Toronto Light Rail Makes No Sense

This is a piece I wrote in 2010 and updated in 2012 in response to the revival of Toronto’s daft LRT plans. Some of it is out of date, but plenty of it is still quite valid.

I was surprised by the positive response to the reincarnation of Transit City LRT plan and now Onecity. These plans really don’t make any sense. There are some serious oddities as to cost, and there does not seem to be a cost-benefit or multiple account analysis which is standard the world over for these types of projects.

Practically all of the debate seems to be about the personalities and the council politics but not the actual pros and cons of the various proposals. For the Eglinton Line, I have not once seen a table the equivalent of the one below (from the Vancouver Richmond rapid transit study):

This is strange. A strange way to plan and debate rapid transit and a strange way to built it.

1. Cost

There is no point complaining about the high cost of transit systems, they are expensive, but I would like to understand why the Toronto estimates are so high. The Canada Line in Vancouver was built for a bit more than $100m per km with bored and cut and cover tunnels, a bridge and an elevated section. This is for a fully automated subway system, a line that provides better transit service than any LRT would be able to provide. That would be like a $3.6 billion Eglinton Line subway from Pearson to Kingston Road. What is going on here when the estimates for a shorter version of a buried LRT line are over $8 billion? This is not an area of difficult construction conditions, and in many respects, Vancouver is a higher cost environment. There would have to be a bridge over the Don, but in Vancouver the line also required a bridge. In Montreal, the high cost of public works has been linked to mafia involvement in the construction industry. I don’t think that anyone has suggested that this is a factor in Toronto, but there must be some reason why costs in Toronto are double or triple what they are in Vancouver. Why isn’t this a story? If this were gas prices, it would be the subject of a royal commission by now.

2. High Risk Strategy

The TTC bumf also lists a bunch of cities that it says have recently built LRT suggesting that subways are passé and LRTs are the new black – the choice that many other cities are adopting.

This list is highly disingenuous. Most big cities build subways. Those few big cities that build LRT-based systems, like Houston and Dallas, don’t have very successful transit systems. Of the cities listed:

• Some are small European cities like Strasbourg that suit LRT well but aren’t realistic comparators for a city like Toronto. Strasbourg is a small, yet very urban, city with a lovely light rail line. Don’t get me wrong, if they wanted to build Strasbourg in Scarborough, I’m all for it, including a light rail line around a transformed Scarbourg, but that isn’t what we are talking about.

• Paris and Madrid have huge metro systems and LRT is being built in the outer regions. And even that is not the whole story. Madrid has recently completed a huge subway expansion, making an already huge system even huger. Paris has launched a 20 billion euro transit expansion programme consisting of subways in the inner suburbs. These will be automatic systems with widely spaced stations with an emphasis on speed. Apparently in Paris and Madrid, they are still chanting “Metros, Metros, Metros”.

• Many are North American LRT systems that are not actually that successful.

• Other cities on the list are just out and out errors – Vancouver is listed and it does not have light rail.

• Some are just proposals which are not yet off the ground or running into serious cost problems.

• The cities not listed are the cities that have built and expanded subways in the last ten years. I suspect that number of riders of subways built in the last ten years would be multiples of the number of riders of LRTs built in the last ten years.

There is no model LRT-based transit system that Toronto can follow. This makes Toronto’s strategy a high-risk strategy.

LRT ridership statistics tell the tale of that risk (the APTA puts out good annual stats, or you can just go to wikipedia). Here are some daily ridership stats for North American systems (in thousands):

Monterrey	418.7
Calgary	252.6
Boston	215.4
Guadalajara	195.1
Los Angeles	156.6
San Francisco	153.9
Portland	127.9
Philadelphia	94.4
Edmonton	93.6
San Diego	92.1
Mexico City	70.1
Dallas	57.4
Newark	57.2
Saint Louis	55.8
Salt Lake City	46.9
Sacramento	44.4
Denver	42.6
Phoenix	37.4
Houston	35.0
Minneapolis	35.0
Baltimore	34.3
San Jose	32.8
Seattle	28.6
Pittsburgh	26.0
Charlotte	20.2
Buffalo	18.8
New Orleans	17.3
Cleveland	7.7

Not many of these systems are that successful and even the successes are tepid. As a comparison, the current Toronto Subway carries 1,054,200 per day and the streetcars carry 285,600 per day. Do not assume that the low ridership systems are just small. Dallas and Denver have large systems and many of the other cities have at least one long line.

