The DART system in action. Source: World Bank Photo Collection. Dar Es Salaam’s New Bus Transit System. 19 January 2017. Photo.
DART, the BRT system of Dar es Salaam, Tanzania, is proving to be transformative. Though just one phase of 6 planned phases is operational, those 21km (13 miles) are crowded daily with passengers. (Or were, at least, pre-pandemic.) As with any new transit system, there have been problems like non-functional technology [1] and expensive fares [2], but these are kinks in a system that will likely work well and eventually cover the entire city.
But if the city is going to continue to rely on BRT as its only rapid-transit solution, it will quickly run into the mode’s limits. Using Dar es Salaam as an example, I’d like to explore the limits of BRT for megacities and how governments in the developing world should approach transit.
Passenger throughput
Years ago, Ari Ofsevit [3] calculated the maximum throughput of a typical two-lane BRT line, finding capacities of between 9,600 and 16,200 people per hour per direction (pphpd). Though four-lane BRT exists, it effectively requires new freeways – hardly compatible with a dense megalopolis. It can accommodate up to 45,000 pphpd.
Image by Ari Ofsevit.
By contrast, according to Ofsevit, a railway system can move up to 60,000 pphpd, if the trains are long enough, on a two-track system, and up to 100,000 on a four-track system. When a lot of people need to get to a relatively small area – like a major city’s central business district (CBD) – rail is the only way to effectively accommodate them. Midtown Manhattan, for instance, is served by 28 subway tracks and 10 mainline railroad tracks. To accommodate the same capacity, BRT would need at least 78 bus lanes – more lanes than all the north-south avenues crossing Midtown. This might be possible, but only in a place with extremely broad avenues and a dense grid-like Manhattan. In a dense CBD with few access points, like in Dar es Salaam, this would be almost impossible.
Speed and geographic sprawl
On top of capacity issues, transit speed limits the geographic size of a city that can effectively be served by a transit network. People are generally only willing to commute about 30 minutes each way to work [4], something called Marchetti’s Constant [5], with diminishing numbers willing to commute longer (just 2% of people are willing to commute 90 minutes or longer each way). For transit, passengers take about 15 minutes to get to and from their stations, limiting how far a transit system can reach.
We can use this to somewhat arbitrarily split up the rings of a city’s development regions:
A central city would extend out to where most people can get to the CBD in 30 minutes or less. Subtract 15 minutes from that to account for transit access time, and you have a circle with a radius of 15 minutes.
A suburban fringe should be at the outer limit of where most people who commute to the CBD would want to live, roughly 60 minutes. After subtracting 15 minutes for transit access time again you get a circle with a radius of 45 minutes.
The exurban fringe, after which less than 2% of people would want to commute to the CBD, would be 90 minutes away, or a radius of 75 minutes on transit. Tendrils would extend further beyond each of these limits to reflect people living closer than 15 minutes from their transit station, but the bulk of the city would be within this range.
While BRT has an average speed of around 20kph (12mph), DART’s express buses run a bit faster, hitting an average speed of 28kph (17mph) instead [6]. Using the radii described above, that translates to a central city about 7km (4 miles) in radius, suburbs with a radius of 21km (13 miles) and exurbs with a radius of 35km (22 miles). For context, this would extend the core of San Francisco to about West Portal, the suburban fringe to Orinda, and the exurban fringe to just past Walnut Creek. Novato, a northern exurb of San Francisco, is beyond the reach, as are the entireties of Sonoma, Napa, Solano, or Santa Clara counties.
Hypothetical developmental rings in the San Francisco Bay Area. Click here for an interactive map.
Other transit systems are faster, of course. By bus, Golden Gate Transit’s express lines average 55kph (40mph) through the North Bay, though slow dramatically once they hit San Francisco’s surface streets. This is close to BART speeds but, unlike Golden Gate Transit, BART moves quickly underground through San Francisco’s dense and congested core. BART pushes the city’s theoretical exurban edge much further, to 69km (43 miles): Rohnert Park, Byron, and San Jose. The other zones of the city grow, too, with the urban core expanding from just a fraction of San Francisco to its entirety and the suburban fringe out to what was the exurban fringe.
Red circles indicate the BART-scale development zones in the Bay Area. The black circle is the BRT exurban fringe. Click here for an interactive version.
In short, BRT functions best in moderately-sized metro areas, whether in terms of population (to deal with capacity limitations) or geography (to deal with speed limitations). In geographically large or very dense metro areas, BRT won’t be able to adequately serve as the primary commuting mode.
What about Dar es Salaam?
DART Phase 1, now in operation.
