Friday, July 17, 2020

Building Melbourne's Useful Network Part 53: SmartBus orbitals - time to split?

The SmartBus orbital routes, introduced about 10 years ago, are easy to explain. They changed the way people saw public transport and made the network more versatile. They made it possible to make long trips without changing (not that many do - most trips are over 2 to 20 km segments). And they were the first (and so far only) attempt at 'big bus' - that is treating buses seriously with large scale frequent service to move millions more passengers.

However the decision to make them long orbitals (rather than shorter 15 to 30 km sections) hampered their benefits due to the following:

1. Length. They're very long. About 4 hours end to end. This makes recovery from delays on the other side of town difficult.

2. Density. Their catchment can range from high to very low density. This variation makes it impossible to arrive at a fair service level for the entire route. The standard 15 min weekday/30 min weekend frequency overservices some portions while grossly underservicing others. Their sheer length makes the SmartBus orbitals the highest used bus routes in Melbourne even if they are not the most productive (measured as boardings per bus operating hour). Shorter routes would have allowed frequencies to be aligned with demand, especially during peak periods and weekends on busy sections.

3. Connectivity. Most train lines that intersect SmartBus routes operate on either a 10 or 20 minute base weekday frequency. Only two lines served by SmartBus operate on a 15 minute weekday pattern. In contrast SmartBus has a base 15 minute service on weekdays. The result is unharmonised frequencies between train and bus and unpredictable connection times.

Having SmartBus every 15 minutes made sense when SmartBus was confined to the eastern suburbs whose trains ran at that frequency. However it's lost relevance since a. the orbitals got extended to areas with 20 minute train frequencies such as in the northern and western suburbs, b. the Dandenong and Frankston lines got upgraded to 10 minute base frequencies, and c. 15 minutes is not really considered 'turn up and go' in a large city, especially one with significant traffic congestion that delays buses.

Weekends have been a problem with the orbitals since their inception. Their mostly long operating hours are good but the usual 30 minute frequency is barely better than the 40 minutes standard on local routes. And it means that the best connections with trains (mostly every 20 minutes) repeat only hourly. Train-compatible 20 minute weekend frequencies should be regarded as a starting point for orbital sections that pass through suburban densities with even some non-SmartBus routes now enjoying this service. However where densities are lower the existing 30 minutes is too much service. Shorter routes would have allowed frequencies to be harmonised with trains and times to be juggled to minimise waiting while also reflecting demand.

4. Overlaps. SmartBuses sometimes inefficiently duplicate and overlap other major routes. The original SmartBuses (all in the eastern suburbs) were done right. There, and in parts of the north, the orbitals replaced existing routes (291, 560, 665, 700, 830, 831, 888, 889) so didn't unduly duplicate.

However much of the 903 orbital west of Northland overlaps existing routes (eg 527, 465, 232, 411) some of which approach or even exceed SmartBus frequency at certain times.  The same applies in the north on sections of the 901 and 902 orbitals. The result is a complex network with low and uneven frequencies rather than a simple frequent network. It might have been cheaper to upgrade the busier of these routes (eg parts of 411, 465, 527) to every 10 minutes, extend operating hours and implement multimodal 'frequent network'  branding than to introduce new SmartBus orbitals. However they wouldn't have looked as elegant on a map and have left some important gaps (eg Heidelberg - Northland) unserved without some route extensions.  

Last year I suggested splitting the Route 903 orbital at Heidelberg. The main purpose of this was to enable the portion west of Heidelberg to be upgraded to every 10 minutes as far as Coburg by merging it with the substantially duplicative Route 527. Another potential benefit could be to allow an extension of Route 903 to La Trobe University to form a direct route from the Doncaster area.

I also proposed another 903 split further west, at Sunshine. This would have transferred long hours Smartbus service from the Brooklyn industrial area (where it attracts low usage) to Footscray via Geelong Rd by upgrading the popular Route 411 to a SmartBus. That would have provided a more useful connection with a better residential catchment. More on that here.

Transdev's try

Transdev also wanted to split the orbitals as part of its 2015 greenfields network. Some of their frequency changes would have harmonised the northern parts of the orbitals with trains. Unfortunately they only considered their routes in isolation and planned to cut service in the west to give to the east. The network was vetoed by Transport minister Jacinta Allan so was never implemented. 

Politically the SmartBus orbitals have been regarded as 'protected species', at least in Labor circles due to the 'nasties' in Trandev's 2015 proposals and lingering sentimentality from before then (given Labor created the orbitals). Still there is merit in splitting them for reasons given before if overall service levels can be maintained or improved in northern and western populated areas (something Transdev's plan did not do).