The more successful systems are the first seven, but none are like Toronto’s plan for streetrunning LRTs that do not go downtown:

• Monterrey and Guadalajara have lower car ownership and both of those systems are radial downtown focused systems. This is low hanging fruit for transit.

• The Green Line in Boston operates more like subway. It is a branching line heading downtown that utilizes railway rights of way and it operates in a tunnel downtown. LRTs in railway rights of way are different from streetrunning LRTs because they are already partially segregated from other surface traffic, so they minimize conflict with other road users and adjacent neighbourhoods.

• Calgary is the standout success. It operates in railway rights of way and in medians, but there is also some streetrunning. The system was also very affordable and is well used. But this success would be hard to replicate because Calgary’s lines run downtown, the most transit receptive market, and Toronto’s costs are far higher. And even with a successful system, there are inherent trade-offs with at‑grade transit – the trains, cars, bikes and pedestrians get in each others’ way – and because of this, Calgary is moving to grade separation for the new line under construction and future lines.

• Most of the lines in LA run in railway rights of way.

• The muni lines in San Francisco all head downtown and operate in a tunnel downtown, but they do have streetrunning operations.

• Portland’s lines all run downtown, and Portland’s ridership is getting to the outer reaches of “success”. Its lines are only partly streetrunning, using freeway and railway rights of way as well.

Vancouver and Portland provide a good natural experiment being cities of similar size, age, and development pattern. Portland built a lightrail system, MAX, regularly held out as a success, and Vancouver built an automated ROW system, Skytrain. Vancouver has triple the ridership of Portland at slightly less than twice the cost. All the other transit metrics are better too.

And as you go down the list you get to the flops. Seattle’s link line cost $2.5 billion and is now only getting to its exceedingly modest ridership of 40,000. It is being expanded, but it might cost $6 billion in capital costs alone to get ridership over 100,000 per day by 2030. This is insane. (The Canada Line in Vancouver cost around $2.1 billion and ridership is already around 100,000 per day.)

3. Problems with Streetrunning LRT

The LRT proponents have been allowed to depict LRTs whisking people here and there while glossing over the negative pedestrian impacts of that whisking. Toronto, and more obviously Metro Toronto, is a big place. In order to be effective, transit service needs to be able to get from one side of the region to the other in a reasonable amount of time. That means rapid transit of some sort. The LRT advocates point out that streetrunning LRT can be made fast by having dedicated lanes, signal priority and grade separation where necessary. This is all true, but they are not fully recognizing the tradeoffs. All those things that enhance the light rail service also detract from the urban quality of the street. Dedicated lanes allow the trains to go faster, but that is like having a train highway in the middle of the street. In a wide boulevard with a green median, that is less of a problem. But on an urban street, highways, be they road or rail, detract from the pedestrian friendliness of the street. Streets with good pedestrian life are usually streets with slower traffic.

The TTC’s information on LRT seriously gloss over the negative aspects of LRT and serves some straight up nonsense:

Fast service also requires closing cross street intersections and giving signal priority to the trains at major intersections. But closing cross street intersections is terrible for pedestrians, and signal priority lengthens signal timing, also not great for pedestrians. The streets are very wide at LRT stops – not at all pedestrian friendly. Pedestrians do best in a fine grained urban pattern with frequent, fast light cycles and not in superblocks with slow light cycles.

The Eglinton Crosstown Transit Project Assessment Study has nary a mention of the pedestrian impacts.

