Sunday, May 10, 2015

The quality bus

For those who have not read my previous posts on the issue of Bus Rapid Transit, let me sum up by saying that I am not a fan. I think it is an innovative solution, but one that is not appropriate for the developed world where capital is abundant and labor is expensive. A proper BRT is not that much less expensive to build than a surface-running LRT, which offers better qualitative service and has higher capacity, but can be affordable where roads already exist that can be simply reserved for buses and where transit operators already have huge bus fleets. On the other hand, it is a valid solution for the developing world, where capital is hard to find, but labor is cheap.

So personally, the idea of spending hundreds of millions on building proper BRT in the developed world strikes me as a particularly bad idea. For example, the Cleveland Healthline is often considered one of the only decent BRT line in the United States, it cost 200 million dollars to build, has a maximum capacity of about 1 800 pphd and an average speed of 10 mph, not that much faster than regular buses. 

But that doesn't mean that buses have no place in transit, or that current local buses are the best we can hope for. It is important to reform bus networks to make service better.

So, here is my vision: what I would call the quality bus, which should involve a fraction of the cost of single-lane BRT yet offer similar speed and capacity.

What would be the quality bus?

Quality bus involves identifying certain trunk lines on a bus network, lines that follow main arterials with an high number of jobs and residents along them. These tend to be the more frequently used lines, so have higher than average frequency, however, as riders are numerous, the lines' speed falls and buses become crowded and unreliable, which has the effect of discouraging potential riders from taking the line.

These trunk lines would be selected for service upgrades, focused on the following:
1- Higher, more reliable speed
2- Higher capacity
3- Better visibility

Let's take this down point by point.

Higher, more reliable speed

What slows down a bus most is the number of stops it has to do, especially a very urban bus. Though a bus doesn't need to stop at every stop, the randomness of stops where people embark or alight makes it harder to plan bus service and predict average speed. On a bus line with 20 stops, one bus could have to stop at 12 of them, the next bus could only have to stop at 8, the first bus would therefore be much slower than the second, as a result, planners often plan for this by slowing down all buses to match the speed of unlucky buses.

Therefore, the first thing to do is to eliminate stops so that the average distance between stops is around 400 meters (a quarter mile), maybe a bit more.
Effect of stop spacing on bus speed

This will mean slightly farther distances for users, but it's not too important, the 1 or 2 minutes people have to walk more is more than compensated by faster, more reliable service. If the bus' speed increases on average by 25% (say 12 to 15 km/h), that means that on an average 30-minute trip, there is a 6-minute saving.

Second, reducing the number of stops is good, but it is also important to reduce the time spent at stops, the dwell time. For that reason, adopting proof-of-payment and all-door boarding and alighting is vital for a high-ridership line. That way, people don't have to all queue up to enter the front door and pay or swipe their pass in front of the driver. This system can be done in many ways, one of them, used in New York for SBS and some European cities, is to have machines printing receipts at stops. For a method that is even cheaper to operate, just have card readers at all doors, so people with passes or tickets on cars can use them, but demand that people who pay cash still have to use the front door.

Third, use bus lanes where necessary in order to protect buses from congestion. Having physically separated bus lanes is often the number one requirement for BRT, but it is not essential for a quality bus service. You can have queue-jumping lanes instead or the peak-hour only bus lanes (though this is for political reasons only, peak-hour bus lanes are absurd, if you can take a lane from cars during peak hours when traffic is heaviest, you can take it from them all the rest of the time too).

Fourth, when headways are low enough (below 10 minutes), consider running buses on headways, not on schedules. When buses run on a schedule, drivers are told to slow down if they are ahead of the scheduled time of arrival at each stop. Running on headways means that you don't work on a schedule, you simply plan to have a certain number of buses running the route, planning for a certain headway between buses, and drivers "ahead of schedule" just keep on without slowing down. This speeds up service and can even help lower headways even more.

Finally, bus priority systems at traffic lights are, largely, not worth it. For one thing, bus priority works best for low-frequency lines, and trunk lines should have high enough frequencies that when they need the priority systems the most, they are least likely to be able to use them because the frequencies are too high. This is actually one of the advantages of LRT, that they can simply run longer trains and keep headways high enough to preserve priority systems at traffic lights.

