Making Electrification Work

Electrification with Electric Locomotives and EMUs

We have used GO's own numbers from the technical appendices to the electrification study and its Summary Report to replicate GO's estimates for electric locomotives on the Lakeshore route (Option 3) and to estimate the potential case for a different scenario: electrification with electric locomotives hauling bi-level coaches in the peaks, with shorter EMUs operating every 15 minutes, all day. We considered the Lakeshore route, and "all other routes." Our findings are summarized in the Appendix (Table A1). Our approach to estimating ridership, revenues, and benefits is explained below.

Boosting Ridership and Revenues

Carrying passengers is, of course, GO's primary purpose and increasing ridership should be the main objective of any investment scheme. For some reason, the GO Electrification Study Summary Report does not give an explicit figure for the additional ridership that might be attracted by electrification, with faster and more frequent trains. This would seem to be the most important criterion, alongside cost, in evaluating any scheme.

Table 17, in the back of the Final Report, does indicate a capital cost per new 2031 transit rider of $260,000, on the Lakeshore route, with electric locomotives. Note that this is not the "Net Cost per New Rider," as much or all of the capital cost might be paid back from increased revenues, and operational savings. It does, however, give us a way to estimate the new ridership that GO is expecting. Dividing this number into the capital cost of $1.16 billion from the same table suggests a figure of 4,461 new daily riders, roughly a 4% increase from current ridership, and perhaps a 2.5% increase relative to the base case 2031 ridership.

GO's technical appendix 8A states that estimated demand uplift from electrification is calculated using a journey time elasticity of -0.6. Although the actual equations are a bit more complicated, essentially this means that a 10% reduction in journey time will bring a 6% increase in traffic, all other things being equal. Electric locomotive-hauled trains have faster acceleration, and GO calculates they will shorten the average commute by 2.4 to 2.8 minutes. But elasticity applies to the total trip (which includes the time required to drive to the station, park, wait for the train, take the train, then walk or use the subway to get to the office). Altogether, a typical GO trip takes about an hour, door-to-door. So a 2.4-minute time saving might be about 4% of the total trip (2.4/60 minutes). Applying the elasticity of -0.6 explains why GO says total demand is expected to increase only 2.4%, with electric locomotives.

U.K. railway operators have many decades of experience improving and electrifying rail routes. The general experience, presented in the Passenger Demand Forecasting Handbook (PDFH),[1] is that demand for non-peak (including both off-peak and contra-peak) travel is more time-sensitive, with more journey time elasticity. A 10% reduction in non-peak journey time will, generally, bring a 10% increase in traffic, sometimes even more. Peak traffic is less sensitive to time savings, because the peak market is mostly "captive." GO rail is already much faster than driving, and much cheaper than driving and parking, so a 10% time saving might indeed attract only 1.2% more peak riders. The -0.6 elasticity GO used in the electrification study is, in effect, the all-day average. This did not matter in GO's analysis, because it did not consider strategies using more than one kind of train for serving the different peak and non-peak markets.

To evaluate the case for using electric locomotive in the peaks, and EMUs all-day (because GO adds peak trains only when it needs to), the peak and non-peak markets need to be analyzed separately. We can estimate the current split of peak and non-peak trips on GO as follows. GO says its total ridership is 55 million per year, of which 80% or 44 million is by rail.[2] GO Rail Lakeshore ridership is about 26 million passengers per year. GO operates about 25 trains arriving at Union Station between 7 a.m. and 9 a.m.,[3] most of them full (because GO only adds peak trains when it needs to), carrying about 1,200 passengers per train or about 30,000 passengers total. The same passengers mostly depart between 4 p.m. and 6:30 p.m., about 250 workdays per year, so the total peak ridership is about 15 million per year. The remainder, about 11 million per year or one-third of ridership, generating revenues of about $66 million per year, is non-peak.[4] Assuming ridership on Saturdays and Sundays is the same as weekday non-peak ridership, the figure of 11 million a year implies about 30,000 passengers a day.

U.K. Rail operators also have experience with the impact of service frequency improvements. According to the Passenger Demand Forecasting Handbook, going from a 60-minute frequency to a 30-minute frequency is equivalent to about a 10-minute reduction in journey time, on a typical 60-minute journey. On the GO Lakeshore route, this is about the same as a 16% reduction to the average journey. So introduction of half-hourly all-day services should bring a 16% increase in non-peak riders.

GO has introduced a half-hourly all-day service, with about 10 more trips each way on weekdays and about 20 more on weekends. The labour cost for each one-way trip is estimated at $400.[5] Energy, rolling stock maintenance, and track costs would triple this to about $1,200 per trip. For the extra 9,360 one-way trips per year, the incremental cost would be about $11 million. Initial counts indicate the service is indeed attracting about 15% more non-peak passengers, about 150 new passengers per additional train, or about 5,000 per day or 1.8 million per year.[6] With an average fare of about $6, incremental revenues will be about $11 million per year, offsetting the incremental labour costs, although not fuel or rolling stock maintenance. The overall growth rate confirms that growth is in line with U.K. experience, and that there is strong demand for increased non-peak rail services.

With development over the next decade expected to double underlying demand, GO's half-hourly lakeshore service will be good for the GTHA and should even be a financial success.

