Is BRT the best answer for bewildered city planners?


I was asked by the New York Times to comment on what future city transportation plans should look like. In a short piece, they could not repeat all I said, so I will expand a bit here.

My main advice to cities is that nobody, including myself, has the exact answer on how transportation will look in 2030 or beyond. (They are making plans for 2030 and even 2040 now.) Because we can't know, my advice is to design to be flexible. Design to be able to change your mind.

To do this, I call upon the internet's "stupid network" principle that changed the world. Simple infrastructure, and intelligence at the edges. This means stupid roads and smart cars. The roads should not dictate the solution as much as it can be avoided. Bare pavement carries pedestrians, bikes, motorcycles, scooters, vans, cars, robots, trucks, buses and vehicles yet to be invented, while rails carry only trains, trains or trains. If building rail, consider having rail embedded in pavement -- or someday even paving over the rail because of how expensive embedded rail is.

My personal feeling is we don't need new infrastructure. Incredible capacity gains in the roads are possible. Look at any road and count the empty seats going by, and then the empty slots, even at rush hour. Replace each car with a 16 seat robotic van and consider the capacity. It's immense -- in theory.

Not everybody believes that, and they see new infrastructure as the only choice. I understand why they might be skeptical of grand visions.


As an alternative, I suggest they consider new variants of Bus Rapid Transit. BRT is already popular in many places around the world, and growing in the USA. It is effective, but vastly cheaper than the darling of transit planners, Light Capacity Rail. For some reason transitophiles like BRT but they love Light Rail, in spite of its high cost, atrocious energy efficiency and light capacity.

BRT can mean a lot of things. In some vision it involves taking out large numbers of regular lanes, but this is not always necessary. There are simpler visions which focus on a few special features such as:

  • Synchronization to traffic lights to assure green lights
  • Mixture of dedicated right-of-way and mixed lanes, regular and with priority
  • Larger "stop" areas and multiple door entry/exit for high boarding throughput
  • Elimination of ticket purchase or fare-gating on the bus and other causes of boarding delay
  • Offline stops -- the bus does not block the road lane when it is stopped (unlike a train.)

My main reason to suggest BRT is that you aren't going to go very wrong by doing it. Whatever shows up in the future can probably make use of whatever infrastructure you build for BRT, whether it's lanes or tunnels or offline stops. It can probably make use of it even while the BRT is still running, so transition can be gradual. That's much harder for subway, LRT or other forms.

In addition, because BRT is low cost, a city won't waste that much money on wrong decisions that are obsoleted by new technology.

Hybrid BRT lanes

One useful option for BRT, as well as at-grade LRT, is to create hybrid lanes which allow use by well behaved private cars. Many cities already allow taxis into dedicated bus lanes so this is not that radical an idea. Robocars can make a promise of good behaviour which could justify their entry into such lanes, but even regular human drivers with something as simple as a smartphone might be allowed in.

In particular, I propose allowing other vehicles into the lane behind the bus. When a bus goes by, let them use the extra capacity behind it. Either for free, or paying a toll (like "HOT" lanes on the highway.) They must not, however, get in front of a bus. In fact, if a bus ever needs to hit the brakes because a car is in front of it, the camera on the bus would take a picture and that licence plate would get a fat ticket.

Of course, full carpools (including those from UberPool) might be highly encouraged to be users of these lanes, along with vanpools and other efficient road users.

To keep people from getting in front of a bus, a smartphone app could know the locations of the buses, and warn drivers to get out of the lane or pull over into spots where the bus can pass them. If a driver is approaching a bus stop where the bus is just about to pull out, they would be warned, and know to slow or pull over so the bus can pull out, and then they can safely follow that bus.

When I first described this this 14 years ago in the 7th ever posting on this blog I worried about the risk of congestion when drivers in a panic try to get out of a lane when a bus approaches. Today the smartphone can give plenty of warning, and robocars can be even better behaved.

Offline stations allow tremendous capacity, another benefit of the smartphone era. If passengers have indicated what station they want, you can generate buses which only stop where they need to and start having a bus every couple of seconds if you need it. Of course, if you do that, there would not be any room for the cars, but so that's because the road is already efficient. 40 person buses going by at 2,000 per hour produces 80,000 people/hour capacity in one lane -- more than any train line, and it's all seated express capacity, too. Off course you need much more offline stop capacity in this case, but it's doable. This is what will also eventually happen with a repurposed BRT corridor lane in the future robotaxi era.


Agree with the premise of your argument. A fellow in Australia developed some open source software to compare light rail to an autonomous taxi approach, which inspired this article that suggests a good way for public transit to adapt is to focus on frequent bus feeder lines with low-speed alternatives for the last-mile.