All these measures can be implemented for little or no cost and can have significant impacts on speed and reliability. Of course, they shouldn't double speed, but if you can get a 20-30% gain, that is significant, both for users and for the transit operator who then has lower operating costs.

Higher capacity - use higher capacity vehicles

There are two ways to increase the capacity of a transit line: increasing frequency or increase vehicle capacity. The number of people that can be carried per hour per direction (pphd) is simply calculated by multiplying the number of trips per hour times the capacity of each vehicle. So on a line with a 10-minute frequency and a 50-passenger vehicle, that means there are 6 trips per hour, with up to 50 people per trip, for a sum of 300 passengers per hour per direction.

Usually, many transit operators tend to favor increasing frequency and using the same standard bus everywhere. The reasons for that are simple enough to understand: this allows the operator to simplify its fleet, to use only one model that can be used on any line, and higher frequencies mean less wait time for users.

So I'm going to go against the flow here. With regards to the quality bus lines, use articulated buses.


First of all, we're talking of heavily used lines here, or at least, lines that we want to see heavily used. However, transit users don't arrive at bus stops at equal intervals, their arrival is, for all practical purposes, random. This causes a problem for users, in that the number of passengers per bus will fluctuate, so it's the luck of the draw: will the bus be half empty or full? In that respect, larger vehicles have a major advantage in that they will limit these fluctuations.

OK, a demonstration is in order. I have used my knowledge of probabilities to create a randomizer in Excel. Every minute, there is 36% chance that no passenger will show up at a stop, 37% that 1 passenger shows up, 18% that 2 show up and 8% that 3 do. In one simulation, I've supposed 1 regular bus (capacity: 55 passengers) every 5 minutes shows up and picks up every passenger accumulated at all stops. Then, with the same randomized data, I've used a different supposition, that every 8 minutes, an articulated bus (capacity: 88 passengers) shows up and picks up every passenger. In effect, both systems have the same passenger capacity. Then I've summed up how many passengers each bus picks up in both scenarios, supposing none alight until the terminus, and calculated their load ratio as % of capacity. Here are my results:
Vehicle load as % of capacity, with the regular bus scenario and the articulated bus scenario
Though in both scenarios the average load of vehicles is 83%, the fluctuations seen in the first scenario are massive: some buses are nearly half empty (63% load), others are at or near saturation (one is at 107% capacity, it would leave passengers on the curb). In the articulated scenario, fluctuations are kept much lower, the bus with the lowest load is still at 76% of capacity, while the most loaded bus is at 93% of capacity. There is no over-capacity bus.

What this means is that the articulated buses, with a similar load level on average, guarantee a more uniform experience for riders. The fluctuations are basically cut in half. So riders have to fear much less the possibility of seeing a bus pass in front of them without stopping due to being at capacity, or of being squeezed like sardines in one bus, just to see another one, half-empty, overtake his overcrowded bus.

Second, articulated buses have on average about 60% more capacity than regular buses, yet the data I've found seems to indicate that the operating cost of articulated buses per vehicle-hour is just 20-25% more. So overall, a seat on an articulated bus is 20% cheaper than one on a regular bus, which isn't surprising considering most of transit's costs is labor. I would actually recommend to run buses on the same budget and not "cash in" the difference by running 33% less buses, but just maintaining the same spending level. The point of that is that, just like congestion can depress car use, an over-capacity line, or even a line that frequently has vehicles over-capacity, will push away potential users.

So a quality bus should aim at offering more capacity than the minimum needed, and to do so, running bigger vehicles is a very good idea. Furthermore, to get people to use transit, first you must offer them a seat to occupy, if you are trying to encourage new users to try out transit but all your major lines are at capacity, even if they want to use it, they won't be able to. So quality bus lines need to offer unused capacity to allow for transit use growth along the trunk lines. Using bigger vehicles is a good way to do it without breaking the bank.

Of course, even if that is true, it will still mean higher headways using articulated buses than regular buses. That will increase waiting time... but again, we're talking of high-ridership trunk lines here, that should have an off-peak frequency of 10 minutes, or 15 minutes max. So if articulated buses are run 20% less often to match the cost of using regular buses, yes, the frequency may be 1 bus per 12 minutes instead of 1 per 10 minutes, or 1 per 20 minutes instead of 1 per 15 minutes, but on average, the additional delay to riders is just 1 minute in the first case and 2,5 minutes in the second. That is not a massive difference, and it gets even smaller as frequency increases.