Streetrunning is never as fast as ROW transit. Because there will always be conflicts with other street users, there will be compromises. Parents will ask for signalized intersections across the lines on popular walking routes to school, cyclists will ask for signalized intersections for bike routes, and motorists will forever campaign for left turn lanes to replace the U-turn manoeuvres that are proposed. Because LRTs run in city streets and have to respect other users, they will have to slow down. The Spadina streetcar has the same issues; signal priority was never implemented because of the impact on other road users.

Streetrunning is also never as reliable as ROW transit. Because the LRT never run entirely in their own lanes, there will always be collisions with cars that stop the trains from time to time. It is true that these are often the fault of bad drivers, but that does not change the fact that it will slow the service down and aggravate riders.

4. BRT Alternative

The cheaper alternative to LRT’s is bus rapid transit (BRT). Instead of light rail vehicles traveling on rails, buses travel on a paved surface. There is really very little that a light rail vehicle can do that a bus cannot. BRT would have separated lanes, signal control, mid street platforms, multiple door loading, platform ticketing etc just like LRT.

LRT advantages:

• more comfortable ride, although that advantage fades as the rails get old

• no street level pollution being electric

• vehicles can be grouped together into trains with more capacity than a bus

BRT advantages:

• buses can pass each other, although this requires an additional lane

• buses can drive around accidents and obstacles

• BRT costs way less than LRT. BRT is estimated to cost $10m per km in the Crosstown Study while the Sheppard line (entirely surface) is estimated to cost $77m per km.

The difference that LRT proponents hang their hats on is the difference in capacity. They say that BRT just doesn’t have the capacity to handle the expected ridership. This is mostly false. The Crosstown Study states at page 16:

Capacity/Reliability

Buses are smaller than rail vehicles and cannot be “coupled” together to operate in pairs or three car trains. As such, a local BRT service – one that services all stops – has less carrying capacity than LRT. High BRT capacities would only be feasible with by-pass lanes to allow some buses to operate express and pass one another at stops, and there is not sufficient space for a 3.5 metres by-pass lane in the Eglinton Crosstown LRT corridor right-of-way while providing a “comfortable” walking environment, bicycle lanes, four through lanes and left turn lanes for traffic.

Moreover, a standard 12 metres bus typically has an average capacity of 50 people per vehicle over the peak period. Given that transit forecast demand is in the order of 5,400 passengers; approximately 108 buses would be required, per hour, to service the demand. Even if articulated buses were purchased, it would only reduce the minimum number of buses to 72 per hour. In a partially grade-separated operating environment, transit service reliability is closely tied to traffic signal operation. As the frequency of service increases to be close to the normal traffic signal cycle length (40 cycles per hour) it becomes increasingly difficult to prevent bunching and gaps between vehicles, resulting in unreliable service.

This passage relies on a number of fudges. First, the capacity of a standard 40’ bus is closer to 80, not 50. Second, no one would propose BRT based on 40’ buses. Everyone would use articulated buses with higher capacity. Third, the actual capacity of an articulated bus is about double the capacity implied in the passage above (5,400/72=75), and they crush load higher. Articulated buses with a capacity of 150 could be run 40 times per hour for a line capacity of 6,000 per hour, enough to handle the projected peak demand.

The Crosstown Study acknowledges actual BRT capacity in the chart on page 15. That shows BRT having a capacity up to 8,000 per hour. That could be done using the double articulated buses used in BRT in South America. They apparently have a capacity of 270 (I suspect that this is crush loaded or with very few installed seats in the bus). If these buses were run every two minutes, the line would have a capacity of 8,100 per hour. And in extreme situations the buses could be run every minute, doubling that capacity. (This would cause some bunching.)

LRT does have lower operating costs because, having higher capacity vehicles, it requires fewer drivers to carry the same number of riders. However, if BRT were run every two minutes and LRT every four, the riders would have the benefit of the higher frequency and lower wait times. So it really is just a matter of lower costs for lower service.

To be clear, BRT would have all the same negative impacts that LRT would have, it would just cost way less. Many people don’t like BRT, and neither do I really, but it is clearly cost effective transit. It shows that all the handwringing about the cost and time to built rapid transit in Toronto isn’t really about rapid transit but about the cost and time to built LRT. A Onecity sized BRT system could be built in a few years for a few billion dollars.

