Metrolinx has published a study of the proposed subway extension to Richmond Hill updating a Benefits Case Analysis done in 2009. The new report is dated May 2013, but it has only recently been publicly released.
Background information in the study gives an indication of the demand challenges facing the transit network in coming decades. The study itself shows many of the shortcomings of Metrolinx analyses in the selective use of information and limited scope of alternatives comparison.
The study looks at four options for the Richmond Hill line:
- A Base Case assuming substantial additions to existing subway capacity, leaving things as they are with buses serving the existing terminal at Finch Station.
- Option 1: Full subway extension to Richmond Hill Centre close to the existing GO station.
- Option 2: A two-stop subway extension to Steeles with buses serving the area beyond.
- Option 2A: A Steeles subway extension accompanied by improved GO service on the Richmond Hill corridor.
Notable by its absence is an option of both a full subway line to Richmond Hill and improved GO service or any analysis of how demand would divide between the two routes.
The study notes that the Metrolinx Board, in response to earlier analyses, requested additional information:
- Possible adjustments in project scope, timing or phasing;
- Consideration of the extent to which improved service levels on the parallel GO Richmond Hill rail corridor off-loads some of the demand on the Yonge Street subway; and
- The cost impacts of the various options on the subway yards strategy, Yonge-Bloor subway station improvements, and a future Downtown Relief Line to bypass the Yonge-Bloor congestion pinch point. [Par. 1.12, page 3]
The 2013 report does not address these requests because it does not include any option where both the subway and improved GO service operate to Richmond Hill. Although parallel studies (such as the TTC’s own subway yards needs analysis) do look at some aspects of the third point above, this information is not integrated into the analysis, nor is there any review of configurations that could avoid some of the cost of increased subway capacity. This should follow in the Metrolinx study now underway of the Relief Line and associated alternatives, but that sort of network-based review is years overdue.
Chapter 2 of the BCA reviews the rationale for the project. Without question, Yonge Street both in Toronto and in York Region will be a major spine for development, and it is already a strong demand corridor, a status it has held for much of the last century. This demand will be reinforced by development along the path of any rapid transit line.
The 2006 AM peak demand from York Region to downtown Toronto was 20,100 split 55% to GO and 45% to the TTC. This reflects the comparatively low level of GO service crossing Steeles Avenue on all corridors and the volume of passengers fed into Finch Station by regional bus lines (and to a lesser extent to other subway terminals). [Par 2.35]
A demand forecast from 2011 predicts that by 2031, there would be over 20,000 passengers boarding the subway extension for southbound trips in the AM peak period. These are passengers boarding north of Finch and the count does not include additional riders at Finch itself (although demand there would be much lower with many feeder bus routes shifted to points north). [Par 2.36] Ridership along the corridor north of Finch could grow by 70-90% by 2031 relative to 2006. [Par 2.38]
Demand southbound at Wellesley is now 28,000 per hour, 2,000 above the nominal capacity of the subway (calculated at 1,000 passengers per train on 25.7 trains/hour, the currently scheduled service). Reading the projections, one must be careful with figures quoted for the AM peak (which depending on the study may be two or three hours long) and with figures for the AM peak hour. As a rough rule, 50% of the projected demand for a three-hour peak will occur within the peak hour, but this can be affected by factors such as capacity constraints (e.g. Yonge subway) and train schedules (GO).
The demand at Wellesley will be affected by other lines feeding into Yonge Street such as the Eglinton line and a possible Scarborough Subway, offset by whatever travel can be shifted to a regional rail network and/or the Relief Line.
The current signal system can support the 140 second (2’20”) headway now operated and, in parts of the line, the headway can fall lower with the strategic insertion of trains such as now occurs southbound at Davisville and northbound at Union in the AM and PM peaks respectively. However, fundamental constraints at terminals sets a lower bound on turnaround times and the number of trains/hour that a terminal such as Finch can handle.
Over 20 years ago, a proposed “loop” subway initially via Sheppard (when that was seen as the northern terminus of the Yonge and Spadina lines), then via Steeles (to a future York U extension), was proposed to eliminate terminal bottlenecks. This would allow the headway to be reduced with the offsetting penalty that many trains would run nearly empty across the “top” of the loop between the two arms of the YUS.
This scheme was dropped because of its potential capital and operating costs, because it would interfere with northerly extensions into York Region, but most importantly because the recession of the early 1990s sliced 20% off of TTC demand and the then-critical capacity issues on the YUS vanished. More recently, very aggressive claims have been made for potential subway capacity in the range of 50k/hour which would require closer headways and greater train capacity than are planned even with the new signal system on YUS.
That recession contributed directly to Toronto’s ability to focus growth on the suburbs while pretending there was no problem with downtown capacity. The riders are back now, but we have forgotten just how close we came to completely filling the subway years ago.
