One of the most important concepts in public transit is how different transit lines can connect to feed and distribute passenger flows across a city. While some concepts like feeder (and distributor) buses are well understood — buses bring passengers into a train station or transit hub to be moved on large vehicles to major destinations, or in the reverse, buses distribute passengers from a centralized station to (ideally) areas of lower density that don't have a direct higher order connection; the lack of terminology to describe how other types of connections can amplify the usefulness of public transit probably limits discussion of them.
One concept I find myself coming back to again and again is the ability for a transit service to move more people when it has more interchange connections. In this article, I'm going to refer to this as the “loading multiplier”, but if there's a more formal term that exists, I’m all ears! I’ve talked about this before in videos, but I really want to dive deeper into the implications for ridership and network design.
What is the “loading multiplier”?
I think the concept of a loading multiplier first occurred to me when I was looking at transit ridership data many years ago. What really stood out to me was the enormous number of people moved by various deep-level London underground lines despite the small trains used on that network. I think a big part of why London is able to move so many people on lines that have relatively small vehicles is that the lines are very long and thus provide convenient connections to an enormous number of people and destinations.
The natural question is: if the trains are small and tons of people are living around the lines, how can we manage to actually move more and more people without becoming completely saturated? While on one hand London has pretty great metro operations and extremely respectful number of people through very old and cramped stations, lots of systems with much larger trains operate similar frequencies to London and don't manage the same level of ridership. This is where the loading multiplier comes in.
What makes London's network so powerful is that it truly is a network. There are generally several ways to travel between any two points, and this allows the city to take advantage of loading multipliers.
When you think of a London underground line in isolation, it generally travels from the suburbs to the city centre and then back out to the suburbs. If you imagine typical passenger flows, a train travelling into the city is likely to pick up passengers on one side, drop them off in the centre, and then pick up more passengers in the centre and take them to the suburbs. Of course, some passengers will get on mid-way into the city and stay on until mid-way back into the suburbs, but generally speaking this is the predominant loading pattern. What's important about this pattern is that it basically suggests that a train travelling the length of a line is probably not going to reach capacity more than twice, and so you can use that to approximate how much ridership a line can max out at. You can think of the loading multiplier as the number of times a train can fill up before emptying back out.
Now, imagine we are talking about real-life London. In this case, obviously passengers aren't only travelling between suburbs and city centre and the reverse, but there are also lots of connections along radial lines that allow passengers to interchange with other services. This means that a train can potentially empty (or partially empty in the real world) many times on each journey, and the more times the train can empty, the higher the loading multiplier and potential ridership.
But what about the passenger kilometres?
It's important to note that I called this the loading multiplier, because the number of passenger kilometres isn't actually changing. If a train is generally pretty full from end to end, it doesn't matter how many passengers got on or off — you're still hitting roughly the same number of passenger kilometres.
At first this might seem kind of bad because you're putting more strain on the trains and stations to hit the same passengers/km numbers. However, I actually think it's very good in the ideal situation where trains are mostly loading and unloading over just a couple of stations at a time; the train line is serving an enormous number of journeys within a given passenger kilometre number — each individual segment is providing trip value.
Essentially, each subsegment of your line is providing a valuable connection or distributing passengers to locations that they would've otherwise needed to find an alternative means of transportation to. Now, of course, your entire network isn't going to operate at this level of congestion (any system running at full tilt for long is unlikely to be reliable or high functioning; slack is virtually always needed to provide space for “moving parts”), but ideally the most expensive parts of it are so that you're getting maximum value for the expense of infrastructure you had to construct.
Looking at this model of “loading multipliers”, I think an interesting question to ask is how you can design a network that maximally takes advantage of them to move more people on each piece of infrastructure.
Taking advantage of Loading Multipliers
There are a few good patterns which, when designed into transit networks, can take good use of loading multipliers.