A look at the orbitals

As background, past local bus network reviews of 10 years ago regarded the orbitals as fixed and did not seek to incorporate reforms to them in their recommendations. This was probably because the orbitals were either new or about to be implemented. And the first (eastern) stages of the orbitals, as mentioned before, were wisely implemented with wider network changes to avoid duplication.

Those earlier bus reviews had a poor record of having their recommendations implemented. Not least because some recommendations were expensive and/or duplicative. Later bus network reviews, such as occurred in Brimbank and Wyndham, were less duplicative 'smell of oily rag' efforts. Future local bus network reviews need to consider the fate of the orbitals and surrounding routes if they are to deliver the best service for the least cost, particularly in Melbourne's north and west.

Below are comments on sections of the three orbital routes. These may be useful if their usefulness is reappraised in future local bus reviews.

Route 901 Frankston - Melbourne Airport

Frankston to Dandenong: The main connection between two key centres. Although there's a lot of open space the size and demographics of these centres makes this segment popular. Not just for people going the whole way but for shorter trips like Frankston to Carrum Downs. Has significant low income catchments in Frankston North and around Dandenong. Also key connector to jobs in Dandenong South. However use as a rail feeder to the Frankston line is constrained by it arriving at an acute angle to it, with passengers having to go south before going north. There is little duplication of service as the 901's introduction coincided with deletion of old 830 and 831 routes.  An improved peak (every 15 to every 10 min) and weekend service (30 to 20 min) would be desirable as per Transdev's 2015 proposal. 

Dandenong to Ringwood: Also an important link. Passes near TAFE and hospital north of Dandenong. Significant lowish income catchment north of Dandenong. The only good north-south bus connection in Knox area, serving Stud Park and Knox City. Apart from the 664 there is only minor duplication between Dandenong and Ringwood as the old 665 was removed when SmartBus went in. An improved peak (every 15 to every 10 min) and weekend service (30 to 20 min) would be desirable as per Transdev's 2015 proposal, although a higher priority would be local bus network improvements in the grossly underserved area east of Stud Rd. 

Ringwood to Blackburn: Roughly parallels the railway. Serves 'big box retail' in Nunawading area. Service levels likely out of kilter with activity. Eg service to midnight might be excessive but there may be a case for a 20 min weekend service given 7 day trading. Case may exist for any replacement route to continue to Box Hill due to that centre's size and role as a transit hub. 

Blackburn to The Pines: This provides a strong train feeder role despite about half of it overlapping the 906 freeway express to the CBD. Case may exist for peak frequency upgrade to every 10 min (as per Transdev's 2015 proposal). 

The Pines to Greensborough: A weak part of the route. Duplicates other routes along Foote St/Reynolds Rd (280/282, 309) and the 902 SmartBus in Templestowe area. High income residential on large blocks reduce patronage potential making the SmartBus service excessive. Transdev correctly wanted to finish 901 at The Pines to remove SmartBus from parts of this low patronage area. More on this here.

Greensborough to South Morang: An even weaker largely semi-rural catchment that should never have got a SmartBus with service through bush until midnight. Only the slavish adherence to the orbital concept ensured that it did. SmartBus in this area guarantees its uselessness as a reliable train feeder as local trains run every 20 minutes versus SmartBus every 15 or 30 minutes. Service is partly  overlapped by some local routes, further depressing usage. Transdev wanted to retain orbital service of this area with a shorter route offering reduced frequency but the same long operating hours.

South Morang to Roxburgh Park: Serves a growth area across the north. With large shopping centres at Plenty Valley and Epping, along with TAFE and hospitals at Epping a SmartBus is definitely justified. However its frequency does not match trains on any day of the week. For several years this portion overlapped a local route (571) but this was eventually tidied up. Transdev wanted to cut service levels here in 2015 but I think there's a reasonable case for a 10 min peak and 20 min weekend service, especially if reforms are made to partly duplicative local routes like the 556.

Roxburgh Park to Broadmeadows: Parallels the train line but serves a key shopping centre at Broadmeadows and has significant low income catchments at Meadow Heights and Roxburgh Park. Again poor train connectivity due to unharmonised frequencies but its parallel geometry makes it role as a feeder less effective though stations are widely spaced in parts. When the route started there was overlap with Route 544 but this was later shortened to start at Roxburgh Park.

