Building Melbourne's Useful Network Part 78: Saving 2 million annual passenger minutes - The Frankston line stations we could possibly do without

The Labor government is getting more brazen with its level crossing removals. Although the policy to remove crossings has bipartisan support, the method did not, with noisy but largely baseless opposition to elevated rail. This is despite advantages at some sites including lower construction costs, shorter disruptions, better ground-level permeability and retention of vegetation. 

Political considerations were a major element of projects designed and built during Labor's first term (2014 - 2018) when it had a slim majority. However the 2018 landslide gave it increased confidence. Hence last month's announcement that as part of grade separations on the Ringwood line, Surrey Hills and Mont Albert (which are 800 metres apart) would be replaced by a single station in the middle.

This would be in the seat of Box Hill, one of Labor's more marginal seats. Although to be fair that was an unexpected win. Labor could lose it and similar surrounding seats without losing office. Especially given the coming electoral redistribution which will transfer seats from the low-growth east to the high growth west, north and south-east where Labor is (mostly) stronger.  

What if boldness prevailed on the Frankston line?

This post is a hypothetical. What if such boldness (some might say brazenness or even arrogance) was in force when Frankston line grade separations were being designed? We might have got an elevated rail solution for Edithvale/Chelsea/Bonbeach. Then all three stations would have provided a similar view that Carrum has newly become famous for. Instead of a bay glimpse at just one station it would be over a longer section of track. And there would be better road and walking permeability with less of a need to close or merge some crossing points. That's locally important as most residents must cross the rail corridor to reach local shops and the beach. 

Even more radical (at least for train users) is the location of stations or even their existence. Changes to the latter was not on the agenda for Frankston line grade separations. However maybe it should have been, especially if we are as interested in building a first-class rail network as improving road traffic flow. 

The Frankston line trend: more frequent but slower

Frankston line service trends over the last 25 years can be expressed in frequency and running time. One trend has been positive while the other has been negative. 

On frequency it's been unambiguously a good news story. Daytime trains are now twice as frequent as they were in the early '90s, with a weekday increase from 20 to 15 to 10 minutes. Weekends went from 20 to 10 minutes for most of the day, with an even bigger increase on Sundays. And evening service, still lagging on most lines, will increase from every 30 to every 20 minutes on the Frankston line at the end of this month.   

While there's less waiting, travel times are a story of continually decreasing speed. Frankston passengers once enjoyed off-peak express running through Hawksburn, Toorak and Armadale. That privilege got swapped with the Dandenong line for admittedly sound reasons (including it being busier). A late morning off-peak trip from Flinders St to Frankston took 58 minutes in 1997 before that change. 

The new Southland Station also added some time, though the timetable was altered well before the station was built. Subsequent timetables have added a minute or so each time in running time. Padding timetables can be seen as catering for increased patronage, improving reliability, making punctuality targets easier to achieve or reducing operator lateness penalties. Today a weekday morning trip from Flinders Street to Frankston takes 66 minutes. However this will increase to 67 minutes with the January 31 timetable. This is for a distance of 43km from the CBD.

Competition with roads

The speed limit for traffic in local streets in Melbourne is 50 km/h. Local shopping strips, with high pedestrian activity, often have a 40 km/h limit. Yet our suburban trains, on their own right of way, have average speeds measured in the thirties, even off-peak where fewer boarding delays can be expected. The Frankston train line is no exception, with an average 38 km/h speed. Peak express trains exist but speeds are only slightly higher, with a typical 59 minute travel time. That's slower than the off-peak service in 1997.  And to compensate for loading delays the all stations trip has been inflated to 111 minutes during peak times. 

Road, rail's competition, enjoys 100km/h speed limits on vaguely parallel freeways like Eastlink, Peninsula Link, and the under construction Mordialloc Bypass. None of these existed in 1997. These compete with rail for many trips, especially from areas with infrequent, slow or backtracking bus routes or to jobs with poor transit access such as at Carrum Downs or in the Monash precinct. 

