Thursday, March 12, 2009

Direct running on the Frankston line - it's worth a go

A Herald Sun article reports a 'secret' plan to make Frankston trains run direct to Flinders Street instead of via the City Loop. The Opposition has slammed the idea on the grounds of the forced transfers that will occur at Richmond.

Before rushing to condemn, is it not better to step back and first ask whether the current timetable is the optimum for Frankston line commuters. Or could things be done better, with direct running one of the tools to add extra needed capacity (and frequency)?

Victorian Transport Plan, Page 53

The first thing to remember is that the Frankston line is part of the Caulfield group and shares City Loop access with lines from Pakenham and Cranbourne. The Pakenham and Cranbourne lines are now Melbourne's busiest, serving major outer growth areas. Their patronage now beats Belgrave/Lilydale, both of which have better infrastructure and superior weekday frequencies.

Frankston Station itself has a huge catchment population though the Mordialloc - Seaford portion does not (water to the west, vacant land to the east and limited buses). Patronage growth is somewhat slower than the Cranbourne/Pakenham lines due to fewer new housing estates and the lack of strong mid-line trip generators (eg a university or a major shopping centre). It's still highly patronised, but comes off second best against faster growing lines which get first pickings of any new services added. Also notable is that all three lines form the Caulfield Group, which typically has about twice as much late running than the other groups. Line-by-Line performance comparisons are here.

Then there's service levels. Peak frequency isn't particularly high. Scheduled intervals between peak direction trains on some parts of the line can range up to 17 minutes, ie inferior to the standard 15 minute off-peak service. The actual number of peak trains is little different from 30 years ago, with the major recent improvements being off-peak and Sunday frequencies.

Can passengers benefit from avoiding peak times? They can on other lines, but not on the Frankston line as peak service is only delivered during a narrow period (about 1 hour, both morning and evening). This provides no incentive to alter travel as frequency falls away or express services cease during shoulder periods.

Consistency is another feature passenges value. Since the November 9, 2008 timetable consistency has reduced for Frankston passengers as two key peak services were taken out of the loop and given to Dandenong line passengers. This has caused schedule gaps of up to 24 minutes for loop services serving stations such as Glenhuntly or McKinnon. Hence consistency has reduced and these gaps do not constitute a legible 'turn up and go' service. This is exacerbated where services either side do not run or, as can happen in the afternoon, late-running services avoid the loop and run direct from Flinders Street.

The above demonstrates that while the existing Frankston timetable may have sufficed with lower passenger numbers in the past, it is heavily strained in 2009. Any affordable ideas for improvement, no matter their source, should therefore be considered sympathetically, rather than dismissed as 'too hard' or 'controversial'.

What are the possible improvements that could come from direct running?

The first benefit is for the large proportion of passengers do not use loop stations at all, and get off at Flinders Street or Southern Cross. These passengers will all receive faster service, with the time saving similar to that of making a service run express in the suburbs (but without the problems, such as reduced track capacity and achieveable frequency). Those already using express services would see further travel time reductions.

While the percentage of direct passengers might be about half the total peak loading, it's likely to increase over time. This is because the city's two main growth areas (Southbank and Docklands) are nearest the non-loop stations. The areas nearest the loop stations grew most in the 1980s, but in the 2000s growth is swinging elsewhere.

Secondly, and most importantly, is the possibility of improved service frequency by making train operation more economical by removing 'time spent in loop' and thus increase the number of trips a train can make in a peak period. Better frequency means improved capacity and legibility as timetables become simpler and more consistent. Even customer complaints about delays and cancellations would fall as customers treat the service as 'turn up and go' rather than worrying about precise train arrival times or having to run for a train (due to the fear there won't be another for 20+ minutes).

With some increase in the number of available trains it might be possible to develop a clear two tier service pattern of local and express services operating over a longer span. Currently express services on the Frankston line have gaps that vary between approximately 12 and almost 30 minutes, with about an hour between first and last express. Stoppers in the pm peak are roughly every 12 minutes, with all now running to Frankston.