Without the loop, an alternate strategy to deal with terminals is required. This involves short-turning half of the subway service at Finch on the Yonge line and at Downsview on the Spadina line. With this arrangement, if core headways drop to 105 seconds (1’45”), the service to Vaughan and to Richmond Hill would be 3’30”, or about 2/3 of the capacity now operated to Finch Station (at 2’20”).
The combined effect of more trains (34.3/hour) and the 10% increase in passenger space with the new “TR” units would bring the capacity of the YUS up to about 38,000/hour. This is still far below the claims once made by the TTC to justify deferral of a Relief Line, a tactic that has left Toronto dangerously short of rapid transit capacity.
Peak hour demand south of Finch is forecast to rise to 25,000 and just over 44,000 at Bloor-Yonge if Option 1, the Richmond Hill subway, is built. [Par. 4.26, page 34] This exceeds the nominal capacity of 38,000. The actual peak depends in part on the degree of dispersion of peak period demand around the peak hour, but in any case, the Yonge subway would be beyond its capacity by 2031 even if all of the enhancements included in the “Base Case” (see below) occur before then.
Metrolinx assumes that the carrying capacity per car can be increased to 220 from the 180 figure now used for the TR trains. This is a dangerous assumption because it requires that more densely packing of passengers is sustained for the peak hour (with presumably even higher densities during the “super peak” or after any delays). This is an irresponsible position and it does not align with good transit planning. Moreover, Metrolinx is effectively telling riders that after spending billions on subway improvements, they will be even more tightly packed than they are today. [See Par 4.27 and footnotes 19 and 20 on page 35.]
A more likely approach would be the “seventh car” scheme which the TTC has proposed with an extra short car (taking the train length out to the full platform length) inserted into the “TR” consists. Although technically possible, this would have knock-on effects on storage locations and maintenance facilities around the system.
Shorter headways require a new signal system, and the first phase (Eglinton to St. George) is now being installed. The full line will not be completed until 2018. The need for this upgrade went unrecognized (or at least unacknowledged) long enough that it became an “add on” to the Spadina-Vaughan project and was not included in its base budget. The 2018 implementation date allows construction and commissioning of the Vaughan extension to occur without conflict over a signalling contract that was not included in the planned work.
Major work is required elsewhere on the YUS to handle the additional pedestrian traffic at busy stations. Bloor-Yonge and St. George are heavy transfer points, but there are other stations where passenger volumes have difficulty clearing the platform between trains (especially if one path is blocked by, say, an out of service escalator). Additional entrances/exits are essential that are truly useful to many passengers, not merely a convenience for a minority of travellers (e.g. Yonge Station, west end). During the PM peak at shorter headways, passengers will arrive at the BD line in greater numbers than they do today unless these transfer trips are diverted to other routes, but even with additional platform capacity (itself a challenge) more trains will be needed on BD.
None of the cost of these changes is included in the Benefits Case Analysis even though at least part of them flow directly from adding to rapid transit capacity north of Finch Station.
This level of analysis will come in the Relief Line study now underway, but its absence here detracts from the credibility of this limited “Benefits Case Analysis”.
The Richmond Hill GO Service
Various studies of the Richmond Hill line have been performed including several for GO as a diesel-based upgrade, the GO electrification study, the GO Rail BCA and this BCA. However, the level of peak service projected is different in each case varying from 30 minutes (6 trains inbound during a 3-hour peak period) down to 10 minutes.
This variation has significant effects on projected ridership and operating cost, although a good chunk of the capital cost (double-tracking the line from Union to Richmond Hill) would be common to each scenario. The GO Rail BCA shows a positive value (i.e. costs less than benefits) for better Richmond Hill service at a comparatively modest level. By contrast, the subway BCA downplays the value of better GO service in conjunction with a subway only to Steeles (option 2A) because, as a package, the “benefit” is not outweighed by the cost. This is only one example of how the assumptions in an analysis can skew the results.
The subway BCA does not examine the degree to which a 10-minute GO service (with a capacity in excess of 10K/hour) could provide relief on the subway system. That relief has a value, but it is not considered here.
The subway BCA does note:
As a parallel route, the GO Richmond Hill line is a potential option for providing some off-load transit capacity for the Yonge corridor north of Steeles. At this time however, with the current fare structure, the GO service may not divert a significant number of riders from the subway. [Par 2.64]
This raises another point. The degree to which projected demand on the subway is a function of having a TTC fare all the way to Richmond Hill is another important issue in network planning. However, GO fiercely defends its high farebox cost recovery ratio while ignoring that this produces offsetting costs for the TTC to handle trips that might otherwise be on the commuter rail network. The attitude might be justified (or at least understood) when talking of inside-416 travel, but Richmond Hill is definitely GO territory. If a cheaper GO fare shifts riders off of the subway, shouldn’t this be considered as an “investment” in alternative capacity?
