Updated April 15, 2018 at 6:00 pm: Comments about projected demand at Park Lawn Station have been added at the end of this article.
In a previous article, I reviewed the Metrolinx technical report on the performance of proposed new GO and SmartTrack stations as part of their overall network. At the time, there was some debate about the validity of the report’s analysis.
Metrolinx has now produced a backgrounder to this report which gives greater details about their methodology and results.
This information is interesting not just in its own right as part of GO’s planning, but also in its implications for the City of Toronto’s expectations for GO/SmartTrack service. The service levels listed in the City’s report date from a Metrolinx plan approved by their board in June 2016. The levels shown in the backgrounder are different, and reflect the change to a mix of local and express trains in the GO corridors. The backgrounder takes pains to emphasize that the service plan is not definitive, but the express/local mix of trains is an essential part of GO’s strategy as approved at a recent Board meeting.
The report begins with an introduction common to such documents laying the basic process for “business analysis” of new proposals. This is summarized in the following diagram. The model focuses on a few key factors:
- The degree to which riders are lost from GO because the addition of stops reduces the competitiveness of GO travel versus driving.
- The degree to which riders shift to a new station thereby reducing their travel time.
- The number of new riders who previously drove and are enticed onto transit by the new station.
This scheme underpins recent changes in the planning for services notably through the benefits conferred by a combination of express services (avoided delays from new stations) and level boarding (reduced station dwell times generally). The technical details of “level boarding” have yet to be revealed, but the analysis assumes a benefit through the elimination of the step between platforms and the interior of trains.
The benefits of electrification in reducing overall travel times and allowing for more closely spaced stops are not mentioned at all, and travel time comparisons are based on an electrified service as a starting point. Metrolinx has effectively discarded one of the arguments they used in advocating electrification in the first place.
The creation of a service plan within a demand model involves related components:
[P]lanning activities are inherently related and must be jointly considered:
- Ridership depends on provided rail service levels
- Service planning and fleet capacity requirements depend on ridership
- Infrastructure requirements depend on the service levels that need to be achieved
Train services are delivered by optimizing train fleet, infrastructure capability, and service timetable planning decisions. Each element can be adjusted independently to yield a different level of train service. There is not a single answer and these factors need to be optimized together in order to achieve the best solution. [p. 7]
The service which would actually be operated is something of a moving target, but in order to model the system, one must create a pro forma plan. This immediately gets Metrolinx into trouble because SmartTrack is not simply an upgrade of GO’s service, but a very political entity with already-promised levels of service and the associated capacity for new ridership. The expected service levels have already been incorporated in the City of Toronto’s proposed agreement with Metrolinx, and there is a mismatch for the eastern part of the SmartTrack corridor. This, in turn, affects the claimed level of service in Scarborough versus existing transit services and the planned Scarborough Subway.
The express-local concept used in the model varies for three corridors:
Barrie corridor: Outer service stopping at all stations between Allandale Waterfront and Aurora; trains will also stop at Downsview Park and Spadina stations, otherwise, express to Union Station. Inner services will serve all stations between Union Station and Aurora.
Stouffville corridor: All-stop peak direction outer service between Lincolnville and Unionville stations; trains will also stop at Kennedy and East Harbour stations, otherwise, express to Union Station. Inner services will stop at all stations between Unionville and Union Station.
Lakeshore West corridor: Alternating trains with bi-directional 15-minute service on the corridor with stops at Mimico and Park Lawn stations. Mimico and Park Lawn stations would therefore receive 30 minutes service inbound and outbound all day. [pp 7-8]
This scheme produces varying service levels at the GO and SmartTrack stations, but these are not supposed to be definitive.
The modelling assumptions in Table 1 of the Appendix do not represent a service plan. The full service plan for GO Expansion will be defined by bidders as part of the procurement process, described below.
On April 3, 2018, Infrastructure Ontario and Metrolinx also issued a Request for Qualifications for the GO Expansion project2. The RFQ is the first step in the procurement process to select a team to deliver GO Expansion, including design, build, finance, operate, and maintain. The project is being delivered as a Design Build Finance Operate Maintain (DBFOM) contract using Infrastructure Ontario’s Alternative Financing and Procurement (AFP) model. A key feature of this choice of procurement method is that bidders will be responsible for planning, designing, building, and operating the train services that are required to meet Metrolinx defined output requirements.