One of the most obvious patterns is local and express service. When you look at — say — a local/express New York City subway corridor, the local service can often have a higher loading multiplier by transferring many passengers onto faster expresses. There’s an obvious barrier here though, and that’s the fact that the local and express trains in New York are always going to be of similar capacity, limiting the degree to which local passengers can be shifted to an express (and ultimately constraining capacity for popular fast express trips). This is why in my eyes the Paris RER or the Elizabeth Line is a much better instantiation of express service, because the comparatively larger trains allow more passengers, and also because the reuse of surface corridors means these projects can be quite cost-effective per passenger moved. On this, the New York model of local and express subways fares much worse because it’s more expensive to build and leaves less room for high loading multipliers.
Of course, the natural place to look at loading multipliers is with loop lines, which are often designed to maximize loading multipliers. This can be done with many interchanges and also trying to ensure longer journeys along a loop are better served by other more direct services.
But why? Well, on one hand, more heavily-used assets are going to be a better value, but also likely more efficient. It’s also sometimes the case that providing a particular service on a particular corridor or at least expanding capacity is very challenging or expensive. Creating new services and interchanges that can take riders off of parts of a network that they are simply transiting frees space for passengers that actually want to go to a destination along that network segment. This approach also allows the creation of networks served with small trains as in Madrid, because parallel corridors that passengers can transfer to provide better journey quality and faster trips. Building new parallel corridors allows metro service to be refocused on local and last-mile trips, which tend to also enable higher-loading multipliers.
Depending on where you create a potential “new” loading (by emptying a vehicle out), you can also create space for more passengers down a line. For example, Toronto’s (I’m contractually obligated to mention Toronto in anything I make) Ontario line will massively reduce loading of Line 2 at Pape station, and this creates the room for dense development and new passengers west of the interchange heading into the north of Downtown Toronto.
Ultimately, high loading multipliers are something I look at when seeking out good transit networks, because they tend to correlate with high-interconnectivity (enabling many potential journeys) as you need to have places where passengers can load and unload from one service to the next, and high-frequency (enabling convenient journeys) because the capacity and convenience needed to encourage passengers to transfer in large numbers requires it.
Nice work!
I guess you took a more technical approach what economists call economies of scope or economies of network. The more destinations or interchanges on a network, the more usage a network will have. Next to more destinations, the thing is traveltime. Having more interchanges allowing more routes and therefor maybe faster trips than networks with less interchanges. And less travel time means more usage in general.
I like the way the loading multiplier tries to explain in more detail what happens and try to make it usable for network design.
A special focus should be on tangential lines. Lines that do not cross the city center but have the main purpose of making extra movements around city centers possible. Their success is mostly dependent on interconnecting with the other main (radial) lines. So a relatively high loading multiplier needed. But also a good frequency, because with too low frequency or transfer time too long, transfer from tangential lines doesn’t function quite well, as total travel time will be too long, compared to a single transfer between the main radial lines (Moscow circle line had that problem, but also The Hague tangential tram 19).
Since you mentioned Toronto in your newsletter, I figured someone HAD to bring up Vancouver in the comments.
I think the Purple Line in Vancouver (I know, I know, just a twinkle in Translink's and the Mayors' Council's eyes ATM) has the chance to be truly transformative in smoothing out the loading of the Expo Line.
Currently, the vast majority of SoF Expo riders are traveling to DT Vancouver, with a small number alighting at Columbia to head to the Tri-Cities or SFU, a slightly larger number getting off at Metrotown to transfer to the R4 or head to work in that Town Centre, and a very large portion transferring to the 99 at Commercial-Broadway. By the time a rush hour King George train reaches Metrotown it is overcrowded and this capacity crunch does not alleviate until at least Broadway if not Granville. And the SLX will only make this worse. On the surface, the Purple Line would do hardly anything to spread the ridership more evenly across the entire Expo Line.
But something frequent riders of the 241, 246, or any other DT Vancouver to the North Shore bus line know that others might not is how many Expo passengers who alight downtown do not reach their final destination there. My daughter's daycare is in North Vancouver and over half of the daycare workers live SoF and commute via Expo + bus to and from work. I presume this phenomena is not limited to that one business, and that workers all across the North Shore from Dundarave to CapU ride often overcrowded, frequently slow, and always off-schedule buses for the last leg of their journey to work. I think it will be very interesting to see boardings/alightments shift from Burrard and Waterfront to Metrotown when the Purple Line is built (2040? 2050?).