5. Crosstown Study

Normally when transit projects are proposed, they are subject to a cost benefit analysis (often called multiple account evaluations because not every factor is reduced to a dollar figure). Various options would be compared in terms of cost, ridership, travel time, environmental impacts, land use impacts etc. This is hardly an exact science but it is useful to decision makers to see the costs and consequences of the options in one place. Whether for instance, the comfort advantages of LRT are worth the 7x cost multiple over BRT. I cannot find a cost benefit analysis for the Crosstown Line. There is really nothing that would qualify as a cost benefit analysis in the Crosstown Study, and some of the analysis that is there is dodgy. It is possible that there is a cost benefit analysis, and I just can’t find it, but even so, such a study ought to be at the core of the public debate. What is the use of talking about this without having the costs, travel times and ridership projections of the various options at the ready? And if there isn’t a cost benefit analysis for this multi-billion dollar expenditure, appalling is the only word.

One thing that needs to be covered by a study is passenger benefit from shorter travel times. Shorter travel times will attract more riders, but those riders will also be having more fun because they will be getting to wherever they are going faster. Valuing time is notoriously difficult, but there should be some accounting of that benefit to the rider.

There are other oddities in the Crosstown Study. For the tunnelled section of the line, it recommends twin bored tubes. But the only options considered were a single bored tube and twin bored tubes. But for this type of line, the most obvious option would be a cut and cover tunnel. Cut and cover construction is cheaper and keeps the stations closer to the service and thus easier to get in and out of. (This is even more important for a buried LRT that is slower than a subway. Riders are more willing to go through some effort to get to a subway platform if the result is a fast ride.)

Cut and cover is much more disruptive of surface activities than bored tunnels (although station construction is still very disruptive), but the obvious question for the study would have been whether the decreased cost and station improvement of cut and cover were worth the increased disruption or whether some cash compensation for the disruption caused by cut and cover would be less than the increased costs of bored tunnels.

6. Subways

A useful thought experiment would be to ask whether people would trade the current Yonge and Bloor subway lines for LRT lines. LRT lines that were slower and every once and a while stopped completely when a car accident occurred on the tracks. I think that we can safely agree that there is no chance. So if subways were the right choice when the city was smaller, why are they not the right choice now? The arguments that the LRT proponents raise are capacity, lower density, cost and time.

The LRT proponents seem to principally base their arguments on the idea that subways have too much capacity for the expected ridership. The number they dwell on is persons per direction per hour (ppdph). Endlessly repeated in all the pro-LRT materials is a chart showing that subways only make sense when peak demand is over 10,000 in the peak direction. In fact it really makes no difference whether the ridership has peaks above 10,000 ppdph. If a daily ridership of 100,000 would justify a line, it is irrelevant whether that were distributed evenly at 6,250 per hour for 16 hours of operation or whether there were pronounced peaks at rush hour and lower over the rest of the day. It just doesn’t matter. Capacity is not the deciding factor, ridership is.

Now the LRT proponents might say that if usage were only 6,250 per hour, then the capacity could easily be handled by cheaper LRT. But people are not attracted to capacity. They are attracted to fast, frequent and reliable transportation. If LRT were as fast, frequent and reliable as a subway, it would attract the same ridership, but it does not. If transit were about nothing but capacity, everyone would walk. All the sidewalks together surely have a capacity of 10,000,000 ppdph.

And if excess capacity were a concern, a solution would be shorter subway trains and platforms. The Canada Line in Vancouver has a maximum capacity of 15,000 ppdph, but right now it is operated at around half of that. (Building smaller stations did lower the cost of this line, but there is already concern that this might have been short-sighted with the strong ridership of the line.)

The most useful metrics are travel times and ridership, but you can hardly find these numbers. The March 15^th Advisory Panel Report on the Sheppard Line does not even mention the travel times and ridership is obscured. As per usual, the report harps on pphpd. It’s only in the appendices that you can find these:

This first option – the subway extension – would provide eight kilometres of rapid transit, with seven stations, serving an adjacent population of 34,000 people, and carrying 27 million passengers per year.

