A couple of days ago I lent an ear to a conversation between fellow bloggers Patrick Reynolds and Matt L over the fate of train service patterns around Newmarket station. They were particularly concerned with the best way to run a service so that people can travel easily between the west, the city and the south… in any direction.
On one hand Patrick was suggesting it was time to embrace the transfer model and create a real ‘network’ out of our rail lines, a network that services travel across the whole region all day rather than just shuttling office workers to downtown at peak hour. This is something I totally agree with. However he also suggested that running all trains from the west to south would be a good idea to facilitate this sort of cross town travel, and that Britomart bound passengers could simply transfer at Newmarket for the rest of the trip. This set of little alarm bells inside my head. While we do need a cohesive integrated network – and transfers are the cornerstone of such a network – we do need to recognise that transfers are usually an inconvenience and it pays to not break the main flow of travel unnecessarily. Going west to south with all the trains seems to be mostly about fitting the services in around the infrastructure constraints, not designing the network to suit the travel needs of our citizens.
On the other hand it seems Matt L agreed with my thoughts. He thought that the core flow was going toward downtown, and changing that would upset a whole lot of people’s trips for a lesser gain elsewhere. But he did recognise the value of linking the south and west lines together. Matt’s suggestion was to do both, send four trains an hour direct to Britomart, and two trains an hour from west to south via Newmarket. This also set off alarm bells. Do we really want to have two service patterns on each line, where 2/3 of trains go one way, but every third train goes somewhere else? That means a funny timetable with funny frequencies, which makes just turning up at the station and heading off on the next train an issue. It’s kinda funny that proposals like this that intend to open up new origin and destination pairs often have the opposite effect: they make catching the right train infrequent and irregular. This kills of the “just turn up and grab the next train” sort of plans you can make with a frequent and regular service.
So I found myself in the conundrum of both totally agreeing and totally disagreeing with both of them. It really is quite a pickle: effectively we want trains that have a single regular service pattern, yet somehow magically take us directly across each line on the network too. We need services the support the main travel demands headed to the CBD, yet also seamlessly support the various travel patterns across the network. And we need to do this all with limited resources and an understanding that there are only so many trains we can fit through our junctions and stations.
Is this asking the impossible? Perhaps not.
Last week I was lucky to attend a transit network design workshop run by the inspiring transport guru Jarrett Walker (of Human Transit blog and book fame). The workshop was based around a series of fun network planning games where teams were given a large map of a fictional city complete with topography, land use and density data and a fixed budget of transit service and other operational constraints. Our task was to design a network within these bounds to meet various planning goals and evaluation criteria. Through much drawing of lines on maps and fiddling with spreadsheets, we all came to grips with the fundamental tradeoffs and hidden efficiencies that lurk in the geometry of transit networks. At the end of the day my team did quite well. We developed a strategy of planning out a gridded network of high frequency routes and focussed on running them at 100% operational efficiency to extend the network as far as possible. This strategy paid off in most regards: we ended up with a nice logical network that our fictional citizens could easily transfer around to get almost anywhere. In terms of efficiency, coverage and legibility we excelled. However there was one criterion in which we were consistently beaten: travel time. No matter how great our efficiency, we just didn’t have the resources to get enough high frequency service to enough areas to make our transfers invisibly quick. In many cases we had connections taking fifteen or even thirty minutes, and that eats into travel time.
The faster groups took a different approach. They had a strategy of providing a lesser level of frequency overall but coordinating transfers between routes around timed connections at key locations. This is called pulse timtabling, and it allows us to do amazing things. The key to it is that two or more routes are timetabled to arrive at the same stop at the same time, this means that anyone can transfer from one line to another at the pulse point without any time delay, simply by walking across the bus stop or train platform. Transferring takes only as long as a regular mid-route stop, and because each of the routes would stop at the pulse point anyway, there is no delay to passengers who are staying aboard and not making the connection.