OK, one last point, why not double-decker buses instead of articulated buses? Double-decker buses do offer similar capacity, and are nice for users on the second level HOWEVER, double-decker buses are terrible in urban service, because they have less doors than articulated buses and climbing up and down the stairs is a major bother, so dwell time is significantly increased. Double-deckers should be reserved for express lines, not for high-ridership lines in urban areas. Yes, I know, London uses them, they have an emotional attachment to double-deckers and a psychological fear of articulated buses. It's their mistake, don't copy it.

Better visibility

Finally, there is a last point: the trunk lines that are elevated to quality bus standards MUST stand out. They must stand out on the street and they must stand out on the maps.

One big barrier to transit use is often that people struggle to understand the bus network. Yes, they can go on Google Maps and find an itinerary for a specific time, but they are uncertain about how the system works, where they can go, which line is reliable and which isn't. As a result, the average user doesn't see the bus network as a network, but as ad hoc services that overlap.

Maps like these...
Bus map of Hamilton

Bus map of a part of Houston
...are really hard to visualize for riders, even when they are on grid-like street networks. They see a mass of lines, sometimes of different colors, but it's quite overwhelming. On the other hand, subway maps are perfectly useful, because each line can be relied on to offer fast, high-frequency service all day long, finding a path to your desired destination is exceedingly easy.
Montréal subway map... whoever gets lost deserves an award
A quality bus network should seek to emulate subways a bit, you can even create a similar map and name each stop so as to simplify path-finding. Even better, have an automatic stop announcement system in each bus to make finding one's way easier. Again, this is something that isn't that expensive to do, it just takes some efforts.

Buses from quality bus lines should also be easily identifiable, painting them a different color is the easiest way to do so, and bus stops should be particularly identified. Again, a great way to identify these bus stops and differentiate them from regular stops is naming them. The existence of these lines must pop out, to impose themselves on people, so that they learn to depend on them.

Why can't we do the same for all bus lines? Because if you try, you return to the original situation of chaos and confusion, there are so many lines that people don't know which is which and they can no longer visualize the network.

An example of quality bus: Québec City's Métrobus

Québec City is the capital of the province of Québec and second largest city, after Montréal... a fact that most residents still haven't made peace with. In the 90s, due to a difficult budget situation, the province cut transit subsidies, forcing the Québec City transit operator to rationalize its overlapping bus lines inherited from bankrupted private bus companies. They took that difficult situation as an opportunity to rationalize the network. Services in the periphery were cut, but they created new bus lines following the main arterial streets of the city, running mainly on reserved bus lanes. Probably in the spirit of competition against Montréal (pretty one-sided competition, Montréal doesn't care much about Québec City, which enrages the latter's residents even more), which had its underground Métro, they named these lines the "Métrobus".

These lines were a great success, they attracted plenty of riders, most of which from other local and express routes. Nowadays, the Métrobus lines transport 35% of all riders of Québec's transit operator.
Above, one of the articulated buses serving the Métrobus routes, below, an usual Québec City bus
Some bus stops were decorated into "stations" and named, like this one, called "station colline Parlementaire"
The Métrobus lines have good elements but lack others mentioned here. Though they use articulated buses differentiated from regular buses, have bus lanes and have differentiated stops that are more widely spaced, they lack all-door boarding, proof-of-payment and other elements mentioned here. Still, they have done great things for transit use in Québec City.

And the best thing is.... it took less than a year between the idea and the realization. Setting up the lines took a mere few months.


So, that was my idea for better bus networks in North America. These quality bus lines require far less investment than "BRTs", yet offer most of the same advantages. They still remain of limited capacity in the end, even with articulated buses, they can't offer much more than 3 000 pphd, roughly the same as one-lane BRTs with articulated buses and without convoys. So the big issue with them is that if we encourage people to use them much more, like people use subway stations, then they may saturate quickly. However, for mid-sized cities with decent cores, they are a good solution to increase transit use, at least, that is my opinion.