This second option – a light-rail line – would provide thirteen kilometres of rapid transit, with 25 stations, serving an adjacent population of 58,000 people, and carrying 17 million passengers per year.

The first thing to note about these is that the subway will carry more people. The second thing to note is that the served adjacent population numbers are silly. It works out to 794 trips per person per year for the subway and 293 for the light rail. If every single person in the subway catchment area rode the subway to and from work every work day, that is only 500 trips per person per year. Nothing gets numbers anywhere near that high. Clearly the served adjacent population numbers do not tell the real story on catchment area.

However as a relative measure, those served adjacent population numbers do show just how much more popular subways are over light rail: 2.7 times more popular. The LRT has more stops and more people adjacent to those stops, yet it will carry fewer people. And these figures also don’t include growth over time, which even the LRT proponents acknowledge favours subways. Nor do they include increased use of adjacent buses as their frequency is increased to handle subway traffic (or course they do catch the transfers).

The LRT proponents say that the Yonge and Bloor lines serve areas that are denser and more pedestrian friendly than the other areas of the city, and that is undoubtedly true. However, Eglinton has high density areas and low density areas that have tremendous development potential, akin to Yonge when that line was built. And population density maps show that there are more people living and working around the Sheppard line than you might think. It’s not the first line that I would build, but it really isn’t the low density suburbs.

Another complaint about subways seems to be that they are expensive and too expensive to build a whole system at once like Transit City. There is no getting around the fact that subways are expensive, and the electors ought to be aware of that. The question is whether good transportation and low carbon transportation in an era of climate change is worth it. The corollary complaint is because a whole system cannot be built at once, it pits one part of the city against the rest. This is surely nonsense. Not everyone’s birthday can be on the same day. Sometimes you just have to wait your turn.

Building a subway system is a process with long time horizons. The question is how do you want the city to work in 40 years? Do you want a city with large areas of pedestrian life and where it is feasible to live without a car? If you want those things, a subway system is probably the best way to get there. People might not wake up wanting “subways, subways, subways”, but they do want fast, frequent and reliable transportation and ROW transit best delivers it.

(Underground LRT lines are not the same as subways. The proposal to just bury the Eglinton LRT was nonsensical, and demonstrates how weak Ford is on this file. This has all the costs of a subway without the benefits: automatic trains, super short headways and lower marginal labour costs. That the province seemed willing to go along with this for a while is also alarming. The expertise at the provincial level does not seem to go much past Mayor Ford’s.)

7. Advisory Panel Report on the Sheppard Line

The Advisory Panel Report was a politicized, quickie justification for LRT. Naturally it is filled with problems. I didn’t see any of them pointed out by the media, only a few quotations from Gordon Chong. Does no one read these things?

The Panel’s Option Analysis starts at page 39:

Transit Service

• Ridership : provide the necessary capacity to meet expected ridership demand in 2031.

• Network Connectivity : provide a transit line that supports better connections with the transit system, improves overall access and network capacity.

• Level of Service : consider the door to door travel time of the end users, including out- of vehicle time (walk, wait and transfer times) in addition to in-vehicle time.

Ridership is completely subsumed into capacity. Transit service drives ridership, not the other way around. The question ought to be “which option will have the best ridership?” and then “does that option have the ability to meet that ridership?” Again, bizarrely, this section does not even quote the ridership of the lines. Only the ppdph. The panel scored the subway low on ridership even though it would have ridership higher than LRT. One reason is the higher cost, but that was already scored in other areas of the analysis, so scoring it again here has the effect of double counting.

Connectivity was scored higher for LRT than subway even though the LRT will require a transfer at to the Sheppard Subway. The reason is that LRT has greater promise of providing a whole network of transit over the whole region. This ignores the fact that the downtown market, because it is concentrated, is transit’s best market. Transit is never going to do as well in more dispersed employment areas.