The Base Case
The “base case” presumes that considerable investment will be made in the YUS including:
- Spadina extension to Vaughan;
- ATO implementation across entire YUS line;
- Full roll-out of Rocket Trains on YUS line;
- Service frequency of 105 second headways;
- Short turn at Wilson Station;
- Rail Yard provision as required; and
- Yonge-Bloor station investment. [Par 3.8, page 24]
Several of these items are currently unfunded, and there is no indication of when or if they might actually occur.
The TTC’s fleet plans include 420 “TR” cars, or 70 trains. The service now operated requires:
- AM Peak: 44 trains on a 2’21” headway with a scheduled short turn at St. Clair West plus 4 gap trains.
- PM Peak: 47 trains on a 2’31” headway with no scheduled short turn plus 2 gap trains.
Wilson Station does not have provision for a short-turn operation, but this has been included north of Downsview Station. Running all AM peak service to Downsview will require 6 more trains based on the current schedule. (The AM peak service uses 25 trains for the Finch-Downsview service. Hence 50 trains would be needed if the short-turn were eliminated.)
The Vaughan extension at 8.6km represents just under 30 minutes of running time at 36km/h (a commonly-cited average speed for subway lines with similar station spacing). On a headway of 280 seconds (double the existing core headway), this would translate to about 12 trains.
The combined total fleet required for the Vaughan extension at the existing service level is, therefore, about 62 trains, a value this is tight for spares by current TTC standards. (This will likely be achieved by having some gap trains do double duty as spares.)
Reducing the headway to 105 seconds from 140 requires 1/3 more trains, or about 23 more than the 70-train fleet, not to mention additional carhouse space and operating costs. Moreover, operation of such a headway at Finch terminal would be challenging, to say the least.
At Bloor-Yonge Station, the cost of improvement could be over $300m [Par. 3.9] not to mention changes needed at St. George, additional entrances at now-constrained stations and the effect of transfer passengers arriving at a higher rate at interchanges with the BD subway.
All of these costs are presumed to have been borne before the BCA even starts to consider the cost and benefit of extending the subway to Richmond Hill even though at least part of them are a direct reaction to the demand that extension will trigger. This results in understatement of the true cost of the extension.
Option 1: A Subway to Richmond Hill Centre
The subway would be extended 7km with 5 or 6 stations from Finch to Richmond Hill. Service would operate every 105 seconds (1’45”) to Finch and every 210 seconds (3’30”) beyond. At an average speed of 35km/h, the one-way trip would be 12′ long. It would add 24′ to the round trip time on the line, and at a 3’30” headway this would require 7 trains plus a spare. (The BCA allows for 12 trains.) Storage will be provided in an underground yard north of Richmond Hill Station. [Par. 3.12-16]
Option 2: A Subway to Steeles Avenue
This option would build only the first two stations on the extension (at Cummer and Steeles) with the Richmond Hill segment left for a future project. Bus services now terminating at Finch Station would be cut back mainly to Steeles Station substantially reducing bus and passenger volumes at Finch and on Yonge south of Steeles, but requiring a large terminal at Steeles replacing much of the capacity now at Finch.
Subway service levels would be as in Option 1 with half of the trains turning back at Finch (an essential requirement to allowing Steeles terminal to operate with a conventional nearside crossover as Finch does today). The extra round trip time would be slightly above 4 minutes, not a major addition to the overall length of the YUS.
Although not mentioned in the BCA, there would likely be an extension of the three-track section north of Finch to Cummer Station to provide additional train storage at the north end of the Yonge line.
Option 2A: A Subway to Steeles Avenue Plus Improved GO Service to Richmond Hill
This option is identical to Option 2 for subway service, but adds a 10-minute service on the GO rail corridor to Richmond Hill in place of the 30-minute service presumed for Options 1 and 2.
Comparison of the Options
The Metrolinx BCAs use a “multiple account evaluation” wherein the options are compared against various possible benefits and costs to arrive at a ratio supposedly expressing the net worth of the capital and future operating investment.
Travel Time Savings
“Time is Money” as they say, and the BCA assigns a value to time saved by travel on each of the options. The value of a person hour is set at $14.24, a weighted average of $38.45 for business travel and $11.83 for everything else. [See Appendix A, Assumptions]
Clearly, the majority of trips are going to be “non business”. Obviously, the option that maximizes the change in trip speed for the most users will generate the highest benefit, and Option 1 easily wins on this basis. [Table 4.1, p. 32] However, the lion’s share of the benefits flow to existing transit users, not to new users of the network. For Option 1, about 15% of the benefit flows to new riders, while for Options 2 and 2A, the values are 10% and 8.6% respectively. A substantial benefit accrues to motorists through supposed reduced congestion, and this is further compounded by a presumed reduction in accident rates.
