GO Electrification

Table A1: GO Electrification Options

We have estimated the incremental costs to operate with a mix of EMUs and electric push-pull trains. Except where otherwise noted, figures are derived from data provided in the GO Electrification Study Final Report[1] and technical appendices.

Infrastructure Capital Costs

These should be about the same as for GO's electric push-pull option. It might be necessary to provide some additional sidings to split and join the EMUs, but we have not included these costs.

In its Appendix 8B, GO presents a capital cost estimate of $1,019,000 to $1,170,000 for electrifying the Lakeshore.[2] This is shown in the Summary Report, Table 12, reproduced above, as $764 million NPV, about 69% of the actual total. This is because GO discounts figures to a common start year, 2010, and the money is actually spent over several years. Infrastructure capital expenditure to electrify the rest of the network is about twice as much, $1,605 million NPV, presumably because although the ridership is similar, it has about twice as many route km.[3]

We have not examined GO's infrastructure capital cost estimates in detail, and take them as given. Given the issues we have found in the rest of the study, a detailed review is warranted, but was beyond the scope of this study.

Rolling stock capital costs

The GO Electrification Study Appendix 8B contains some very conservative assumptions that we think are biased against EMUs. The consultants who prepared the estimates may have had limited experience with EMUs, which are still quite rare in North America. Moreover, although some commuter rail systems are operated by private contractors, as is GO, almost all of the North American electrified systems are operated by public-sector organizations and are encumbered by inefficient work rules and procedures.

Specifically, GO assumes a minimum spares ratio of 20% for EMUs, higher than the 16% spares ratio allowed for diesel locomotives. While the 20% may be "standard" among U.S. operators, U.K. train operators work successfully with spares ratios as low as 5% for EMU fleets. If the operator is, occasionally, short an EMU, it can substitute a short locomotive-hauled train to provide the service with similar performance. And with a high-frequency service, every 15 minutes, a train service might be missed only every month or so. We assume a minimum of 10% spares and at least 2 spare units, allowing one for heavy maintenance.

GO estimates the rolling stock cost to convert the Lakeshore route to electric locomotives as $91 million NPV, which does not seem unreasonable. Existing diesel locomotives would be cascaded to meet growing traffic on other routes. Electric locomotive do generally cost a few million dollars more, each, than diesels, and GO will also need to spend money to modify its maintenance facilities.

Rolling stock capital costs would be higher with EMUs. Although EMUs would be bought in place of some electric locomotives and bi-levels, GO is correct that that EMU cars are more expensive to buy than electric locomotives and unpowered cars, when configured as 10-car trains. However, even using the figures in GO's own reports, the premium is more like 25%, and not the 40% GO claims.

GO assumes a base cost of $2.6 million and an "extended cost" including engineering support, spares, etc. of $4.6 million for each single-deck EMU car. We think this figure is probably too high, especially for a new large fleet that can be procured with competition. In 2009, Paris purchased a fleet of 300 double-deck EMU cars (actually 60 five-car trains) from Alstom and Bombardier for EU917 million ($1,175 million), or about $3.9 million per car; this seems to have been closer to an "extended cost," including spares, support, etc.[4] Sydney, Australia, has recently purchased new "Waratah" double-decked EMUs built in China, apparently offering high quality at an even lower price.[5] These are built in Changchun Railway Vehicles Company, which produces subway cars in joint venture with Bombardier.

We assume an "extended cost" of $5 million for powered bi-level EMU cars, 60% more than LTK's $3.1-million cost for an unpowered bi-level cab car and 25% above the price supplied to Paris. We think this figure is probably still too high.

GO's "extended cost" for unpowered bi-level cars is $2.75 million, or $3.1 million with a driving cab. GO's cost for an electric locomotive is $11.2 million, versus $7.8 million for a diesel. A 10-car bi-level push-pull train with an electric locomotive and with one of the bi-levels having a cab, costs about $40 million.

An all-day service every 15 minutes between Hamilton and Bowmanville would require 16 EMU sets, plus two spare sets. We assume these are actually 3-car sets, each with two powered EMUs and one unpowered bi-level.[6] The 18 sets, or 54 cars, would substitute for 5 electric locomotives and bi-level trains. GO might also incur costs of, say, $20 million to provide maintenance facilities for this new type of train. The net additional capital cost should be about $133 million NPV.

GO estimates the rolling stock cost to convert "all other routes" to electric locomotives as $452 million NPV,[7] more than 5 times as much as for the Lakeshore route. This figure seems to reflect the loss on disposal of 52 diesel locomotives, which the consultants assume would be sold for $1 million each and replaced with new electric locomotives costing $11.2 million. We question this assumption. Our understanding is that there is a reasonably liquid market for good, slightly used locomotives. GO has replaced all its old locomotives and now operates entirely with MPXpress locomotives that are used on 12 other existing systems across North America. These also seem to be the locomotive of choice for new operators, which start up every few years in cities across the United States. GO should be able to get $3 million for each used MPXpress locomotive, perhaps more. GO currently has 65 locomotives, and will certainly have continued use for several of these as it extends services to places like Kitchener, St Catharines, and perhaps Peterborough. And GO has several years to find the right buyer. The rolling stock cost for conversion to electric locomotives, for all other routes, should be reduced $2 million for each of 52 locomotives, or $102 million NPV, to $252 million NPV.

On "all other routes," traffic will be lighter and 32 two-car, bi-level EMU sets should be sufficient for likely traffic with a 15-minute all-day service. There is an issue whether to power both cars, to achieve higher performance and reliability, or have one powered car and one trailer. Both options can work. We assume, conservatively, that both cars are powered EMUs with cabs. The 32 two-car, bi-level EMU sets displace 6 locomotive-hauled trains, so the net cost increment would be $411 million NPV, compared with diesel operation, or $554 million for the entire GO rail system.