Train service requirements are expected to be defined in the Request for Proposals (RFP) that will set out the train service needs, run times, and capacities that are to be achieved through the concession period. Bidders will be responsible for proposing the infrastructure that needs to be built, the type and number of trains, and develop detailed train service timetables that meet the mandated requirements. This will constitute the full service plan for the GO Expansion program. Over the life of the DBFOM contract, Metrolinx will continue to update the service output requirements based on emerging demand patterns, policy decisions, and customer requirements. The successful proponent will be expected to continue to respond to these changes through further optimizations to the overall fleet, infrastructure, and timetable solution. [p 8]
In other words, we really don’t know what service will be offered to “meet the mandated requirements”, or even what those requirements are. A basic problem faced by any bidder is that the infrastructure is already in place, or at least designed, based on the service plan GO had intended a few years ago. Moreover, earlier reports were quite clear in rejecting express services because of the infrastructure requirements.
The table below compares the service levels used in this report with those in the City of Toronto’s SmartTrack report.
|Corridor||Stations||Metrolinx Trains/Hr||City Trains/Hr|
|Stouffville||Milliken, Finch, Agincourt, Lawrence East||4||7|
|Lakeshore East||Main, Gerrard/Carlaw||8||11|
|Weston||Liberty Village, Bloor, St. Clair, Weston||6||6|
“Fare integration” is also assumed, but for purposes of modelling, this does not mean GO/ST services as a part of the existing TTC tariff, but only that the fare to switch between the two networks would include no barrier, no disincentive. Whether TTC base fares would have to rise to make this possible is not part of the discussion. Basically, the fare component of trip choices within the model is simply “turned off”. This, of course, does not match the current arrangement with co-fares and discounts.
The modelled fare integration scenario represents a future state where customers move seamlessly between services and do not need to worry about the logo or colour of the bus or train that gets them to where they need to go. This was achieved, conceptually, by modelling a scenario where there is no difference in cost to take the TTC, GO, or use both. A specific dollar amount or fare structure assumption was not necessary to support the modelling scenario that was adopted to represent Fare Integration. The analysis removed the fare barrier by giving forecast TTC transit riders in Toronto the ability to choose GO if it is a more convenient, faster option for all or part of their trip. No additional modal shift to transit from auto or walk was considered in the analysis. [p. 9]
For the purpose of the model, a base run without any new stations was done to establish the projected demands on the network for 2031. Separate runs were done each with one new station to determine the change from the base numbers. This established the new ridership attracted by the station, the shift in ridership between stations, and the auto/transit shifts due to the attraction or disincentive of changes in travel times. The to-and-fro of this process is summarized here:
Although the new station provides a new access opportunity, it increases travel times for upstream riders that now have to stop at the station. The delay associated with stopping and travelling through a new station is small for most longer-distance trips (e.g. a 1.5 minute delay is less than a 5% increase in travel time for a 40 minute trip). Although this is a small increase in travel time, a small number of riders may consider switching to alternate modes in response to the delay, primarily switching to auto or other transit modes. The number of users that would consider switching largely depends on whether alternative modes provide competitive travel times relative to GO, which can vary significantly by station location. For example, although the auto network is congested in peak periods, the TTC subway can be a competitive option for travel to downtown for trips that start in southern York Region either via walk/transit access or Park-and-Ride (PNR). [p. 12]
Park-and-Ride remains an important part of the Metrolinx model because it is the only way that many would-be riders can actually reach a station. This produces intriguing effects as we will see shortly.
The ridership projections are shown in two tables which are consolidated for convenience below.
- The set of numbers on the left is taken from Table 5 [p. 20] which gives the projected boardings (ons) and alightings (offs) for each station, as well as the all day total. The AM peak total shown here is simply the sum of the ons and off. The ratio shows the factor Metrolinx used to scale up modelled AM peak numbers to all day numbers.
- The set of numbers on the right is derived from Table 4 [p. 19] which shows the gross count of new GO trips, and the net value when riders lost to other travel are removed. The “Diverted” riders are those who formerly used another station (i.e. they are existing GO users) but shifted to the new stations. (This is the total riders less the new riders.) Finally, the “Lost” riders shows the difference between the gross and net new riders values.
In all of the numbers above, remember that “All Day” figures include the return trips and so the number of people represented is roughly half of the number of trips shown.
