Wednesday, July 16, 2014

BRT and LRT: the transit war

One of the most contentious debates currently raging in urbanist circles about transit is the BRT vs LRT debate. BRT stands for Bus Rapid Transit, LRT stands for Light Rail Transit. Both are forms of rapid transit, or maybe better put, semi-rapid transit. Both seek to provide faster, more reliable and higher capacity transit by implementing many features like:

  • Greater distance between stops
As I presented in another article, the speed of a transit line is highly dependent on how often it has to stop, even more so than the maximum speed it can reach, at least in urban situations:

Average speed of a transit line versus the average distance between stops

  • Transit right-of-way
This is often considered the most important feature, the transit vehicles must travel either on a lane where only they have the right to travel and with priority at intersections where other vehicles cross their path, or on a path that they alone are allowed to be on, others can't even cross it (elevated or underground rail for example). The point of it is to reduce the number of times that transit vehicles have to stop or slow down by getting rid of other vehicles from their path, and to allow them to travel at higher speeds, which is mainly useful for subways and commuter rail.
  • Paying before boarding and boarding through all doors
This means that people pay not when the vehicle has arrived, but when arriving at the stop or station. The most well known example of that is the turnstiles at subway stations, but on stops on the street, it can be simply a machine at the stop that prints out a receipt which you must carry with you but do not need to show to the driver, based on an honor system (proof of payment, POP), with severe fines if you're caught on a vehicle without one. This also allows boarding from all doors.

The point is to allow for fast boarding and deboarding, reducing the time spent at stops or stations. This is especially important when you have vehicles with high capacities. Imagine a subway with a capacity of 1 000 passengers who all have to pay when boarding through just one door. That would be insanely slow, and unreliable.

Anyway, all these features aren't dependent on what kind of transit vehicle is involved. Though we generally do it with trains and subways (so-called, Heavy Rail Transit), nothing stops us from doing it with buses too. Which is what Bus Rapid Transit is, bus routes that take on many features of subways in order to provide better service.

Light Rail Transit is the same, except using light rail vehicle instead of heavy rail. This means lighter, shorter cars that can travel on the street, as they are more nimble and can tolerate tighter curves. Something more like streetcars of old than passenger rail.
Sapporo tram, a light rail vehicle (a vintage streetcar behind it)
One car of a Tokaido train, much longer, not segmented and heavier

What is the issue?

The issue is that both LRT and BRT were conceived in recent years as ways to offer an intermediate transit option between slow, low-capacity buses and high-capacity subways. Buses had low initial costs, subways had very high initial costs, and there was little in between. LRT and BRT came up as solutions to fill that gap, for much less than subways, they could offer much better service than local buses.

So when there is a transit corridor that has the ridership to justify better than buses but not high enough to justify subways, LRT and BRT butt heads. Both technologies have their fans.

BRT supporters are especially vicious versus rail transit, some declare rail to be obsolete and completely useless, expensive to build when we already have roads we could appropriate for buses. They claim that BRT can be a dirt cheap method of building rapid transit and could allow us to build much more complete rapid transit systems much faster.

LRT supporters on the other hand recognize that buses have a role to play in transit, but they point out that there is a marked user preference for rail transit as more comfortable, less noisy, less polluting and more predictable in its movements (a boon for pedestrians and cyclists). They also are pretty skeptical of BRT as they claim that BRT is not that much cheaper to build than equivalent LRT lines for significantly inferior qualitative service.

Are there objective differences?

Putting aside qualitative subjects of perceived quality (which greatly advantages LRT), there are a few main factors that differentiate these technologies.

1- Capacity

Capacity of a rapid transit line is very important if you want to have transit-oriented developments along the line. No matter how fast a line is, if it can't carry enough people, it will stunt the growth of the zone it's built in, just like congested highways stunt the growth of suburbs around them. This is a touchy subject because if you look at the numbers only, both seem to deal with the same capacities. However, this isn't quite correct, because these capacities are obtained differently.

Okay, here's how a transit line works for the most part:
Standard transit line
So a vehicle moves down a set path, stops at pre-determined stations or stops (in blue), stops, lets people out and in, then goes to the next station, rinse and repeat. On a rapid transit line, vehicles stop at all stations.

Doing it like this, LRTs have a major advantage over BRTs in that you can attach cars together and obtain trains of vehicles. That means a single driver can carry 500-600 people, maybe even more, since the vehicles are on a guided path, they remain stable as you make each train longer. You can't tie two buses together, as they are not on guided rails, they would jump up and down, move right and left and maybe even cause accidents. The most you can do is use articulated or even biarticulated buses, limited to 120 or 200 passengers.

