As Melbournes train lines increase their frequency to keep up with demand the passengers certainly appreciate the shorter waiting time and less crowding, but could we do even better? Although a few years old the train station patronage data from PTV allows some insights into how passengers as getting to the stations and from that we can estimate what effect station closures or timetable changes might have on the network.Included in the daily passenger volumes per station is a breakdown of the mode of travel passengers used to arrive at each station. As a simple estimate of a station closure we will assume that they continue to use the same transport mode but travel the additional distance to the nearest station along the line at a slow speed for that transport mode (3km/h walking 40km/h driving etc). This over estimates the additional transport cost as few passengers detours would be the worst case of travelling up to the closed station first.
The other side of the equation is estimating how many passengers travel through each station, here the assumption is that half of all the passengers getting onto stations further out from Flinders St Station than the station in question will travel through it. This could be improved with access to the complete passenger data but for a first look at the data it will still be illustrative. Closing a station saves all those passengers the time it takes to stop the train, wait, and return to speed, approximately 2 minutes lost time for a typical station. When this is balanced against the costs of asking all the passengers of a specific station to go somewhere else it looks like this:
A strong correlation is seen in the distance between stations and the possible savings in transport time for the system, for some stations we could ask those patrons (a small percentage of the passengers) to extend their journeys by a small amount and save the majority of people more time overall. Since these numbers are a little rough we wont pull out any specific stations with figures but instead discuss the underlying characteristics of each line.
South Morang & Hurstbridge
This is a largely simple example of stations far too close together for a mass transit system, having many stations with spacings of 500-700m close in towards the city while the majority of passengers are already on the train from stations further out. Even with the high 60% mode share of walking to the stations they are so close that closing almost every second station on the South Morang line appears more efficient. The stations are so close that even of those people who would need to travel further after closing their stations, the majority of them would have a shorter total journey after counting the savings in train travel. The Hurstbridge line does have more practical 2km station spacings towards its end and those stations are viable as is through distance and a lack of passengers already on the train to delay.
Sunbury, Craigeburn & Upfield
The densely packed Footscray stations on the Sunbury line stand out as a place for rationalisation, with the close in South Kensington station serving a similarly small percentage of passengers compared to the rest of the line and delaying the majority of passengers for the convenience of a few. This contrasts with the Craigeburn & Upfield lines that have no stand out stations which could make large improvements until they all converge onto North Melbourne.
Passenger access modes at North Melbourne are majority transfers with over 70% reported as passengers originating from another train. As the last stop before the city loop where passengers will have greater choice of transfers, and its close proximity to other stations this model suggests that North Melbourne should be mothballed. Shifting more transfers to Southern Cross station is currently difficult due to poor connectivity between the platforms, but the shuttered pedestrian subway that remains between the platforms could cope with the demand as they are no more narrow than the facilities at North Melbourne.
Lilydale & Belgrave
These lines have all their problem stations on their shared section but they span a range of different causes. Starting from the furthest out at Laburnum, although 2/3rds of the passengers walk to the station they’re a small number compared to the stations around them only 800m away. A similar situation arises at the sequence of Chatham, Canterbury, and East Camberwell stations all serving tiny numbers of passengers while being close enough to each other to suggest merging the first two, and then closing East Camberwell.
As the trains progress towards the city and fill up the model suggests even busier stations such as Auburn and Burnley could be closed with resultant reductions in overall travel time.
Alamein & Glen Waverley
Very few stations on these lines show much benefit in closing, the already bypassed East Richmond being the obvious case that stands out. Beyond this Glen Iris is a different case as its weighted by 60% of passengers arriving by bus and car, modes with lower costs to move to the next station if the parking spaces could be moved the short distance to the adjacent stations.
Pakenham & Cranbourne
With these lines more than half the passengers arrive by car or bus and the widely spaced stations further out are well matched to their use, except Sandown Park which is a defacto parking lot. There are two exceptionally quiet stations on these lines, Officer and Yarraman that simply through their limited patronage are inefficient and closure would be an improvement overall. The close placed tuple of Hughesdale, Murrumbeena and Carnegie could do with rationalisation to 1 or 2 stations between them. Once these lines merge into Caulfield they join the larger MATHS group to be discussed together later.