We adapted that approach to the idea of creating dedicated lanes on freeways as the feeders. At major intersections over the freeways, transit oriented developments/transit centers would be created (covering the freeway).

and more detail on the freeway cap is here:

While there is nothing specifically wrong with a TOD, that is an investment presuming robocar transit will not arrive. Robocar transit is not station oriented -- rides can start and end at any address. In Robocar-oriented development, you can produce high density development, but you include lots of easy boarding locations away from traffic, and even allow electric vehicles to enter the complex and even enter buildings for high resident convenience. Connection to express lanes is of course a virtue for any robocar oriented development but the cost of building it over the highway is probably too high.

We are having this debate in New Zealand at the moment. The anti-car people are completely in control and they want a light rail line going from the central city through the airport in Auckland. At a cost of $2.6 billion NZD (~$2billion USD) currently. But presumably substantially much than that. Their argument is that our central city streets are already so clogged with busses coming and going from all over the place that light rail is the only solution. This is the first article I read that suggests anything opposite to that.

We we already have a busway coming from the North Shore going to the central city. Runs beside the motorway. The only things allowed on it are buses and ambulances (there was a proposal from the previous pro-car government to allow electric cars but that didn't go very far). The speed limit is 80 km an hour and it is completely separate from surrounding traffic. It's at ground level with stations about every 2 km apart.

The politicians are saying it needs to be converted to light rail before it reaches capacity in 2030: My understanding about that capacity limit was that they physically couldn't fit any more buses on. But you are arguing that light rail has less capacity? Having never heard this before it's quite intriguing. I think one of the arguments for LRT is because the trains are longer you can get people on and off more quickly. Can't think of any other reason why they would be arguing that.

Thanks for the very interesting article! It's so nice to read something different than what we get over here in New Zealand.


Light Rail trains can be longer than buses, so in terms of capacity per driver it can be the winner. You can get a lot of people in a bendy-bus but not as many as in the small train. However, today, we could easily build "bus trains" where extra buses, as many as you like, follow a main bus, slaved to it, on a dedicated busway. And of course, in just a small number of years, autonomous buses.

If you have the drivers, BRT today can surpass the LRT in terms of capacity per dollar of cost. Alas, in many cases, the capital budget to build the system and the operating budget to pay drivers come from different budgets, so the wrong economies are sought. LRT is almost always electrified, while for bus the temptation to use diesel (again, operating and building budgets) can be strong, making it less green. It can, of course, be electrified or hybrid.

By 2030 the technologies to give the busway capacity are a pretty sure bet. In fact, by that time we will have fully automated transit fleets with smaller sizes.

One nice thing about smaller sizes is you get express trips for almost everybody, like the modern elevators where you enter your floor before getting in the elevator. Buses do offline stopping far more easily than trains. So you go to the station, punch in you want stop 14, and it tells you to get on bus "C," when then only stops at a few stops on the way to stop 14, possibly none at all if that's a popular stop. Shorter travel time is of course fantastic, but it also increases capacity as each bus does its loop faster.

Ottawa has had "BRT" since the 1980s. We are building LRT now. It's really Light Metro, in a tunnel at a higher cost.
BRT suffers from a huge number of inefficiencies that prevent from scaling. The major one is that it's flexible, and it can go into any kind of $*@#$ car-oriented development, so there is simply no pressure to design intelligently. People want to sit down, and want a single A->B bus, with no transfers, because the transfers are unreliable.
If you want to see BRT at it's worst, please come and visit Ottawa. Stay at a downtown or Kanata Hotel, and take BRT from downtown to the Cisco or Newbridge buildings. Or from Kanata to a government building. Please make sure to do this in early February.
BRT is transit for poor people who have no power: you'll never see a politician who voted for it on it.

It has a wikipedia page: Bus rapid transit creep -- when people turn a BRT slowly into something closer and closer to regular bus service.

I agree that is a real thing, that the political realities must be factored in. However, in the article above, I propose BRT to cities precisely because it will be the easiest thing to turn into the transit of the future, whatever that is. I don't expect the BRT to last, nor do I expect LRT or heavy rail to last. People who put in LRT will spend immense amounts of money for something shiny that doesn't last nearly as long as they planned.

I have similar fears for HSR and conventional airports as well. Both because I would do an airport entirely different today, and because we may start to see VTOL passenger liners which fly 99.9% of the route as conventional fixed wing jets but can do vertical takeoff and landing, no longer needing runways. It's all going to change.

Brad Before you say another word about BRT, take a look at the bus-train crash in Ottawa and tell me if you still want the bus. The same thing with the recent Saskatchewan hockey team crash. Commercial buses over 15,000 pounds are exempt from crash worthiness standards. School buses are not.

Trains are huge and heavy and can't stop very quickly. They can be hugely destructive and as such, we can't let them mix with traffic and have to have gates when we dare an at-grade crossing. Happy to be rid of them for the reason you show.

Add new comment