So in this way of doing things, LRTs win without a doubt as they allow much more capacity per train than buses can provide. The Calgary LRT works this way and provides a capacity of around 15 000 passengers per hour per direction (pphd). With simple buses (not articulated), the highest possible capacity is perhaps around 2 000, supposing a bus every 2 minutes, with articulated buses, the capacity is around 4 000 pphd. With bi-articulated buses, maybe around 7 000 pphd, but bi-articulated buses are rather rare as they become difficult to maneuver around on city streets, not having the safety that rail tracks provide.

How BRTs can provide greater capacity, passing lanes for express buses

But BRTs have an ace in the hole, at least according to their promoters, buses are more flexible, they can change lanes easier and go out on city streets. So in order to provide higher capacities, one trick that some jurisdictions have used, most famously Bogota's Transmilenio, is to have two lanes instead of one and allow buses to pass each other. This allows them to pile many different lines on one another, piling local buses, limited buses and express buses on top of one another.

With just one lane, if a vehicle is delayed or stuck, vehicles behind it cannot go around, but with two lanes, they can. The express lines can also be faster than the lines that stop more often. This has advantages and disadvantages. 

It increases capacity, yes, but it requires much larger infrastructure to do so. LRTs require just about 3 meters per lane (10 feet), as they are on rails, they don't need lanes to be much bigger than vehicles. Buses require wider lanes as they are not on a guided path they can't deviate from, so a strict minimum of 3,3 meters (11 feet) is expected. Plus, you need passing lanes, so you need two lanes, for a total of 6,6 meters minimum per direction, for 2 directions, that's about 14 meters (46 feet), add space for a station between the two directions and you're easily up to 20 meters (nearly 70 feet), which is how wide Bogota's Transmilenio is. The reality is that you just cannot fit the Transmilenio in most cities, not in their urban cores at least, unless you have gigantic streets you can completely give over to the BRT. It's nothing short of a gigantic bus highway.

BRT supporters would say also that it gives buses more flexibility as you can tailor make lines to suit demand. A lot of people from point A go to point B? You can build an express instead of having them stop at every station on the way. You can even take buses off the trunk line and into side streets.

But for users, this "flexibility" manifests as complexity, here is just one part of the system map of ONE trunk line of the Transmilenio in Bogota:
Part of the Transmilenio's map for one trunk line
This is about 20% of the actual map of the trunk line, there are 12 similar documents. Good luck finding your way around as a tourist or even as a resident who has to go somewhere he's never gone before. Try not to take the wrong bus.

Not only that, but as express buses have to merge to the right then merge left when entering and exiting stations, this creates congestion at stations when there are many buses. As a consequence, unlike subways and LRTs which can maintain the same speed even at full capacity, the speed of the Transmilenio falls during peak hours because of station congestion.
A graph I found illustrating average speed versus the number of buses per hour on the Transmilenio
Lastly, this means that even if there is one bus every minute at station, if there are in fact 5 different lines stopping there and you want to take one line specifically, then the effective frequency for the user is 1 bus per 5 minutes, as most buses that stop at the station will be useless to them.

So, in other words, LRT can get more capacity out of a narrow right-of-way, BRT can however leverage its flexibility to obtain similar capacity at a cost of simplicity and requiring a lot of space, which can be hard to find in dense urban areas. So if you can only find the place for two lanes, there is no question that LRT will provide significantly higher capacity all else being equal.

2- Economics: labor vs capital

One major difference between LRT and buses is in their cost structure. Putting rail in and building the electric system to allow LRT to run is more expensive than just painting bus lanes on existing roads (though real BRTs with high capacities may need a concrete road to be built, else they will tear the pavement to shreds quite quickly, a loaded bus being nearly as bad as a truck on roads). How much more expensive? Well, it really depends, in France, they build their modern tramways (LRTs) for as low as 16 million euros/25 million dollars per kilometer (40 million per mile). On the other hand, Gatineau in Québec built a BRT for around 15 million dollars per kilometer (25 million per mile). But it really varies, transit projects vary widely depending on what is included in them, many of them end up having a lot of beautification projects tied to them, ballooning costs.

There is no question that LRT has higher initial capital costs than BRT, vehicles are more expensive, rail is more expensive than reusing roads, electric lines are more expensive than not installing any. If the analysis stopped there, BRT would win without a hitch. But in terms of long-term costs, LRT looks much, much better.

For one, light rail vehicles can last much longer than buses. The average life expectancy of a light rail car is about 40 years, that of a bus, about 15 years. Hiroshima in Japan still ran a couple of streetcars that survived the nuclear bomb in 2010. Vehicles with internal combustion engines and tires just aren't that reliable compared to trains. So though buying the fleet is more expensive at first, as the vehicles last longer, the higher cost over the long term is a wash, maybe an advantage in favor of LRT.