Finally, here is one thing that is NOT a good idea: WiFi and other similar gadgets. Maybe on long-distance commuter buses, it would be a good idea, but for high-ridership urban lines, this kind of amenities is irrelevant. They are the hardened motorist's idea of what transit riders want. Transit users want reliable, adequately fast service that is also affordable and comfortable. WiFi and other such amenities are way down the list of what they want.


  1. In Zurich, they run buses on schedules down to a bus every 7.5 minutes, and are adamant that even in the 10-20 minute range, clockface schedules with predictable intervals dividing 60 are critical.

  2. In Zurich the maps are typically hierarchical, where the Kantonal maps show only rail lines, the city maps show rail lines and some major bus and tram lines, and each transit stop has a local-scale map which only shows transit lines within 0.5-1.5 hours of walking distance. No map has an overwhelming amount of detail, and it is clear that the easiest way from point a to point b is using the transit hierarchy (and using a bus to rail to bus route).

  3. I know this is quite old, but where did you get those pphpd stats from?

    Guangzhou's at grade BRT system does 30,000pphd. In fact, if we take Siemen's best LRT unit it can apparently take 800 people "ultra crush" which requires 38 vehicles per hour to meet the BRT level. Which is very difficult because LRT systems have a single platform each direction, so each vehicle must wait for the next to finish with the station.

    To be clear LRT has its place, within an urban area for example LRT provides a much higher aesthetic and renewal opportunity.

    However, BRT seems best more like a low cost commuter rail or metro alternative. For example the North Shore BRT made bus service on the North Shore *way* better, but didn't require a new tunnel/bridge for rail, nor an incredibly expensive CBD tunnel/station system.

    It's true the North Shore BRT cost $200 million. But estimates for rail run from $3-4 billion (TransportBlog, based off the Vancouver System) to $11 billion (NZTA, pulled out of their ass). I'd also wager a significant portion of that was spent on future proofing it for heavy rail.

    Also the North Shore BRT runs at 80km/h (50km/h within station zones).

    1. I've got the pphd from various sources and theoretical extrapolations.

      Guangzhou's BRT, like Bogota's, is 4-lanes wide, with express buses skipping stations on the second lane. That's how they can get this kind of capacity, well, that and super-crush loads in buses, like 160 people in a bus designed for 100-110 people (by developed world standards). LRT can equal that capacity on a single lane, if you compare apples to apples, meaning suppose super-crush loads too and shorter headways than most transit authorities are willing to run. Manilla's LRT matches the Transmilenio's maximum capacity without having a second track per direction.

      The bigger problem with this kind of BRT is that you need a 20-meter ROW (excluding sidewalks and car lanes) to build it in an urban area. That's a kind of space that is rarely seen in most cities... thank god. I'm of the opinion that highways have no place in urban areas, and that counts for bus highways like Guangzhou's or Bogota's BRT too. OK, if cities made the terrible mistake of building these immense asphalt rivers that cut their cities in two, reusing them for 4-lane BRTs makes sense, but these huge roads shouldn't have been built in the first place.

      BRT is not really that much lower cost. It has lower initial costs, but requires much higher labor costs. So in the developed world, focusing on BRT is a very bad idea, Ottawa in Canada is paying the price for that.

      But even looking at initial capital costs, it's important to understand that the difference is not really between technologies, but between chosen ROW alignment.

      Laying tracks is no more expensive than building concrete roads to support buses' weight. Rail transit is almost always electric though, so this has costs in the form of power lines and power stations, and rail vehicles are 2-3 times more expensive than buses per capacity, yet have a life expectancy of 2-3 times that of buses, so overall, it's a wash, they cost more initially, but last longer.

      Anyway, if you compare apples to apples, meaning a BRT and LRT on the same alignment, the cost difference is not so great, maybe 30-40% more for rail, which is partly made up by longer lasting vehicles and less fuel costs. The reason why most rail projects are so much more expensive is that they tend to have better alignments, more protected ROW, more elevated and underground portions. Buses can also sometimes re-use existing road infrastructure. When rail projects have the same opportunities, they can be dirt cheap, Ottawa's O-train LRT cost a mere 27 million dollars for a 8-km line. That's because they reused existing tracks and bought diesel-powered LRTs, so they didn't have to build tracks or electric infrastructure.

  4. This comment has been removed by a blog administrator.