To summarise, Frankston line rail travel is unattractively slow, and it's been getting slower over time. And Frankston line patronage has been falling, particularly on the outer portion. Some could be attributed to frequent shutdowns due to construction works but not all. 

The main train lines are public transport's version of freeways yet their speeds are more like local traffic. If a trip needs connections to slow or infrequent buses then travel time blows out further. This is particularly an issue for Frankston due to the extension of near continuous suburbanisation, mostly on two increasingly narrow corridors, a further 20 to 30 kilometres beyond where frequent rail stops. Public transport travel time to Melbourne from much of the Mornington Peninsula is rarely under two hours, even during off-peak times.  It can be faster to get public transport from Melbourne CBD to Ararat than to parts of the peninsula, despite the latter being about half the distance. 

Around Australia, similar comments may apply for the Gold Coast, whose train is also relatively slow. Mandurah is nearly twice the distance from Perth as Frankston is from Melbourne yet its trains are vastly faster due to wide station spacing. While Mandurah's local buses are infrequent (like Frankston's), the rail they feed offers a vastly faster trip than driving, something that the Frankston line does not necessarily do. The trade-off though is that with many more stops, including shopping centres like Southland, local connectivity for radial non-CBD trips is better than for the Mandurah line.  

Optimum station spacing for the Frankston line

Which brings us to what is ideal for the Frankston line. More stations equals more walk-up coverage but less speed. Fewer stations equals less walk-up coverage but more speed. The latter can still have wider coverage with a good feeder bus network but rarely are the buses as frequent as trains. Plus there's a transfer penalty that increases door-to-door travel time. You also need to bear in mind that walking is the main access mode to most Melbourne suburban stations and distance is a key factor of whether the service is considered useful. But when stations are too close they eat into each others walking catchments and don't add much new catchment. 

One approach is to keep all the stations but have express services. This reduces frequency at the stations skipped for a given number of trains. Skipping one or two stations doesn't greatly reduce travel time (although it somewhat reduces peak capacity on a two line system) while skipping three or more is not practical if a high frequency operates without signalling upgrades and, ultimately, extra line capacity. 

Where stations are very close so that large parts of their catchments are within a short walk of two stations on the same line, stations can eat into each other's catchments. With only a small unique catchment it's worth considering whether all stations should remain open since many passengers would almost as easily be able to walk to stations either side. When you reduce the number of stations you can speed the service without having to introduce confusing skip-stop express running (tried but abandoned in Perth) or reducing service frequency at some stations due to express running.

As I mentioned before, the Frankston line is an increasingly slow railway that is not necessarily attractive to those with the choice of using a parallel freeway. With good 7-day daytime frequency and reasonable (but still not great) evening frequency instituted from January 31, the next major priority for improvement should be travel time. 

Improved travel time involves various small and large changes. These are the sorts of things that a government focused on big infrastructure can overlook. And some, like reducing the number of stations, can provoke a political backlash, even though the vast majority of passengers might benefit. 

Removal options

Which stations on the Frankston line might one remove? 

Station usage could be helpful information. Numbers are available from this Philip Mallis blog post. Though note potential volatility in some years due to line closures and bus replacements. 

Bonbeach, Aspendale, Kananook and Edithvale are in the quieter group. Bonbeach is 1.3km south of busier Chelsea and 1.5km north of Carrum (before it was rebuilt and moved south). Aspendale is a similar distance north of Edithvale which in turn is 1.7km north of Chelsea. However Aspendale is distant (2.6km) from Mordialloc.

Proposed grade separations will move Chelsea south and Edithvale north, making spacings more uneven than previous, with Edithvale closer to Aspendale. Edithvale loses denser residential catchment than it gains to the north (part being a golf course) as well as access to the area's main bus (902). Hence both it and Aspendale may lose catchment from the grade separation, although Edithvale gains somewhat by users no longer having to wait for trains to reach the platform (Aspendale doesn't as it, like Chelsea, has an underpass). 

