Updated February 3: In a previous version of this post, I was using 4-car trains for the Base Case and therefore claimed that the fleet was undersized. This has been corrected.
On January 16, the Metrolinx Board approved release of the Benefits Case Analysis (BCA) for the replacement and extension of the Scarborough RT. This document is now available online.
The original TTC proposal, was simply to replace the current technology with Mark II RT cars on the existing alignment. This is now referred to as the “Base Case”. Four alternatives, all considered superior to the base case, were evaluated in the BCA.
- 1: Extend the RT 5.4 km to Malvern Town Centre using the current technology.
- 2: Extend the RT to Markham & Sheppard where it would connect with the Sheppard East LRT including an LRT branch north to Malvern.
- 3: Replace the RT over its entire length with LRT and extend to Malvern on approximately the same alignment as the first RT option, completely in an exclusive right-of-way.
- 4: Replace the RT with LRT and extend to Malvern with a partially exclusive right-of-way east of McCowan.
Major Flaws in the BCA
To save readers from combing through the rest of the text, here are major points where the analysis does not hold up.
- The peak demand for which the line is designed, 10K/hour, is substantially above the modelled peak demand of 6.4K/hour in The Big Move. The Metrolinx regional plan includes frequent commuter rail services through the extended RT’s catchment area, and this likely attracts some riders away from the RT. Designing for higher capacity than required inflates fleet, carhouse and operating costs. It may also affect train lengths and the cost of retrofitting existing stations.
- LRT options presume the construction of a dedicated carhouse for the Scarborough line even though, by the time an LRT would operate, the Sheppard line’s new carhouse would be operating and could act as the base for Scarborough trains. This inflates the capital cost of LRT options.
- The alleged cost-benefit ratios are highly sensitive to the presumed value of travellers’ time. This value is orders of magnitude greater than the value of environmental effects (reduced car use) and dominates the calculations. The model overall favours proposals that serve long trips at comparatively high speeds (e.g. with widely spaced stations) that may not be conducive to the type of neighbourhood preferred in the Official Plan.
- The values assigned to savings from reduced automobile use are based on a much higher factor than in the VIVA Benefits Case report, 95 vs 23 cents per km (2031). The effect is to grossly overstate the savings from reduced auto usage for all options.
- Economic benefits include the money spent on labour during construction. This value for all options is a disconcertingly low percentage of the total project cost (well under 20%). This shows that a substantial portion of any scheme is consumed by planning and design, materials, vehicles and system component costs. Moreover, the idea that spending more on one project is “good” because it generates more work is valid only if one ignores other projects that could be built with the same money and labour. This will be an important factor when projects are weighed against each other.
- Estimates for the length of time the RT would be closed for upgrade or restructuring are “at least 8 months” for RT and “up to 36 months” for LRT. These figures need to be reviewed in detail to determine where the differences lie. The numbers are taken from the original TTC study (which did not include the qualifiers) when the physical changes needed to handle Mark II RT cars were considered to be trivial. This may no longer be true.
- Overall the analysis looks at the RT in isolation from the surrounding network and ignores alternative ways that the demand might be served on the network, not just on the RT corridor. Although the report shows LRT as the less expensive option, the difference versus RT options may actually be understated.
For the Base Case, the existing fleet would be replaced with 36 new MKII RT cars to handle demands up to 5,400/hr. These cars would operate in 3-car trains. (MKII, or “Mark Two”, is the longer Vancouver version of the RT car.)
Physical changes to the RT infrastructure include:
- platform edge cutbacks to accommodate larger vehicles
- replacement of Kennedy Station with a more convenient connection for passengers and a double-track configuration for better terminal capacity
- various structural upgrades to accommodate the new trains
- upgrades at McCowan Yard for a larger fleet and the physical requirements of the new cars
- upgrade of the ATC system, and possible replacement of some infrastructure that may be at end-of-life
The planned headway for 2031 is 120 seconds, with a fleet of 36 Mark II cars (12 3-car trains). Current service is provided by 6 trains on an alleged peak headway of 210 seconds (3’30”, actually considerably longer given operational problems). Getting this down to 120 seconds would require 10.5 trains just for service. A small saving will accrue from improved terminal operations, but this would only bring us down to 10 trains.
The service plan tells us that the design capacity for a 3-car train is 180 passengers (30 trains/hour gives 5,400 capacity).