It’s quite common in smaller European and American towns as a strategy to provide a good interconnected network with minimal transfer delays, in places where the town cannot afford to provide the high level of frequency needed to make transfers simple. In places like this you have a central hub (usually the main street or an intercity railway station) where every bus line terminates or passes through. Most of the time there are no buses around, but every thirty minutes every bus in the town arrives at the same time. Some people get off, some hop on, and some step out of one bus and straight onto another. The buses all then depart for their routes, and return a thirty minutes later to repeat the process, over and over all day. The critical thing is that it provides a tiny time delay when transferring that is independent of the frequency the lines are run at.
There are some tradeoffs with pulse timetabling. For one you actually have to plan and coordinate a timetable across the services you want to pulse. In Auckland that should be feasible with the rail network and perhaps the ferries (the Waiheke buses are actually pulsed to the ferry already), but perhaps not on the rest of the buses (not until we have the new PTOM system in place anyway) which are timetabled and run by various independent operators as independent routes.
Secondly you need to run a consistent clockface timetable. If one of your pulse routes runs every fifteen minutes, all the connecting routes need to run every fifteen minutes too. If you want to drop the service to hourly on one line to safe money, they all have to drop to hourly or you lose the timed connection. This also has implications for efficiency, you have to plan at a network level which in some cases means an individual line might be quite inefficient. Imagine you need a bus route to turn around every hour to meet the pulse connection, but the route itself take an hour and ten minutes to cover. In that case you’d either have think about cutting the route short to make up the time (losing coverage and paying passengers), or you need to pay to put on extra buses to cover the whole route (which also means they’ll spend most of their time sitting around empty waiting for the timetable).
The third major constraint is you need a fare system that allows for transfers. We’re not quite there yet in Auckland, but we should be soon. On the rail system it could be quite straightforward to extend the existing paper ticket stage system across all the lines. For example if a trip from Kingsland to Britomart is one stage, and Britomart to Greenlane is two stages, they could simply say that Kingsland to Greenlane is two stages also. All that would require is a little transfer ticket to show the clippie you had already paid on the first train.
While there are some constraints here, the benefits are huge and with a little effort we could see timetabled connections in Auckland. Going back to our Newmarket conundrum, a pulsed connection could be the answer we need to meet our impossible goals. So how could this work?
Well to start with we’d need to assume the western and southern line ran to the same frequencies at the same times of day, which sound quite reasonable. Then it would simply be a case of shuffling one timetable back or forward until they meet at Newmarket at the same time. So every fifteen minutes during the peak a citybound western line train heading to Britomart arrives at Newmarket platform 3, and at the same time a southbound southern line train arrives from Britomart arrives on platform 4. Both trains open their doors and people can transfer from the western line to the southern line by simply walking across the platform to the waiting train. A few minutes later the opposite happens: a westbound train arrives from Britomart on platform 1 at the same time a train from the southern line arrives on platform 2. The doors open as usual and people can hop on or off either train, or connect between them. Again because both lines stop at Newmarket anyway there is no delay to passengers on the regular service, and your overall time from west to south or south to west is exactly the same as if the train actually ran through that way. And at the end of the day making a transfer from one train to another across the same platform is about as hard as shifting from the couch to the lazyboy on the other side of the lounge, so it’s hardly even counts as losing a one-seat-ride.
So is this our solution? Well it could be. A pulse timed cross-platform connection at Newmarket probably represents the best option for excellent west to south rail connectivity while still keeping a frequent and reliable ‘turn up and go’ service pattern along the main route to the CBD. The only issue I can see is at the Britomart end. With only twenty train slots an hour across the network and very busy trains, we don’t really have any spare slots to hold trains around to fit the pulsed timetable. I guess it would come down to whether the transit times happen to work out in favour and an efficient timetable could be put together. At the end of the day, if we can wriggle the timetable to give a timed connection at Newmarket then we should do it. It would cost nothing but a little planning yet gain an almost seamless link between all the stations on the western and southern lines. That’s a huge benefit for almost no cost.