Kananook is further from stations either side. Seaford is 2km north and Frankston is 2.5km south. 

None of these distances are super close by Melbourne station separation standards. As comparison, Riversdale and Willison on the quiet Alamein line are just 600 metres apart.

Assuming station walking catchments are 800 metres, and they are seen as a rough circle around each station, stations need to be 1600 metres apart for them not to have any overlap where stations eat into each other's catchment. Of the distances quoted above, the closest two are Chelsea and Bonbeach, though this spacing is likely to widen as Bonbeach is moved significantly south and Chelsea moved slightly south. The movement of Edithvale north opens a gap between it and Chelsea (which is populated) while overlapping Aspendale near the unpopulated golf course.

We also need to think about those away from the rail line whose walking distances may be increased if a station is moved (or even removed). A factor here that's important is that stations align with the main east-west road. In the case of Edithvale, Chelsea and Bonbeach all three stations are being moved further from their main east-west road. This makes access to them less legible for those coming from 1.5 to 2 km away. The same has happened for Mentone, which I regard as a flawed project due to the new station's less convenient position south of the main activity area. 

To summarise, if we view the Frankston line as a walk-up medium speed railway (some might argue we shouldn't with something like the fast Mandurah line, being a pace-setting example) then the station spacing on this part of the line is not overly close. However that is not so further north, where spacing is more like a metro system in a dense European city. European cities are more compact and they don't have the metro doing double duty as a regional rail type service (as the Frankston line tries to, being the main feeder to areas as far as 70km from the CBD, via the infrequent 788 bus).

Patterson

Patterson is another of the quieter stations. It is 800 metres south of Bentleigh and 900 metres north of Moorabbin (for which big things were planned in the '50s but little happened). A few shops surround the station. It is served by no buses. For a long time it suffered as being the first station exclusively in Zone 2, with much of its catchment walking north to Bentleigh to take advantage of cheaper fares. 

Patterson opened in 1961, during a time when train patronage was falling. It would have brought trains closer to parts of newly developing Bentleigh East. However one could query whether it added more patronage than was lost by Frankston line trains being made slower. Closing it could save passengers one minute each way, or two minutes per day. It doesn't sound much but it is when multiplied over a year. Such savings would be at the expense of the relatively small number of Patterson Station users who would need to walk in the very worst case 800 metres more (and usually much less).

Patterson features an elevated platform somewhat cut off from its surrounds. The road already passes underneath so it has not needed to be rebuilt as part of the level crossing removal program. The removal of higher Zone 1+2 fares may have brought back patronage that was previously lost to Bentleigh. However too much of its walking catchment is overlapped by either Bentleigh and Moorabbin. Still, little has been spent on the station for years, and it shows! 

McKinnon

The same cannot be said for McKinnon Station. Relatively minor McKinnon Rd got grade separated around the time that busier North and Centre Rd did. That meant the station got rebuilt into something grand. Watch the opening festivities here. 

Should McKinnon have been rebuilt? It's busier than Patterson. And for that matter stations down the line like Bonbeach and Edithvale. However it is much closer to stations either side as per the map below: 

The 800 metre spacing gives it very little unique catchment given that stations either side are 1600 metres separate. The unique catchment would have been even less if we had provided Bentleigh with an extra northern entrance. Unfortunately multiple station entrances is something designers too often scrimp on. 

Unlike Patterson, McKinnon does have a bus (the 626). However if the station was closed it could be rerouted via Bentleigh. This would probably make the route busier due to a new connection to the thriving Bentleigh shops. And it would provide a direct route to Carnegie, another developing area. 

Benefits and disbenefits 

Station closures to speed travel typically have asymmetrical benefits. That is many passengers gain slightly while a small pocket of passengers near the closed station lose a lot more. The latter can make station closures bad politics as people who have had something taken away make more noise than those who would stand to gain. 

Here are some numbers to help quantify whether closing the two Frankston line stations with the most overlapping walking catchments would be for the greater good or not. 