Updated February 3: In a previous version of this post, I was using 4-car trains for the Base case and therefore claimed that the fleet was undersized.
In this option, a much larger fleet of 108 Mark II RT cars would operate frequent service (108 seconds) in 22 4-car trains from Malvern to Kennedy. The total fleet would be 27 trains, a rather generous spare ratio compared with today’s single spare.
For this option, two additional major changes are required relative to the base case:
- extension of all existing platforms to handle 4-car trains
- replacement of McCowan carhouse to handle larger fleet
This is the most extensive of the “all RT” options.
This is a much shorter extension of the RT only to Sheppard where it would connect with the planned LRT. A branch of the LRT would provide service to Malvern.
As with Option 1, this scheme requires lengthened platforms on the existing line as well as a new carhouse. Projected demand is lower because of the transfer from LRT to RT an Sheppard, and again from RT to subway at Kennedy.
A fleet of 80 Mark II cars would operate every 120 seconds using 16 4-car trains from Sheppard to Kennedy. The total fleet would be 20 trains including spares.
This is the “all LRT” option. In this version, the line to Malvern is on a completely separate right-of-way and shares no trackage with the Sheppard LRT. This option includes a dedicated carhouse for the RT line even though this could be shared with the Sheppard LRT.
All of the existing infrastructure would require modifications to accommodate LRT vehicles.
Service would be provided on a 139 second headway with 3-car trains. The fleet would be 62 LRVs including spares.
The projected ridership for this option is the same as for Option 1 because the two services are essentially the same.
In this option, the LRT line would run in its own right-of-way in Progress Avenue east of McCowan and turn north into Malvern. The service design is for 2’15” headways west of McCowan carrying comparable demands to those projected for option 3. Beyond McCowan, half of the service would continue to Malvern on a 4’30” headway. The fleet size is projected to be equal to that of Option 3 with this service design.
Depending on the alignment chosen, this option may required shared running with the Sheppard LRT line. Whether this is operationally feasible given the heavy service projected for Sheppard itself is still under study.
As with Option 3, the BCA assumes a dedicated carhouse for the new Scarborough LRT even though it could be shared with the Sheppard LRT at a lower combined cost.
For the Base Case, Metrolinx cites a capital cost of $452-million (2008$) up from the TTC’s estimate of $360-million (2006$). Options 1 and 3 (the “all the way on exclusive right-of-way schemes) have capital costs of $1,612M and $1,404M for RT and LRT respectively. Option 2 (RT to Sheppard only) is less expensive at $1,233M, and does not include the cost of an LRT branch off of the Sheppard line into Malvern. Option 4 (LRT with partially segregated right of way beyond McCowan) is $1,225M.
Cost is not everything, of course, and the BCA looks at a variety of factors to weight the options against each other. As we will see, the analysis has the odd effect of rewarding expensive projects (because they generate more short-term economic stimulus) and projects that attract many new riders who would otherwise drive (thereby causing savings in various environmental measures and reduction in commuting costs).
The peak point demand anticipated for Options 1 and 3 is 10,000 per hour, while that of option 2 is 9,000. Option 4 is a special case because only half of the service would operate north of McCowan Station, although it would still provide 10K/hr over the existing route south to Kennedy.
The BCA is unclear about what new GO services are included in the demand model, and whether the projected RT demand is artificially inflated by the absence of parallel GO routes taking some of the long-haul load. However, the Modelling backgrounder to Metrolinx’ regional plan shows how the network as a whole would behave. The projected demand in that report is 6,400 per peak hour, very substantially below the demands cited in the BCA for the RT, whatever form it might take.
As with other projects studied in isolation from network effects, we may be evaluating a capital investment and service level proposal far in excess of what we would actually see if related network elements were considered as a package. Providing capacity that is not required skews the analysis in favour of higher-cost options, and overstates benefits such as user time savings. With frequent GO service to downtown, the real time savings for many in northeastern Scarborough would accrue to GO, not to the RT.
The model seems more attuned to projects that break new ground for transit system coverage rather than making trips for existing users more convenient. It also double-counts some variables because of their effects on multiple components of the overall evaluation.