The Level of Service section does not mention the travel times anywhere. I suspect that if they were any good, this panel would have mentioned them. Instead, the panel discusses concept of total trip times including station access instead of only the time actually in the transit vehicle. The panel states that station access time can be valued much more highly that time in the transit vehicle, and that the more closely spaced stations of the LRT will be better at providing good total trip times. But if this were true, the LRT would have higher ridership. But as we see above, this isn’t true. The LRT ridership is lower even though the line is longer with more stations. (It is quite true that some out-of-vehicle time like waiting is valued more than in-vehicle time, but not always, and as the numbers show, sometimes the faster travel overwhelms this effect.)

These are the scores the panel gives for a few of the variables:

	LRT	Subway
Ridership	4.57	2.29
Network connectivity	4.71	3.14
Level of service	4.14	3.57
Environmental sustainability	4.14	3.57

Except for network connectivity, which is probably more of a toss-up, the subway option is superior to LRT on the other criteria. Yet the panel apparently arbitrarily gave it a lower score. And the use of these numbers gives this scoring the air of precision when really these are panel members’ rankings.

8. Streetcars and LRTs

LRTs and streetcars do have their place. Streetcars are good in pedestrian areas of the city where they run slowly and have the effect of slowing all the traffic down. This can be very desirable and can be a good argument for building streetcars in areas that should be more pedestrian friendly. But those streetcars will not be rapid transit. They will be slow, short-haul transit. The current streetcar system in Toronto serves this purpose reasonably well in the area south of Bloor. For that reason the removal of the Yonge and Bloor lines on completion of the subways under those streets was probably a mistake, and I would not advocate removing the Queen Street Streetcar if a subway were built under it.

LRTs to do best when operating on established ROW’s that don’t damage pedestrian fabric of the city or in very suburban areas where pedestrianization is never going to happen. Some of Sheppard is a long way from being urban, but I still don’t see why LRT is preferable to BRT. There would seem to be some potential in Toronto to electrify some of the GO network and turn it into an LRT system, and that is worth investigating. But that would not be cheap because of the need to deal with conflicts with the freight trains also using those tracks.

9. Decision Process

Minister Bob Chiarelli says that the time for debate is over and that the government is going ahead with the revived Transit City. But the debate to date hasn’t been a very good one. Much of the debate has been tendentious. LRT proponents give examples of light rail expansion around the world when none are very relevant to Toronto. Nor do they mention all the cities building and expanding subway systems. Instead we get pictures of Strasbourg. And Ford hasn’t contributed anything useful. To properly debate this, everyone needs to be looking at useful transit studies that compare ridership, frequency, time, land use and cost.

For Eglinton, I would expect that a study would look at 10 different options including options that went from Pearson to Kingston Road, an elevated line, BRT and LRT with a tunnel in the centre, purely surface BRT and LRT, a full subway and various combinations and lengths with all the pros and cons of the various options.

A lot of this debate seems to be driven by Mayor Ford by pushing people into thinking “the enemy of my enemy is my friend”. He is a terrible advocate for transit because he thinks about transit from a car driver’s perspective: “[We] don’t want these damn streetcars blocking up our city.” This obviously isn’t the perspective of a streetcar rider. But just because LRT is Ford’s enemy does not mean it should make it anyone else’s friend.

10. Some questions for Karen Stintz and Bob Chiarelli

Why is transit infrastructure so much more expensive in Toronto than it is in Vancouver? What is being done about this?

Why can Vancouver build subways and ROW transit, but Toronto, a city twice the size, cannot?

What is the capacity of an articulated bus?

What is the capacity of BRT running every two minutes?

What is the cost differential between BRT and LRT?

Whether a complete BRT system could be built in 3 years for money already budgeted?

Is it TTC practice to build on the basis of a cost benefit analysis?

Is there a cost benefit analysis for the Crosstown Line? What about an internal cost benefit analysis in the provincial government before the project was funded? What are the ridership, travel times and costs for all the various options for all the Transit City lines? Is this information important to you? And if you don’t know, what are you doing to find out?