A few stations, East Harbour and Spadina/Front, are destination stations serving major employment areas. They originate fewer AM peak trips, proportionately, than stations that draw from residential areas. King/Liberty is half-and-half being both a major residential and job node.
The daily:peak ratio reflects the use of different factors to scale up the AM peak model numbers based on anticipated ridership patterns. GO Transit is primarily a peak period service, and so relatively little demand occurs in the off-peak. However, a higher factor reflecting TTC patterns was used for stations within Toronto.
For most stations an AM peak to daily factor of 2.8-2.9 has been adopted which reflects existing patterns on the GO rail network, specifically the Lakeshore corridors which currently have 30-minute all-day service.
Given that the Fare Integration scenario represents a significant shift away from the status quo to a future where travel between TTC and GO is seamless and where GO Expansion provides frequent service in Toronto, it is reasonable to expected that the future looking GO system will start to behave more like the TTC subway system within Toronto. In light of these considerations, an AM peak period to daily factor of 3.85 was adopted under Fare Integration for Toronto stations. This factor was derived from system wide figures in the 2015 TTC Subway Platform counts. [pp 14-15]
Whether this assumption is uniformly valid depends on the nature of each station’s catchment area and the type of traffic it will attract. It is not clear that a service operating every 15 minutes in the off peak will be able to attract a TTC level of off peak demand relative to the peak, and so the all day numbers could be overstated here if the actual performance is more in the GO Transit range.
Some of the numbers here deserve much greater scrutinty:
- The Lawrence/Kennedy station shows an all-day demand of 9,200 riders compared with existing SRT usage of 7,850 [see TTC station usage counts for 2016]. Given that the bus-to-train connection will be considerably less convenient for the new station, and that the SmartTrack trains will stop only 4 times/hour, it is extremely difficult to believe that today’s level of demand will be sustained if riders have the option of taking a bus to a subway station.
- Kirby station, one that had originally been discarded by Metrolinx as not worth the effort, now is the origin of more trips than any other new station. The number of 3,800 is simply not credible. The preliminary business case states that two thirds of this demand is expected to be Park-and-Ride, and this would mean about 2,500 riders would arrive by car and park for their morning commute. This would require a massive new parking facility for about 2,000 cars. In other words, Kirby doesn’t work as a station unless it is force-fed by pulling car commuters (many of whom now use adjacent stations) to this site. This is a classic planning gerrymander.
The report does not give any information about the forecast demand at existing stations notably Unionville, Milliken, Agincourt, Kennedy, Main, Union, Bloor and Weston all of which lie on the SmartTrack corridor. Similarly there are no data for the GO stations beyond the ST area that would be affected by new service patterns and additional nearby stations.
Moreover, there is no projection of on train accumulated demand to verify that the service plan and the projected ridership actually fit together.
I intend to request the ridership projection details for all stations from Metrolinx.
As an indication of the regard in which I am held at Metrolinx, a recent email from a senior staffer (whose name I will not disclose) stated:
A technical briefing doc will be shared in the coming weeks that you’ll be able to trash thoroughly.
I do not intend to “trash” any document, but to provide analysis and criticism, something to which Metrolinx is particularly averse, especially when politically sensitive issues are at stake.
Updated April 15, 2018 at 6:00 pm:
Park Lawn Station Demand
Metrolinx has been quite adamant in its opposition to a station at Park Lawn, and has relented only to the degree that a split service with Mimico, each station getting a train every 30 minutes, is now proposed.
The demand projection for Park Lawn suggests that it will be quite popular.
- New GO riders (ons+offs) projected: 2,000 gross, 300 net
- Total daily demand: 10,000
The fact that the net number is quite low indicated that most of the new riders attracted to Park Lawn would have used Mimico absent a Park Lawn station. Because Metrolinx has not published a projection for the Mimico station, we do not know how much of its demand would shift east to Park Lawn.
The total daily demand at Park Lawn (10,000) is roughly the same number as the infamous Kirby station (10,600) which gets by only by dint of a massive parking facility. Similarly, Lawrence East (9,200) rates a station, and like Kirby would see 4 trains/hour. Park Lawn gets only 2.
There are riders today at Park Lawn and at Lawrence East while Kirby is an empty field. It is difficult to accept that Kirby makes sense and Park Lawn does not, but then I’m not the Minister of Transportation plunking down a station where it suits me.