Patterson

2018/9 annual station boardings (Patterson): 354 000

2018/9 annual station boardings (Moorabbin - Frankston): 8 317 000 

I will assume the following: 
* 100% of access to Patterson is by walking (an overestimate)
* Patterson passengers will walk an extra 5 minutes on average to Moorabbin or Bentleigh
* Expressing trains through Patterson will save an average 1 minute in travel time
* One-third of Moorabbin - Frankston passengers will benefit from above (those who don't are either on express trains or are using the train for local trips) 

Annual extra minutes for Patterson passengers: 1 770 000 minutes extra
Annual saved minutes for Moorabbin - Frankston passengers: 2 772 000 minutes saved
Overall annual saved minutes: 1 000 000 (but could be higher)

McKinnon

2018/9 annual station boardings (McKinnon): 450 000

2018/9 annual station boardings (Bentleigh - Frankston): 9 398 000 

I will assume the following: 
* 100% of access to McKinnon is by walking (an overestimate)
* McKinnon passengers will walk an extra 5 minutes on average to Ormond or Bentleigh
* Expressing trains through McKinnon will save an average 1 minute in travel time
* One-third of Bentleigh - Frankston passengers will benefit from above (those who don't are either on express trains or are using the train for local trips) 

Annual extra minutes for Patterson passengers: 2 250 000 minutes extra
Annual saved minutes for Moorabbin - Frankston passengers: 3 132 000 minutes saved
Overall annual saved minutes: 900 000 (but could be higher)

Conclusion

Patterson station should be demolished and McKinnon should not have been rebuilt

Very roughly closing Patterson and McKinnon stations could save Frankston line passengers nearly 2 million minutes per year. That number may well be higher as it doesn't count those boarding at stations north of those considered travelling towards Southland/Frankston. Neither have I counted Stony Point line passengers  (around 100k boardings/year) as I'm assuming that their travel is primarily local/Frankston/Southland. Plus non-walking access (normally faster with a wider choice of surrounding stations) hasn't been considered. Hence overall time savings could well be more. 

Two million saved minutes per year is over 30 000 saved hours or 1300 saved person-days. That's a big number! Transport infrastructure and service planners should be considering these sorts of benefits when designing projects and services. That includes smaller infrastructure and signalling projects that can save a minute here and there (eg train speeds on the approach to Frankston often appear slow), along with more efficient operational practices. If these are done time savings could be even higher. 

A bolder Frankston line grade separation program could have included works that closed Patterson and not rebuilt McKinnon. Bentleigh could have got a northern entrance to assist current McKinnon passengers. And dingy Moorabbin could have got a needed facelift, with better access to the north, including across South Rd, to assist current Patterson passengers. 

Other offsetting project could have included (a) improved 7 - 9am Sunday morning train frequencies, especially outbound direction, (b) full time staffing at Ormond Station and (c) major bus operating hours and frequency upgrades on major routes such as the 630, 703 and 824, local routes (like 625, 626 and 823) and a new East Boundary Rd SmartBus and (d) better, ie multimodal, passenger information at stations. The bus changes, especially would benefit a much wider area, particularly suburbs just beyond walking distance from station such as Bentleigh East. And they would aid a lot of short distance local travel. 

Wider station spacing where walking catchments eat into one another would provide all-day faster service on the Frankston line without having to compromise frequency by having some trains operate express and others stop all stations. Also coverage would be substantially preserved. 

Politically acceptable alternatives, assuming we are planning on keeping Patterson and do not wish to close the near-new McKinnon, could be an all day two tier express/stopping service with each tier operating every 10 minutes. This would be operationally dearer but give some travel time savings. And frequency would be compromised per operational dollar spent, especially at night when driving is so much faster. But if we're not closing too-close stations, finding alternative means of speeding the service is essential given the current slowness of Frankston line travel and competition from road, which is faster for some trips it shouldn't be. 

Comments are invited and can be left below. In particular discussion on the pros and cons of this and other station closures or mergers is encouraged. 

See other Building Melbourne's Useful Network items here