The train saving by not running via the loop has already been touched on, though not quantified in terms of achieveable frequency increases. However every little bit helps, even if it's (say) two extra trains per peak and improved shoulder peaks.

In the afternoons it may be possible to obtain further economies by running local services terminating at Cheltenham and Mordialloc, as is done in the morning and used to be done in the afternoon. While this may reduce frequency south of Mordialloc it boosts train efficiency by getting it into the city sooner to do another peak trip. Those south of Mordialloc should be content with an express train every 10 minutes (or ideally 8) over a much wider 'window' than at present. Meanwhile, those north of Cheltenham should be glad if their current 8 to 17 minute morning service (12 minute average) was boosted to a flat 10 minutes, and over the moon if 8 minutes (as was provided in the 1970s) could be achieved.

Also gaining will be those who make inner-city trips such as Caulfield to Footscray; they will no longer need to change trains or cope with unharmonised headways as the same train will run through to Laverton/Werribee. Fast cross-suburban travel will help to reduce the perception that trains are only good for CBD or 'same-line' travel. Admittedly this is a much smaller proportion of passenger numbers than CBD passengers, but the vastly improved service direct running would provide should generate extra patronage.

The Frankston line cannot be treated independently of the Cranbourne/Pakenham lines but direct running of the former should free loop 'slots' for additional Cranbourne/Pakenham trains and hence ensure any extra services can be run via the loop (aiding consistency). It is this line that presents the greatest challenges to schedulers due to its high patronage, population growth, limited infrastructure and need to share with country trains, so the additional 'wiggle room' should be welcome.

What about existing users of loop stations, which direct running would make transfer? In the morning this should not be a problem since there will remain frequent services through the loop from Richmond and Flinders Street. Some trips will take longer but improved frequency and more consistent patterns in the suburbs is a major sweetener.

The evening though is a little harder. This is because transferring from a less frequent local service to a frequent loop service imposes less transfer penalty (and milder consequences of a missed connection) than the reverse. Nevertheless if efficiency gains permit frequency increases so unique stopping patterns are consistently 8-10 minutes apart (or better) then even these problems should be mitigated as all parts are 'turn up and go'.

The above is not intended to be a thorough analysis of train scheduling, and there may be difficulties in implementing the service patterns and frquencies suggested. However it is intended to lend support to plans, 'secret' or otherwise, that could improve service on a line that needs it.

If the Frankston line is to be made to run direct, it would be desirable that this apply to all trains (including off-peak and weekend), not just an added few. An 'all or none' policy would assist network legibility, maximises frequency and provides new cross-suburban travel opportunities that current service configurations tend to hinder. The experience with the Werribee line has shown that despite some doomsayers passengers will accept a change provided other benefits such as better service frequencies are provided.


Loose Shunter said...


Great analysis as always. Unlike some other commentators, you grasp the interconnectedness of transport and land-use with the link between the non-loop stations and the centres of development for the CBD.

Keep up the good work.


Somebody said...

Curious - do you how long the Dandenong line has been busiest for? Not surprising given the fair low pop growth along the Ringwood corridor.

I assume Frankston would come somewhere after those two, but it gets a fair bit less daytime patronage.

With the Loop issue, some of the main trip generators are halfway between stations, hence many pax could use FSS just as easily. Like with the Clifton Hill loop rotation, most Parliament users started going to FSS.

No complaints from me about the idea and agree it should be full time. Seeing as Werribee is getting 10 min offpeak headways later this year, assume they would need to extend to the Frankston line for an even pairing.

Peter Parker said...

Somebody - I think Dandenong overtook Ringwood about 4-5 years ago but I don't have the figures to confirm.

Frankston would come somewhere after those two. Not necessarily No 3; Craigieburn and Sydenham would be up there as well.

Frankston really excels for weekend and night patronage; but during the weekday off-peak it's lowly patronised like Sandringham and Werribee for reasons we know.

I can't help commenting on the proposed 10 minute off-peak Werribee frequency.

While I like better frequencies network wide, the opportunity cost of giving Werribee a 10 minute service is very high as better patronage potential is available by boosting other lines first.

