Monday’s approval of Metrolinx’ plans to run diesel trains on the Weston/Georgetown corridor stirred up lots of discussion here, in the mainstream media and at City Hall. If this approval rested on solid data and projections, we could simply argue the fine points and debate rollout plans. However, the claims made by Metrolinx for emissions from the project, comparisons with auto travel and supposed reductions by redirected auto travel depend on calculations that are transparently wrong.
In brief, Metrolinx assumes that every GO train trip, both ways, every day, all day in the corridor will be completely full of passengers, all 1,900 of them (a fully seated load on a 12-car train). This absurd premise overstates the likely ridership by a factor of at least 4, probably greater (details follow later in this article) with the following effects:
- Pollution caused by the trains is a fixed number determined by how many trips they make. If there are fewer passengers, the pollution per passenger trip is much larger than claimed by Metrolinx.
- If there are fewer passengers, then fewer auto trips are diverted to rail. This does not affect the pollution saving per trip (presuming that one even agrees with this premise), but the total saving is greatly reduced because so many fewer trips are diverted.
Opening day (2015) traffic projected for the corridor is 184 GO trains and 140 UPRL (Airport) trains. The total trips calculated by Metrolinx for the corridor GO services is 349,600 per weekday. To put this in context, the entire GO rail system carries about 180,000 passengers per day today.
In practice, the trains will carry nowhere near 1,900 per trip on average. Peak travel will be heavily inbound in the AM and outbound in the PM, with lightly loaded trains in the counterpeak. During the off-peak, loads will be much lower than at peak, and some trips (notably inbound late evening runs) will be almost empty. The same patterns can be seen on the Toronto subway system.
I am inserting the break here for those who don’t want to read the gory details, although the conclusions are down at the end.
The source material from Metrolinx on which this discussion rests includes background material on the calculations of emissions.
What is the All-Day Demand?
At this point, I must engage in some “back of the envelope planning”. Transit lines have a fairly predictable demand curve through the day, and this can be used to extrapolate from peak period demand projections to all day loads.
Peak hour demand generally represents half of the peak period demand given that the shoulder peak hours carry fewer passengers. A three-hour peak can be estimated at twice the peak hour. Double this to get both AM and PM peak demands.
The off-peak service on a well-used line (such as urban routes in Toronto) carries about half the all-day ridership provided that the service is reasonably frequent.
If we have a peak hour ridership of 1,000, then the all-day number will be somewhere around 8,000. (Double the peak hour to get the peak period, double again for AM and PM, double a third time to include off-peak riders.) The 8:1 ratio will be lower if the service and/or travel demand is strongly skewed to peak travel due to other factors such as an inability to reach the line thanks to poor feeder services or full parking lots.
Note here that I am trying to be as generous as possible in adjusting the Metrolinx figures to avoid overstating my case.
If GO will be carrying 1,900 per train at peak on a 15-minute headway, that’s 7,600 per peak hour. The corresponding all-day ridership would be 60,800. This number could be affected by various changes in the assumptions:
- If peak service runs at 12- or 10- minute headways, and off-peak demand stays at the same ratio, then all-day ridership would go to 76,000 or 91,200.
- If off-peak demand behaves more like commuter rail as we know it rather than like a heavy urban line, then the all day riding will be substantially lower. A fifty percent reduction in projected off-peak demand would reduce the ratio from 8:1 to 6:1 and the figures above would scale accordingly. This would reduce the 60,800 number to 45,600.
- Some counterpeak demand will exist (again by analogy to the subway system). However, counterpeak riders will almost certainly need frequent local transit at the outer end of their trips.
- The ability of GO to attract riding strongly depends on good transit service at the outlying stations and intensive development within walking distance of them. This is vital for counterpeak trips as well as for eliminating the need for parking for all inbound riders. (Again, compare to the subway network.) Nothing in Metrolinx’ plans suggests that this will actually occur or be funded, but it should be an integral part of their plans. They talk about “Mobility Hubs” but don’t make any provision for actually bringing them into existence.
This gives us a range of all-day demand between roughly 45,000 and 90,000 trips, not 349,600. Therefore demand-based calculations are off by a factor somewhere between 4 and 7.5.
By contrast, the demand modelling backgrounder to “The Big Move” projects much higher peak hour demand on lines in this corridor. However, this could only be achieved with headways in the 5-6 minute range (10-12 trains per hour per direction) which is substantially more service than the model parameters for the emission calculations.
Those 184 GO trains are not all destined for Brampton and beyond, but would also run on the Milton and Barrie lines (all day) and the Bolton service (peak). If the Metrolinx demand estimates are believed (many planners feel they are exaggerated), then the combined peak point demand for the peak hour is about 55,000 passengers. This is considerably more than the subway which strains at much above 30,000. At 1,900 per train, this would require one full GO train every two minutes, or 30 per hour. Clearly, the opening day 184 trains will handle far fewer passengers.