This is hard to credit if most of the transit benefit is for existing, not new transit riders. This also begs the question of the degree to which the extension would divert auto trips to transit rather than simply improving service for current users (a laudable goal, but not the primary intent of the Metrolinx “Big Move”). Moreover, if the majority of the benefit flows to existing riders, where does all of the new demand southbound at Wellesley come from? Something is out of whack here in the calculations.
Other Financial Considerations
Figure 4.1 on page 35 is intriguing because it shows the uneven demand at stations. Most of the riders on the extended subway will arrive by feeder bus with the demand being concentrated at Richmond Hill, Steeles and Finch Stations. This begs the question of whether the oft-cited link between development and rapid transit is the fundamental factor in generating ridership here. There is an existing population and demand for travel to central Toronto, but most of it will not walk into the new subway stations.
The incremental forecast annual revenue for the subway is $8.4m [Par 4.28, page 35], but the cost of its operation will be $14.8m [Par 4.33, page 36] for a cost recovery ratio of 57%. However, that calculation assigns all of the fare revenue to the extended subway and ignores the cost of providing capacity for the new riders elsewhere in the system. This is a misleadingly rosy portrayal of the economics. Of course, a good chunk of the operating cost goes to serving existing riders (who contribute no new revenue) by providing them with a faster, more convenient trip north of Finch Station. The cost recovery on their share is zero. There is a benefit in reduced operating cost to York Region for bus service (and in York Region fares for riders who can walk to the subway), but this does not accrue to the TTC.
For Option 2A, there is a considerable net loss due to the provision of more in GO service to Richmond Hill. The BCA does not state the additional fare revenue GO will obtain, nor does it give an estimate of GO ridership. The benefits diversion of trips to GO might provide are similarly not included. This is a fundamental misstatement of the value of Option 2A, and added to the absence of an Option “1A” (full subway plus GO) might make one think that GO improvements are deliberately downplayed here to improve the case for the subway alone.
From this point onward, any attempt to quantify the costs and benefits is meaningless because of errors and omissions in the methodology. I cannot, however, ignore the “economic benefits during construction” component of the analysis.
In brief, the premise is that every $1b of investment in infrastructure generates jobs either directly in construction or indirectly through supplies, and the project has a lasting economic impact in the jobs needed to operate and maintain the line. Self-evidently, the more expensive the project, the greater the economic benefit. However, the type of jobs (e.g. what social segments benefit) and the geographic distribution (little of the high tech components of control systems and vehicles comes from Ontario or even from Canada) are not considered in comparing the effects of capital spending.
What is missing is an analysis of how else the money might be spent to other effect either on alternative transit projects and services (both on the capital and operating account), or elsewhere in society as a whole. One could build a different line, or simply leave the money in unmarked small bills in strategic locations around the region.
Such an analysis is fundamental to the comparison of various transit projects, but it is completely missing here and in all other Metrolinx BCAs. Capital intensive projects may help the construction industry, but they do not necessarily help the transit system — that is the heart of the debate over subway proposals for Scarborough today.
Every proposal has its potential to increase land values and support regional growth in a transit-friendly way. This of course assumes that development will actually occur around transit lines and increase demand for them, and indeed that municipalities will force such development to occur and leave empty fields far from transit. This is rarely what happens because most development is car-oriented.
Without question, building a subway and other municipal infrastructure makes land worth more, and that is a public investment that will be recouped by private landowners. Without question some of that profit should flow back to public coffers through Development Charges and future taxes. However, there is typically quite a lag time between provision of the infrastructure and actual development, not to mention an unwillingness from the development industry to actually pay toward what is provided to make their lands more valuable.
It is important when looking at the potential of various schemes to raise land values to remember that the up front public investment has a cost. Moreover, the Development Charges regime (provincial legislation) demands that only the share of costs due to new development go into the calculation of DCs. The value to existing riders cannot be included even though the investment might not have happened without hope of the future development.
The Metrolinx Benefits Case Analysis for the Richmond Hill Subway is fundamentally deficient in that it does not review an option for both a subway extension and substantially improved GO service. Moreover, the BCA assumes a potential capacity for the subway system in excess of what realistically can be provided at reasonable levels of comfort and attractiveness, and it presumes that substantial investment in this capacity will occur separate from the Richmond Hill extension.
There is little question that a subway to Richmond Hill will be built some day, not just for political reasons, but because Yonge Street is a long-establish travel corridor. However, more is needed than just this line, and as a region we cannot enter into the project without understanding its implications for the network as a whole. Metrolinx does us a disservice by failing to present these implications and costs for a target year, 2031, that is not that far off in terms of major network planning and construction.
We ignored the need for more capacity into downtown over two decades ago, and now we must pay the price with subway and GO Transit expansion.