Dar es Salaam is a moderately large city of 5 million people, with its exurbs sprawling out roughly 30km (19 miles) at their furthest. Its CBD is constrained by some challenging terrain, including a floodplain to the south and west and the ocean to the north and east, and so like Manhattan is only accessible at a few points. As a result, Dar’s transportation problems are already acute, but it is properly sized for its DART system. The Phase 1 line is crowded but still below capacity [1], and the exurban fringe line is beyond the city’s current edge.
The Dar es Salaam CBD, with just 6 roads entering. Image from Google Maps.
However, this will quickly change. The city is expected to double in size in the next 10 years, to 10 million people, and at least one estimate [7] has the city growing to 20 million by 2050 and 62 million by 2100. The suburbs of Tokyo, a metropolis of some 30 million people, extends out some 80km (50 miles) from its CBD. One can expect a city twice that population to sprawl out at least that far, requiring far faster transit than BRT can provide.
Alongside population growth will come economic growth and economic transition. As Tanzania’s service sector grows, the CBD will grow in importance as a commuting destination, increasing the share of the population that needs to commute to it. Even if the Dar of 2050 is half as centralized as the New York metropolitan region it will still swamp the capacity BRT lines.
DART-scale development zones for Dar es Salaam.
In other words, BRT is an excellent start to Dar’s transit system, but it is just one piece of the vast transit network the city will need. Judging by crush loads in Tokyo, Guangzhou, and New York City subways, rail will be an absolute necessity. But just as Tokyo, Guangzhou, and New York could benefit from a robust BRT system, building one is equally necessary for Dar es Salaam to thrive.
Plan for what’s next
In fast-growing metropolitan areas in the developing world, like Dhaka, Manila, Lagos, and Dar es Salaam, governments need to take a systematic and very long-term approach to urban planning.
Plan a city for your grandchildren. Today’s rapidly growing cities will probably not be this way in 80 years, but the infrastructure built today will likely still be in place then and our grandchildren will be responsible for it. Time horizons must be suitably long to account for that.
Planning takes a long time. It was 14 years between the Dar es Salaam’s decision to build DART and the launch of operations, and during that time the city doubled in size. Over the next 10 years, it will double in size again, and so the DART system needed to be planned to accommodate not just the needs of the city of 2.5 million that it was in 2002 but the city of 10 million it will be in 2030 and the city of 62 million it may be in 2100.
It is cheaper and easier to plan from the start than to retrofit what already exists. Alon Levy writes:
One of the positive features of the Paris Metro is that nearly all of it was planned as a whole, and as a result, difficult stations like Chatelet, Etoile, and Nation were built with the intention to have multiple Metro lines serve them. This meant that the stations could be built once, rather than multiple times, once for each line. Two lines, Metro 8 and 9, even run alongside each other for 1.7 km under the Grands Boulevards, which are wide enough for a four-track subway. [7]
Tanzania needs to start planning Dar es Salaam’s rail transit system today, both to serve the needs of the generations to come and to inform the construction of the BRT system of today. To wait until Tanzania can better afford metro construction or until the BRT system is largely complete would be too late.
One should not take this piece to be oppositional to BRT. BRT cheap, can be built quickly (though apparently not in Tanzania), and the need for high-quality transit is immediate. But the government must not ignore a mode simply because it won’t be built soon. The future is coming fast.
Works Cited
All sources are archived. While I prefer live links, please comment below if any have succumbed to link rot and are no longer active.
[1] Alex Malanga, ‘Fresh Move to Sort out DART Chaos’, The Citizen, 20 February 2020.
[2] Rizzo, Matteo. ‘Dar Es Salaam’s New Rapid Bus System Won International Acclaim – but It Excludes the Poor’. The Conversation (blog), 22 January 2019.
[3] Ari Ofsevit, ‘What Is the Actual Capacity of BRT?’, The Amateur Planner (blog), 21 June 2015.
[4] Cesare Marchetti, ‘Anthropological Invariants in Travel Behavior’, Technological Forecasting and Social Change 47, no. 1 (September 1994): 75–88.
[5] Tim De Chant, ‘Marchetti’s Constant, or Why the 30 Minute Commute Is Here to Stay’, Per Square Mile (blog), 13 September 2012.
[6] Ulises Navarro and Gabriel Oliveira, ‘Keeping the Rapid in Bus Rapid Transit: Improving BRT Operations’ (Workshop, Mobilize Dar es Salaam, Dar es Salaam, Tanzania, 2018).
[7] Daniel Hoornweg and Kevin Pope, ‘Socioeconomic Pathways and Regional Distribution of the World’s 101 Largest Cities’, Global Cities Institute Working Paper (Toronto, ON: University of Toronto, January 2014).
[8] Alon Levy, ‘The Lagos Metro’, Pedestrian Observations (blog), 20 July 2017.