Operations and maintenance costs

GO's electrification study Appendix 8C[8] Table 1 indicates a net O&M saving of about $14.6 million per year with electric locomotives on the Lakeshore. The corresponding NPV figure in the Summary Report is $337 million or 23.6 times the annual figure. According to Table 1 in Appendix 8C, the annual figure appears to be made up of $18 million reduced energy costs, and $1.1 million rolling stock maintenance savings, offset by $4.5 million costs to maintain the wayside electrical system. GO also uses an NPV figure of $337 million, 23 times the annual figure. For all other routes, the saving is slightly less, $10.3 million per year in Appendix 8C, although in the summary report GO's NPV figure is $360 million, somewhat higher, and 35 times the annual figure.

To estimate the O&M savings from operation of smaller EMUs but at a higher frequency, we need to look in more detail at the energy, maintenance, and labour costs.

Energy

Appendix 8C of the GO electrification study indicates that electricity consumption is assumed to be about $0.00756 per ton-mile.[9] A GO bi-level car weighs about 55 tons, with 100 passengers. Locomotives weigh about 135 tonnes, so the weight of 10-car train is about 685 tonnes or 68 tonnes per car. EMU cars, fully loaded, weigh about 60 tonnes, so slightly less.

The energy cost for 3-car EMUs operating every 15 minutes, or 12 cars per hour per direction, non-peak, will be about $4.3 million per year less than for 10-car electric locomotive-hauled trains every 30 minutes. Capacity is also 40% lower, but the capacity is not needed anyway in the off-peak. The actual saving will depend on driving, and one benefit of EMUs is that they have faster acceleration, but this also uses more power. On the other routes, the energy saving from operating two 2-car EMUs every 15 minutes instead of a single 10-car locomotive-hauled train should be even larger, about $5.5 million per year.

Rolling Stock Maintenance

According to Appendix 8C of the GO Electrification report, the maintenance cost savings with electric locomotives will be almost negligible, about $1.1 million per year. We think this is a serious underestimate. While the maintenance costs for an electric locomotive may only be about $40,000 per year less than for a diesel locomotive (which explains the $1.1 million per year saving with 34 locomotives on the lakeshore), electric trains have faster acceleration, so about 10% fewer trains are required to provide the same service. GO acknowledges this difference in its fleet capital cost analysis, but seems to have overlooked it when estimating O&M costs. Maintenance cost for a diesel locomotive and 10 coaches is about $1.7 million per year. Assuming, very conservatively, that 5% fewer trains are required (locomotives and coaches), the saving will be about $4.3 million per year or $99 million NPV.

Rolling stock maintenance calculations are complicated, because costs depend both on time and usage. The consultants who prepared GO's Electrification study, Appendix 8C, are not very clear about how they derived their estimates of maintenance costs for electric and diesel locomotives and EMUs. Firm estimates are difficult to obtain, because maintenance costs depend on how equipment is used and most operators have experience with only a few types of equipment. While the estimates presented in Appendix 8C of the GO study, Table 3, are given to the last dollar, implying great precision, it is not clear how they relate to actual costs either of GO or of any other operator. They can best be described as educated guesswork, with a lot of underlying calculations for support.

The study also provides estimates of EMU maintenance costs, also in Table 3 of Appendix 8C. At $254,811 per year, they are about 50% higher than for unpowered bi-level cab cars and almost double the rate for bi-level coaches. The problem here is giving prices per year, without regard for usage. Some maintenance costs do depend on time, but at least half relate to usage (mileage). We use GO's figures, but assume half are mileage-related, and convert them to a mileage rate, assuming they reflect average use of 80,000 miles per year.

With 3-car EMU sets operating non-peak services on the lakeshore, savings would be about $5.65 million per year or $130 million NPV. Savings on all other routes are similar; although there will be more EMU cars required, 2-car sets will replace 10-car bi-level in the off-peak services.

Labour Costs

GO's analysis assumes no change in operations, and so there is no change in labour costs with electric locomotives. In fact, this is not correct. Electric locomotives have faster acceleration, so crew hours per trip will be reduced by about 10%. With EMUs, the time savings will be even greater, about 20%, but the savings will be wholly offset if a higher service frequency is operated.

Crewing policies are an issue. In the U.K., train driver costs are about $60 per train hour, and most EMU commuter trains are operated by a single driver. GO pay levels are similar, but working practices require a crew of three staff on each train, an engineer (driver), conductor (guard), and ticket inspector. Apparently the cost is about $70 per hour each for the engineer and conductor, and $40 per hour for the ticket inspector. Altogether, the labour cost is about $180 per train hour.

Going from half-hourly to quarter-hourly service will require about 40% more train hours. This would be a good time to negotiate more efficient labour practices: in return for 40% more train hours, crew per train will be reduced to $110 per hour with single-driver operation, but still with a conductor/ticket inspector on every train. With much smaller trains, there should be no reasonable case to oppose this arrangement.

Single-person operation, with occasional on-board inspectors paid closer to a true market price, is normal practice in most western European countries where EMUs are in service, and would reduce the incremental cost roughly by half. Certainly the labour arrangements should be agreed with the unions before the final decision is made to acquire the EMUs. An agreement should be possible, because although there will be fewer crew members per train, there will be more crew overall and the jobs will be more sustainable.

Track costs

GO's analysis assumes no change in operations, so there is no change in labour costs with electric locomotives. GO now owns most of its track, and maintenance costs are therefore internalized and not separately calculated. Generally, track costs relate to the speed and weight of trains running over the track. Replacing 10 car bi-levels and electric locomotives with two three-car EMUs should reduce track wear and tear. However, GO has provided no data, so we have not included this in our estimate.