Adjusting for Demand Levels and Tier 4 Diesel Standards
For the sake of argument, I will take the base data used to calculate emissions at face value and will not challenge the calculation of pollution outputs from the diesel trains nor from the autos they might replace. The real issue here is the overstatement of demand.
- The NOX per passenger trip (see page 2 of the document linked above) for GO trains is shown as 5.71 grams versus 6.61 for autos. However, this is for a Tier 2 diesel. If we adjust for Tier 4, the train value drops to 1.14. However, correcting for the actual demand places the per passenger value at between 4.56 and 8.55.
- Particulate Matter (PM) per passenger trip is shown as .20 grams for Tier 2 diesels and .56 grams for autos. Tier 4 diesel is expected to drop this by 90% to .02 grams. Correcting for the demand level gives a range of .08 to .15.
- Carbon Dioxide (a greenhouse gas) is shown as .51 kg per passenger for diesel (the engine tier has no effect on CO2) and 6.08 for autos. Correcting for demand raises the diesel value to a range of 2.04 to 3.83 kg.
In brief, PM and CO2 are still better than autos, but not by as much; NOX may be better or worse depending on assumptions regarding demand.
Pollution from the Airport Link
The Airport service is in a much worse situation because the ratio of horsepower to passengers is much higher and, with that, the pollution per trip. Metrolinx uses Tier 3 ratings for their calculations. Even without adjusting for possible overstatement of passenger volumes, the UPRL figures (page 3 of the linked document) are sobering.
- NOX for airport trips is 77.54g by train, but only 6.61g by auto. Adjusting for Tier 4 only reduces the train value to about 15.5g, over twice the auto value.
- PM values are 1.41g and 0.56g respectively. Correcting for Tier 4 (roughly 20% of Tier 3) brings the PM value for trains down to .28g.
- CO2 numbers are 6.93kg versus 6.08kg.
For NOX and CO2, the airport service, even at Tier 4, will generate more pollution than the auto trips it replaces. This directly contradicts claims that the airport route will reduce pollution.
Metrolinx assumes that every trip to and from the airport will carry 56 passengers. This doesn’t sound like much, but is in fact a seated load for the cars that will be used in their new configuration. Any decrease in the average load per trip raises the emissions per trip accordingly.
Diverting Auto Trips to Reduce Pollution
Now we must turn to the auto side of the balance sheet. I will take the calculated emissions per trip as given by Metrolinx, but of course the number of trips diverted to trains will be much lower than claimed. Therefore, the emission reduction due to this diversion (assuming you accept the premise of such diversion) will be much lower than claimed by a factor somewhere from 4 to 7.5.
As an example, total GO CO2 emissions (which are not affected by Tier 4 standards) are given as 46 kilotonnes per year for rail and 553 kt for auto, a difference of 507 kt. Once the demand correction is applied, this falls to a range between 68 and 127 kt. There is a saving, but not as much as claimed. This will be offset by the higher pollution from train operations on the airport service.
It’s important to remember that this “reduction” is in the airshed of the Barrie/Georgetown/Milton corridor while the new pollution due to the trains is concentrated along the rail lines themselves.
The calculations underlying the Minister’s approval of the Metrolinx proposal are deeply flawed to the point of invaliding claims made about environmental impacts. The Minister and the Government have committed to a major policy decision that is not supported by materials filed by the proponent.
Whether this is an error of oversight (nobody recognized the flaw in ridership estimates) or an error of misrepresentation (someone hoped that this flaw would go unnoticed) is not for me to say. Either way, this shows extraordinarily poor analytical skills in Metrolinx and in the Ministry that this was not caught.
To reiterate (for those who scrolled all the way down to skip the details), correcting the Metrolinx data for Tier 4 emission standards and for their overstatement of riding in the corridor:
- For GO services:
- NOX emissions per passenger trip will lie in a range straddling the projected emissions for cars.
- PM emissions per passenger trip will be much lower for trains than for autos.
- CO2 emissions are lower for trains than autos, but the difference is much less than claimed.
- For Airport services (with no adjustment for possible overstatement of demand):
- NOX for train passengers will be over twice that for auto users.
- PM per passenger trip will be half that of autos.
- CO2 per passenger trip will be greater for trains than for autos.
Because the demand in the corridor is overstated by a factor of at least 4, claims of pollution reduction due to diversion of auto trips to rail are similarly inflated.
Metrolinx and the Ministry of the Environment owe everyone an explanation for this situation. Community groups have worked tirelessly in support of electrified transit and endured much criticism rooted in the belief that any trains, no matter what their technology, were vastly superior to auto travel. Electrification was something for tomorrow, something that needs a two-year study, something about which oddly little was known beyond its being “too expensive”.
Do the study, but structure it so that important information — basics like a validation of capital costs and operating savings — are available to guide policy decisions as soon as possible. If past studies (some as recent as Metrolinx own evaluation of the Lake Shore proposal in 2008) are wrong, explain why we should believe a new set of consultants more than the old ones.
We have already had one ministerial resignation this week in Ontario, and the Minister of the Environment would do well to encourage Metrolinx’ participation in reviewing the validity of their environmental claims.