But the main advantage of LRT is labor. As I have said, a single driver in a LRT train can drive around 500-600 passengers and more, meanwhile, a bus driver in an articulated bus (bi-articulated buses are rare in North America) can carry about 120 people. So to carry the same number of people, you need 3 to 6 times less drivers in LRTs than in buses. This isn't negligible, drivers represent about 40% of the cost of running buses in Montréal currently (going by the STM budget). So everything else being equal, LRTs could offer an operating cost per passenger of maybe 30% lower than buses.

So here is a simple example of this dynamic, supposing that:
  • A BRT costs 15 millions per kilometer to build, the operating cost is around 4$ per user
  • A LRT costs 60 millions per kilometer to build, the operating cost is 3$ per user
  • A subway costs 250 millions per kilometer to build, the operating cost is 2$ per user
Here is what the average cost per passenger would look like over 30 years, depending on daily ridership
Cost per passenger according to daily ridership
If you sum up all the spending over the years, you get this:

Total 30-year cost depending on daily ridership
In this thought experiment with rough estimates, BRTs are only cheaper than LRTs until a ridership of around 50 000 people per day, above that, LRTs are cheaper in the long-term, and with very high ridership, subways even become cheapest of all.

So yes, BRTs have lower inital capital costs, but if the cost to run them overwhelms the transit authority's operating budget, they may face cuts to service.

But this experiment is only valid in the developed world, where capital is plentiful and labor is expensive. In the developing world, the opposite is true: they lack capital access but have plenty of cheap labor. So what happens if I adjust costs for the developing world?

Cost per passenger according to daily ridership, developing world

Total 30-year cost depending on daily ridership, developing world
Here, BRTs remain cheaper at much higher riderships and seem like a much better deal, especially since most clients are captive clients who can't protest rough rides too much as they don't own cars to seek refuge in.


Personally, I find LRTs more promising for the developed world, LRTs are a good solution for a high-capital, expensive-labor world, whereas BRTs are more suited to developing countries as they are low-capital, high-labor solutions to transit. The future of full-fledged BRTs in North America isn't promising, at low riderships where they are cheaper than LRTs, the low riderships don't actually justify the investments in BRTs, limited buses with bus lanes could achieve more or less the same quality of service. When ridership increases to the point where investing in BRT would be really better than just limited bus services or BRT-lite (like the SBS in New York), you'd probably be better off investing in LRT anyway, at least in the long-term. 

Turning your back on LRT to build BRT with high expected ridership sounds penny wise and pound foolish. That is what Ottawa has realized, as it had the most complete BRT system north of Mexico for the past 3 decades and is now realizing its mistake and converting it to LRT in its downtown area. A move that is supposed to save the transit operator tens of millions every year in operating costs AND provide higher capacity and speed.

Still, my preference is to HRT. I think that highways have demonstrated that building fast transport links prior to urban development is a winner as infrastructure is cheaper when it is built and it allows the transport link to shape developments successfully. So my ideal would be to build regional rail running on their own rails, not as a mere afterthought on freight rails. These trains could then provide a skeleton for transit-oriented development later on, and most of the lines could be built on the surface. But that's mainly just a dream, inspired by the Japanese mode of development.


  1. My BRT experience is mostly in Mexico City with a little more in Boston.

    The One line on Insurgentes runs articulated busses with capacity of 100-200 riders running about 75 busses per hour at peak, making a capacity of 15,000. The busses are crowded and often people on the platform wait for two or three busses to stand on one that isn't already full; there's little chance of getting a seat. In 2008 and 2009 I watched this line upgraded block by block overnight from warped and beaten asphalt to deeply entrenched reinforced concrete. The city wasn't confident enough to build proper concrete lanes originally but needed them badly very soon. Lucky for Mexico City, efficient overnight concrete work is a local specialty.

    The One line probably can't be upgraded to light rail because train control systems won't support the consistent lead times necessary to beat the busses' capacity. Busses, of course, don't have train control systems and drivers are happy making 48 second lead times that engineers and their insurance companies would never approve.

    But the One line has enough ridership and unmet demand to justify conversion to full underground subway service at least from Doctor Gálvez to Indios Verdes, most of its length. In fact, that upgrade is badly needed. The extension to the south climbs up several grades over 5% south of Perisur so I don't know what kind of system would work there and integrate with any kind of rail.