The same is also true of upping the Frankston line to 10 minutes off-peak (assuming no station at Southland) compared to the Sydenham and Craigieburn lines which need it more.

Upping Craigieburn and Sydenham (both of which have much higher off-peak patronage and larger mid-line catchment populations than Werribee) from 20 to 15 minutes off-peak would provide far more 'bang for the buck' and headway compatibility with the Caulfield and Burnley groups.

This could be awkward with Werribee still on 20 minutes, but if the latter was taken out of the loop, through-routed with Frankston then it too would still benefit (though only to 15 minutes not 10 minutes).

Altona could become an issue, but if the worst happened and three relatively quiet stations dropped to 30 minutes off-peak (until duplication), the 'greater good' argument of an almost network-wide 15 minute service standard (compared to 10 minutes off-peak on one quietish line) is compelling.

Altona can also be compensated in other ways, especially if the opportunity is taken to rationalise and improve local bus routes after Route 903 (which duplicates part of 411/412 in the area) starts soon.

Loose Shunter said...


The rise to prominence of the Pakenham line in metro rail patronage in recent years shows the foresight of the Federal Government predicting the growth focus along Melbourne's south-east corridor in the Building Better Cities program of the 1990s.

Even if the growth has really hit its straps in the last 5 years, the wisdom of having Cranbourne electrified in advance of the development is a great example of an attempt to integrate transport and land use planning in Australia.

The Feds pushed for Cranbourne electrification and the upgrade to stations on the Dandenong-Pakenham corridor against the wishes of the Victorian DOT/DOI which wanted to extend electrification to Sydenham and Craigieburn instead.

Somebody said...

On the topic of that area, I've been curious for a while about when Dandy-Paki first got a proper suburban service - yes I know they commenced in 1975, but AFAIK it wasn't the extension of every second Dandenong suburban, just a shuttle. Any clues?

I've never thought of Dandenong - Cranbourne as a planning success in that way before, but compared to others where rail has come late, putting in a basic extension was a smart move to cement in a transport corridor for the growing area.

Peter - agree on Werribee & Frankston both being less deserving. I think with the former, the railway only serves as a long-haul corridor out of the general area which explains the lower daytime patronage.

I think the topic of population growth is an interesting one and should be factored more into transport planning. You say there is limited population in the Mordy-Seaford corridor; this is not surprising with the bay on one side, and a narrow band of suburbs on the other. What is the likelyhood of any more new estates going up at the other side of Wells Rd which would add to the catchment?

Peter Parker said...

Peter Parker said...
Somebody, the issue of new suburbs in the area was mentioned in the bus review meetings. From memory the highest growth in the City of Kingston was going to be Waterways and Heatherton.

A major problem is that the 'edge developments' along Wells Rd do not relate at all to the established rail-based suburbs to the west due to many missing connections and an urban structure too coarse grained for anything but the car.

This is largely due to an awkward road network and limited local buses. For example, Aspendale and Bonbeach lack a really direct bus/pedestrian/cycle route to Aspendale Gardens and Chelsea Heights.

Nevertheless more could be done with buses linking these areas to the main stations of Mordialloc, Chelsea and Carrum, all of which have roads to the east.

Another rail catchment growth area is Carrum Downs. Though the bus service has improved greatly in the area commuters must still backtrack to Frankston which would discourage some due to long travel times. Extending local routes to Carrum Station via Thompsons Rd would help in this regard. Similarly Seaford has no significant feeder route to the east, yet is suited to a service to Cranbourne via Carrum Downs.

Unknown said...

The Frankston line would really benefit from a permenant two-tier operation, with local trains running from the City to Moorabbin, and limited express trains from the City to Frankston (stopping after Moorabbin)

Peter Parker said...

Damo, I'd agree, but only subject to more resources that don't cut service frequencies to skipped stations.

Up to a certain point frequency is better. Beyond that point expresses are better. But paradoxically it gets so popular that expresses are harder to fit in. Off peak though there's lots of local trips and expresses don't help that much, and their users likely value frequency more.

I could go on for ages, and do so in a later post!