With EMUs, journey times are reduced a further 8%, because of the faster acceleration and braking. Going from half-hourly to quarter-hourly service brings an even larger benefit, equivalent to a 15-minute saving or about 25%. With a service of four trains per hour, passengers have a "turn up and go" service - they do not need to adjust their schedules to match the train. And they can cut their arrival time at the station finer, because if they do miss the train they were planning to catch, another one will come along soon.

So the overall revenue increase on the Lakeshore, compared with diesel trains, would rise almost $3 billion NPV with a mix of electric locomotives and EMUs. The increase is due to more frequent and faster all-day services. Weekday ridership could rise about 57,000 by 2023 compared with diesel locomotives, to 249,200 per weekday.

Assuming complementary policies, including fuel taxes, parking levies, land use controls, integrated fares, and an improved network of feeder public transport services, non-peak traffic could grow at 5% to 7% per year. This is the rate achieved in Greater London with similar policies. We assume, conservatively, that both peak and non-peak ridership increases a further 50% from 2023 to 2033, to 372,000 passengers each weekday.

Note that we are not assuming any increase in peak ridership, over and above GO's base case assumptions, as the journey time improvement from electrification would be small. In fact, many passengers travel one way in the peak and one way outside the peak. There are also peak-hour commuters who travel in the reverse direction, who are therefore by our definition "non-peak" passengers.

We are, however, forecasting that by 2023, non-peak ridership, including contra-peak, off-peak, and weekend ridership, would exceed peak ridership. This pattern is consistent with experience in Europe, on lines where frequent all-day services are offered.

Some may question whether GO can ever attract sufficient two-way all-day traffic volumes to justify a more intensive service. Outside downtown Toronto, most GO stations are surrounded by large parking lots, often in industrial areas with few jobs or other destinations within walking distance. Many stations do not even offer frequent regular bus connections to local destinations. We see this as a chicken-and-egg problem. At the current level of service, this situation is likely to persist. But with improved service and more people using it, change begins to make economic sense. Station car parks can be redeveloped at higher densities. Feeder bus services can be improved. Already, GO has attracted impressive all-day traffic onto the Lakeshore line. With coordinated planning policies, and a proactive operating strategy, GO can do the same on the rest of the network

The ridership and revenue uplift from frequent and faster all-day service with EMUs should be similar on the rest of the GO rail network. Existing non-peak ridership is very low, as on most routes, trains serve only peak trips. We assume that with electrification, the other GO rail lines can generate as many passengers as the Lakeshore route, with similar benefits. Essentially, we are saying that if they are given the same sort of rail services, then development and traffic will grow along these routes in more or less the same way.

Altogether, the total ridership increase from GO electrification with EMUs to 2033 might be about 125,000 daily passengers. Even this number might be conservative. As the entire network is upgraded, there will be "network effects." For example, it will be possible to go from Richmond Hill to Oakville in about one hour, with a train every 15 minutes, a trip that currently takes about 2 hours by public transport and sometimes that long by road. Many of these trips would have been made otherwise by car on the regional expressways. Rail will become the preferred mode of travel for trips not just to downtown, but also between many other destinations on the GO Rail network.

Further benefits to increasing ridership on the GO system would include reduced congestion on the roads, and the option to increase peak fares for faster trips, while lowering fares for non-peak trips to encourage greater use of the system. Land use intensification around more frequently used stations may also occur.

This is a very rough analysis, but it suggests there is a very good case for electrification of the Lakeshore line, immediately, with electric locomotives used in the peak and with 3-car EMUs operating a frequent all-day service. The up-front investment of about $1 billion NPV can be paid back entirely from addition fare revenues, and reduced operating costs to GO Transit.

After reviewing an earlier draft of this report, Metrolinx staff told us that they now recognize that different rolling stock strategies might generate better business cases, but at first chose to test only a standard operating plan in the electrification study. They are now studying further strategies, and are looking specifically at electrifying the Georgetown corridor and possible use of EMUs for the Union Pearson Express service. It is our hope that this new study, which has not been disclosed to the public, indicates the importance of Regional Express Rail in helping Metrolinx achieve its goals.

Electrification of other GO routes

Further detailed analysis is required to determine whether only some or all of the "all other routes" should be electrified. While the case for electrifying all other routes is not as strong as for the Lakeshore route, this is probably only because it has been treated as a single project. There could be a very strong case for electrifying some of the more heavily used "other routes," while diesel haulage or dual-power trains will be appropriate for some of the longer, less heavily used services, and outer sections of the network such as Oshawa-Bowmanville. But completing this analysis should not delay electrification of the Lakeshore, which is long overdue.

Electrification of the GO network with a fast and frequent all-day service could be transformational.

[1] This handbook was developed by the Association of Train Operating Companies (ATOC) in the U.K.
[4] Note this includes weekday off-peak and contra-peak, and weekend all-day ridership, all of which is more sensitive to journey time and wait times than peak ridership. While road congestion is worst at peak times, the GTHA also suffers severe congestion in the contra-peak direction and even on weekends. While our forecasting method works from the off-peak and contra-peak, changing travel patterns will mean the benefits of improved GO rail services will accrue to all travellers.
[5] Each one-way trip, from Burlington to Oshawa, takes about 2 hours. GO rail labour cost per train hour is about $180, as estimated in the Technical Appendix (Go Electrification). Allowing for driver rest time, labour cost would be about $400 per trip.
[6] Information provided by a senior GO Transit official to the author, 26 July 2013.