In the case of the SRT project, unlike most other BCAs, there is a “base case” that will be undertaken if nothing else is done. Therefore costs associated simply with refitting the existing line with newer RT technology and any spinoff effect this may have such as greater ridership is the starting point for all evaluations. The results would not be the same if we were starting completely from scratch because there would be no existing infrastructure to recycle or modify.
Transportation User Benefits
This measure includes the following components:
- Travel time savings both for transit trips and for road users who will benefit from reduced competition for road space.
- Reduced automobile operating costs for new transit riders who would otherwise drive.
- Reduced accident costs from traffic diversion.
- Enhanced comfort of new, less-crowded vehicles.
Time savings are rated at a nominal value of $13/hour, indexed for inflation, and calculated over the period to 2038. This methodology tends to favour network designs that provide fast-moving service for people travelling long distances because this will attract more car drivers and give each rider more time savings relative to the base case for those who do not live on the existing line. Later, we will see this reinforced by environmental savings from diverting the same cars off of the road.
The value of reduced auto operating costs has serious problems in this analysis. First off, the assigned value for year 2031 has risen from 23 cents/km in the VIVA BCA to 95 cents/km in the SRT BCA. The higher value is cited as “operating and ownership costs”, but saving the entire amount assumes that the total cost of ownership disappears — in effect, that all of the savings by redirected travel produce a corresponding reduction in car ownership. This very substantially overstates the “savings” from redirected travel.
The metric for accident reductions is dubious because roads have a well-known habit of backfilling any diverted demand with new traffic. In any event, this element of the avoided costs is quite small over the life of the study to 2038, about 2.5% of the transportation user benefits.
Although this analysis includes a value for road capacity released by a transit investment, it does not consider whether transit capacity in other corridors might be similarly relieved. For example, any GO services to northeast Scarborough should include their potential for reducing capacity requirements on the RT itself, but this sort of analysis is not on the table because we are only evaluating the RT.
The cost comparison is brewed out of various factors:
- Capital costs discounted to 2008$ from the years they are incurred
- Annual operating costs in 2021 discounted over a 30-year period
- Annual savings in bus operations replaced by RT/LRT similarly discounted
This yields a Net Present Value (NPV) of the capital and operating costs and hence the incremental cost over the Base Case. The highest values, of course, are for the RT and LRT options going all the way to Malvern, although LRT is about 10% lower.
Projected ridership is equal for both Options 1 and 3 because they provide equivalent service to Malvern regardless of the technology. Riding for Option 4 is projected to be higher than for Option 3 because the latter imposes an LRT/RT transfer at Sheppard. As discussed above, all of these projections are substantially higher than the peak demand shown in Metrolinx own network Modelling paper. There is no discussion of origin-destination patterns that could reveal what proportion of riders in the catchment area would be better served by a GO line to downtown versus a local service connecting to the Danforth subway.
In all cases, the projected additional revenue, expressed in NPV terms, is low, a few percent of the NPV of the capital costs, because many riders in this corridor are already on transit as part of the base case. This is an interesting outcome for an area where we would expect there is a considerable market for new transit use.
This calculation produces a number, the sort of thing that people who don’t like to think just adore. However, it’s important to understand how this number works.
We begin with the “transportation user benefits” which, in this case, are almost totally derived from reduced travel times of existing transit riders. These benefits have NPVs ranging from $1.2 to $1.6-billion over the study period, and this is obviously dependent on the value assigned to a rider’s saved time.
Next we have the “incremental cost” relative to the base case, and subtracting one number from the other gives the “net benefit”. The study observes that these numbers are positive for all options, an unusual situation caused by exclusion of the base case costs from the overall scheme.
Finally, we get the benefit-cost ratio which is identical for both options going all the way to Malvern. This value is higher for the other options because they provide less benefit but at even lower cost. In other words, the cost-benefit ratio for a dedicated extension to Malvern does not produce returns at a comparable level to the portion of the line south of Sheppard.
“Comparable” is a difficult word here both because ridership estimates are suspect, and because the process is highly sensitive to the assumed value of a rider’s time and of savings in auto-related costs. The further we build the line, the more time we save for trips at the outer end of that line, but the density of new demand probably falls off even as the cost of construction remains constant for fully exclusive rights-of-way.
As we already know, riding on the RT will primarily come from existing transit users, or more accurately, users who would have been on transit anyhow even if the RT never went past McCowan (the base case).