Swinging back the other way with only 40% of passengers arriving by car or bus the Frankston line still has stations that could be consolidated even with the larger walking passenger share. Aspendale and Patterson are close enough to major stops and serve small enough groups that they could be closed, and the close spaced tuple of Bentleigh, McKinnon and Ormond could be rationalised to 1 or 2 stations efficiently. Then the line joins into the larger system from Caulfield:
This group between the interchange stations of Caulfield and Richmond condense a huge number of trains and passengers from the Pakenham, Cranbourne, and Frankston lines to stations with few boardings. The innermost station of South Yarra also has significant interchange volumes from trams and trains but the tuple of Armadale, Toorak and Hawksburn are ripe for consolidation. This is matched with the current timetables running express trains non stop between South Yarra and Caulfield to skip these stations but more granular and intelligent patterns could be an improvement.
Werribee & Williamstown
The same pattern appears from the Sunbury, Craigeburn & Upfield lines with North Melbourne jumping out as a candidate station which could be closed to the benefit of the system as already discussed, otherwise the Williamstown line appears well balanced. Werribee line has some poorly patronised stations of Aircraft and Altona which show up as candidates for closure.
An absolute model of a train line with the stations uniformly spaced just far enough apart to spread the passengers but close enough to maintain high levels of active travel with 70% of passengers walking or cycling. The shorter line has fewer passengers to delay with the frequent stops and works excellently as it is.
There are many stations identified here which are inefficient uses of the train network. Considering only time spent and externalising the other costs of the added transport for those passengers who would need to travel further paints a slightly distorted view but one that is rational for the network operators. If stations were to be closed they would need to be retained for future reopening as land use and communities change, as tempting as it is to sell off the land for a quick profit.
Trying to sell the idea of inconveniencing a few to the grand benefit of all others is a tough political standpoint but there are some underlying benefits to the network by reducing the required staffing, maintenance, and with faster journeys either fewer trains to maintain the same throughput, or more throughput (higher frequencies) while maintaining the same number of trains in operation. Its this last point which could be the convincer, although some people would need to travel further to their station there would be more frequent and faster trains for everyone including them.
But what if I told you there was a subtle way this could mostly be achieved without closing stations?
As more frequent services are added to the lines in Melbourne there have been some ideas floated around better ways to run the network than simply adding more of the same scheduled services. Daniel Bowen considers benefits of running trains continuously without a timetable so that time spent waiting for a train is minimised, but the travel time remains the same. I’ll propose flipping this around and increasing waiting times for a faster service.
Instead of (or along with) closing stations outright create a new timetable that splits the high frequency lines into two (though it can be extended to more than 2 patterns to better match demand) interleaved services, say the Frankston A, and Frankston B. These would stop at all stations at the large activity centres but then alternate stops at the minor stations removing 10 stops from each timetable and reducing travel time by 20 minutes end to end. The trains run with the same 10 minute headway as before but alternating between the two stopping patterns adding a worst case 10 minutes additional wait time for the promise of a reduced travel time for the majority of passengers greater than the additional wait time. This runs “faster” trains within the existing limitations of signalling, speed limits, and tracks, maintaining the current headway and without the need to choreograph complex express running trains passing others.
This could replace the poorly balanced system currently in place on the Frankston line during the evening peak where comparatively empty trains depart Flinders St to terminate at Mordialloc (the same 20 minutes faster than running to the end of the line as a proposed interleaved run) or Carrum while the only trains to run to the end of the line are express to Caulfield-Cheltenham which fill to capacity with passengers getting off between Cheltenham and Moridalloc/Carrum to save 10 minutes, squeezing out customers wanting to get off beyond Moridalloc/Carrum. A timetable interleaving stations is able to balance the demand much better between the services and provide faster running to all passengers.
These faster interleaved services offer savings in trains and drivers that can be made while maintaining the current throughput, opening up vehicles to add capacity to the network. It only becomes viable once a certain service frequency is reached but we’re right there on the tipping point for the Frankston, South Morang & Hurstbridge lines.