    In 2009, the Two line running E-W was added and isn't as crowded. Then in 2011 the Three line that mostly parallels the Green subway line was added. The Green line is the busiest of the Mexico City subway, so the extra capacity is useful and the northern end of the BRT line turns off and runs on the alternate route the Green subway might have taken if it had been planned differently in the 1960s, so that provides mass transit out on a different spur.

    In 2012 the Four line extended to the airport. On the Three and Four lines I see an essential point. A few buildings moved, a new overpass, and a dedicated shunt through the airport (already legal for some service vehicles) would make an enormous difference in trip length on those lines. They would eliminate waiting at three or four busy intersections or cut three sharp turns out of the routes. But BRT is sold as cheap because all that has to be done is building stations and marking exclusive lanes, so that doesn't get done the way it would for rail.

    For instance, the airport route has to pass two terminals, so it needs to leave one and get onto a road through a twisty onramp or left turns and then get through traffic into the other. Regardless of your terminal, you must run through this gamut twice on a round trip. It makes the airport route uncompetitive when compared to the speed of taxis or walking to the metro station. A simple infrastructure improvement with 200m of dedicated roadway would allow an exclusive cutoff to avoid the wait and make the BRT very fast, but the investment was promised to be small.

    Now Mexico City has among the cheapest taxis and lowest transit wages of any first world city which creates funny incentives, but that probably won't last forever. And there already is a light rail system running, but it's very limited and not exciting.

    1. My light rail transit experience is mostly from Denver, Salt Lake City, and Mexico City.

      Light rail gets enormously more investment in rights of way improvement, exclusive routes, straightening, and quality vehicles. But the experience isn't really all that much better than a good BRT running over reinforced concrete. The worst thing about busses is jerking out of traffic and descending to the curve while you wait for people to creak on and off the stairs and pay and then stopping every block and twisting around a maze of slow corners. BRT already eliminates most of that with pre pay and level boarding. The advantage of LRT is mostly straighter rights of way, but BRT could have that, too.

      The LRT systems in Denver, Salt Lake, and Mexico City all run mostly on old railroad and utility rights of way or dedicated corridors built for LRT. BRT is stuck with existing roads.

      Mexico City's BRT stops every 400m. That's far, far too close spacing. Just going to 800m would really improve service and make infrastructure cheaper. I've seen ordinary city busses really improve by refusing to stop on even numbered street corners, too. Transit planners always put in a lot more stops than is optimal for riders, as if density of stops could substitute for quality of service. LRT and heavy rail suffer much less from the extra stops impulse from planners.

      So I think BRT, LRT, and subway each have their place. I enjoy riding LRT the most but the BRT experience could often be drastically improved by spending heavily in a few focused spots to open bottlenecks with direct, exclusive BRT cut through lanes just like you would do with rail. Everyone is building too many stations on BRT lines.

      I'm fascinated by that Bogotá map and I want to ride that system just to see how it works. Less than 5% of Mexico City's streets are even 20m building to building, though, much less with sidewalks. All the Mexico City BRT lines run through narrower streets than that except the One line.

    2. "But the experience isn't really all that much better than a good BRT running over reinforced concrete"

      That depends on people, I think. I have yet to experience a motion sickness, but a large chunk of my acquaintance can get carsick or bus-sick, especially if they read (which takes out the "work during transit" advantage) or aren't facing forward. Trains don't wiggle and seem notably more comfortable for such people. For me the comfort advantage of a mass transit train is modest, but for others, big.

  2. I am an independent transit/transportation planning consultant who has lived and worked in Ottawa for 24 years. One of the issues the Transitway (the marketing name for our BRT busway) has never been able to shake as the amount passenger traffic increased, the labor costs to O.C. Transpo (the transit operator in Ottawa) greatly increased. Yes our Central Transitway has a peak hour passenger level of 10700 passengers/hour/direction (p/h/d) however, it takes 185-200 buses/hour/direction to do this. Could they improve the efficiency of the BRT operation mode and reduce the number of buses per hour without losing passenger capacity? In short yes but it would only delay the inevitable rise in bus numbers and would force massive wholesale changes to the entire bus network and how it operates. The main issue became the cost of operating that many buses and the effect on bus availability for the network outside of the Transitway. The ever increasing numbers of buses, bus drivers and especially bus mechanics (whom are worth and cost their weight in gold to have) made Ottawa a city of 925,000 and area population of just 1.2 million (if you include both sides of the Ottawa River), need one of the largest bus fleets in North America. The city could no longer afford the current fleet size, let alone the planned future number of buses needed. Thus, the switch to LRT began, made easier because the Transitway was designed to rail not road standards, easing conversion to rail. The change has been expensive and painful but appears to be on track for Stage 1 to open in September 2018 and Stage 2 somewhere around 2023-24.