Therefore the reduction in emissions due to lowered auto use in this evaluation is low, and its NPV is similarly small in relation to the overall project cost.
This measure includes both the spinoff benefits of employment during construction, as well as the permanent improvements through stimulation of development and increased job density on a new route.
Moreover, diversion of trips to transit is presumed to free up income that would otherwise be devoted to operating a car. This is a faulty model because transit often enables people to reach jobs that would be out of their grasp — there would be not offsetting saving in car expenses because there would be no second car. Of course, anything that reduces car purchases and operating costs has an effect elsewhere in the economy as we see all too clearly today.
The figures for direct employment are rather interesting because they are a small part of the total cost of each of the options. Option 1, the all-RT scheme, racks up $250M in wages on a total cost of $1,612M, less than 16%. This shows the substantial costs due to planning and design, materials, vehicles and components for major subsystems such as signalling. These have economic benefits somewhere, but not necessarily in the GTA. We must guard against spending money just for the sake of it, and expensive projects are not necessarily good ones.
As for lasting economic benefits, the employment and income impacts are calculated to be relatively small, roughly 5% of the total capital cost. This is not a major factor in selecting options for the RT.
Finally, we have land value effects (a dubious science in the current market, but one we might expect to be relevant in the long term). The BCA includes an intriguing table based on an extensive study of property values and the effect of nearby transit services. This study shows considerably better effects from rail-based modes than from Bus or BRT, although with some intriguing variations. For example, the more exclusive the right-of-way, the higher the impact on residential prices presumably because of capacity and speed effects. However, LRT holds its own against subways for retail uses because it encourages a stronger pedestrian environment and brings the transit benefit to a larger area with stations closer together than subways.
The effect of this model is to show an advantage for Option 4, partially exclusive LRT, because it has better station access. However, the ranges cited for all of the options are wide, and none of them has a clear advantage.
Social Community Impacts
All options will support land use changes with those for options 1 and 3 (exclusive rights-of-way) being more likely to encourage high density development around widely spaced stations. Option 4 should encourage more pedestrian activity because it is at grade in a semi-exclusive right-of-way like other parts of the planned LRT network.
Health effects accrue both from encouraging people to walk to a transit stop, and from reduction in car use and associated pollution. Schemes that encourage more ridership are credited for generating more walking and less pollution, but the four options are close enough in likely demand that this is not a major factor. Also, as I mentioned earlier, it is unclear whether moving trips off of roads will really reduce pollution or merely free up capacity for more cars.
Accessibility can be seen two ways. Lines with few stops encourage dense development that might be incorporated into the stations making them highly accessible at least to those who live there. Lines with more stops better serve communities along the route. What is not discussed here is the overall population density and the degree to which the neighbourhoods will already be established by the time the line and its extension are completed.
Visual and noise impacts exist for all options in one degree or another. Alignments with elevated structures present a visual intrusion, and alignments running along a former rail corridor past backyards have issues with both visual and noise effects. Although it is not mentioned, it is worth noting the noise effects of the existing RT where it runs through residential areas, some of which are considerably further from the tracks than the proposed extension via the old Canadian Northern right-of-way.
Finally, of course, we have the length of time the existing service would be replaced by buses for construction. This is estimated to be at least 8 months for the RT, and at up to 36 for LRT. When the TTC originally claimed an 8-month project for upgrade to Mark II RT cars, the extent of work needed to upgrade the infrastructure was less than is now acknowledged.
These estimates need to be substantiated and refined to demonstrate why they are so different, and whether the difference between the RT and LRT time is lower than cited in the BCA analysis.
At the end of such an exercise, one hopes that a single option emerges as the clear favourite in most if not all dimensions under consideration. In this case, the RT has benefits in reducing the shutdown time for the line, but at a substantially higher cost. Any hope that the line might go further north as shown on some regional plans would be totally quashed by the high cost of an RT extension north from Malvern, and this is not even mentioned by the BCA.
The report recommends a “wait and see” approach pending the outcome of the Eglinton BCA with the technology selection to be fitted into whatever is appropriate for the longer route. We need to be sure that Metrolinx models Eglinton correctly as part of a network, not just as an isolated line.
The advantage of LRT is that it is not tethered to an exclusive right-of-way, and a network of LRT lines is greater than one project. I will leave this debate for another day.