? Thank you
Steve: The constraint on turnaround time is the length of time it takes a train to traverse the crossover south of the station. This and other crossovers were designed with long tracks so that the turnouts would not throw passengers around much, but this increased the length of time needed to completely clear the area where train paths will conflict. I wrote about this six years ago in the early days of this blog.
Having operators at the ends of the train can help by eliminating the walk between a guard and operator position which can be the source of some delays, but this can be offset depending on where the washrooms are located at the terminals. Another approach is to use step-back crewing where an incoming train’s crew does not take out the same train, but rather the train one or two behind. In theory, this means that a new crew is ready to take a train out as soon as it arrives (although it will stand on the platform for at least one headway while the train on the other track leaves). This can get tricky when trains are off schedule and some of them are short-turned.
One solution proposed for Finch is a double pocket track beyond the station. There is already one such track, and a second one would be needed further north. The idea is to eliminate the crossover conflict by treating Finch as a line station and taking the trains into pocket tracks via paths that do not conflict. However, this will require the construction of roughly 700ft of additional tunnel at a minimum, and will add to the number of trains needed to operate the line (because the round trip must now include the extra time spent north of Finch Station). The problem of giving operators breaks at the terminal still remains, and this would definitely require some sort of step-back crewing scheme. (A terminal layover in a tunnel is no use as a crew break.)
If, eventually, the line is extended further north, it could be operated with only half of the service running to the new terminal and half turning back at Finch. That would still require step-back crews at Finch so that operators would have a break. On the University-Spadina side of the line, this is not an issue because half of the service will not run all the way to Vaughan Centre.
Another related problem is caused by variation in running times over a long route. If the actual time needed to make a trip is less than the scheduled time, that extra has to be burned up somewhere. During quiet off-peak periods, this is less of a problem because the headway is wider and two trains can sit at a terminal with a third approaching just as one of the other two leaves. When headways are shorter, there is no place to store the “extra” trains and a queue approaching the terminal forms. Unlike surface routes, scheduled running times on the subway are not extended in the peak in part to avoid this problem. The offsetting issue is that trains routinely run late, and a lot of crew shuffling is done to put the crews (if not the trains) back “on time” by swapping trains at centre platform stations. For example, a crew northbound at Eglinton can swap with a southbound train and, thereby, gain about half an hour. This sort of thing is done on BD too, but with fewer centre-platform stations, this changeover can take a while as crews walk to an exit, cross over via the station mezzanine, and then walk down to and along the opposite platform.
Some but not all of these issues would be eliminated in a totally automated system, but we are years and a lot of capital investment from that sort of arrangement.