Broadmeadows to Melbourne Airport: Overlaps the 902 SmartBus to Gladstone Park then goes north-west to Melbourne Airport to provide it with its only long-hours public transport connection. Poor signage and remote terminal location at airport reduces visibility of service. Again poor connectivity with trains, particularly on weekends when the SmartBus is only every 30 min, inconsistently meeting trains every 20 minutes. Gets steady but not particularly high patronage.

Route 902 Chelsea - Airport West

Chelsea to Keysborough: Replaces old 888/889 in area (which became local route 858).  Attracts steady patronage. Used as a feeder from Waterways/Aspendale Gardens to Edithvale Station however the level crossing removal, by moving the station north, will break this connection, requiring backtracking via Chelsea.

Keysborough to Nunawading: By far the 902's busiest section. Serves major train stations at Springvale, Glen Waverley and Nunawading. Large low income catchment at Springvale. Overcrowding an issue. Easily justifies a 7.5 min peak, 10 min interpeak and 10 - 15 min weekend service especially on the portion between Springvale South and Glen Waverley. Transdev proposed 10 min peak and 20 min weekend service in 2015 on this section.

Nunawading to Doncaster Shoppingtown: Mostly runs east-west, overlapping the busy 907 SmartBus along Doncaster Rd. Provides connections to Shoppingtown and a rail feeder to Nunawading. Few passenger go beyond Shoppingtown on this segment which is why Transdev wanted to finish it there in 2015.

Doncaster Shoppingtown to Greensborough: On paper this links the area's two largest centres but takes an indirect way to get there, backtracking via Eltham. The partly overlapping 901 goes a more direct way to Greensborough but from the further and smaller Pines centre rather than Shoppingtown. The long-established route 293 overlaps a lot of the 901 and 902 orbitals between Doncaster and Greensborough. It's the best and most direct route alignment of the lot but was unfortunately not made part of a SmartBus route. Overall, with its indirect duplicative routes that don't consistently meet trains, services are performing below potential. Transdev's 2015 plans would have partially tidied this up. More on this here.

Greensborough to Broadmeadows: Here we return to more suburban density. There's significant shopping centres and train stations at each end of this segment. Plus trams to LaTrobe University, light industrial area jobs near Settlement Rd and low income areas east of Broadmeadows. The 902 mostly gets 'clear air' as it replaced the less frequent 560 along this alignment. However there is some overlap with the contemptible 566 and poorly used 538.  Also holding the 902 back is the timetable's unharmonisation with trains and the lack of a station at Campbellfield.

Broadmeadows to Airport West: About half of this segment overlaps the 901. 902 then continues south to Airport West Shopping Centre. It partly overlaps the 477 from Broadmeadows that does a roughly similar thing. Melbourne Airport has city and in fact state wide significance as a key destination. Airport West Shopping Centre has a much smaller catchment. Swapping 901 with 902's destination so the 902 becomes the airport bus might be generally beneficial given 902's straighter alignment across the north.

Route 903 Mordialloc - Altona

Mordialloc to Mentone: A scenic but generally quieter part of the 903 as half its catchment is salt water. Its inclusion in the 903 is a legacy of its predecessor, the 700 SmartBus. However it does provide handy connections to schools in the Mentone area. 

Mentone to Oakleigh South: Serves a mix of postwar light industrial areas including jobs near Warrigal Rd, Moorabbin. Attracts steady patronage. Use as a rail feeder to the Frankston line is constrained by it arriving at an acute angle to it, with passengers having to go south before going north.  Transdev's 2015 network would have increased service on this segment, especially on weekends. 

Oakleigh South to Doncaster: This is the busiest segment of the busiest orbital. It serves key stations at Oakleigh and Box Hill, TAFE at Holmesglen and large shopping centres at Chadstone, Box Hill and Doncaster. It also intesects with some tram routes. Many students without cars live and work in the area. Service should be turn-up-and-go seven days per week. Weekend service, in particular, falls short, though there are some trips on Saturdays that are unevenly slotted  between the full route's half-hourly service. Transdev's 2015 plan would have operated a 7.5 minute service on weekday peaks, 15 minutes weekday interpeak, 10 minutes Saturday and 20 minutes Sunday. 

Doncaster to Heidelberg: A direct and handy link across the Yarra via a residential area. Quieter than the previous segment but still worth having. Potential feeder to Heidelberg station but unharmonised with train timetables off-peak and weekends. 

Heidelberg to Coburg: Largely operating along Bell St and Murray Rd this segment serves many closely spaced destinations including stations (Heidelberg, Preston, Coburg), hospitals and shopping centres. Also  links numerous tram lines and quite good residential demographics for buses. I think it justifies a 10 minute 7 day service rather than the current unharmonised-with-train 15 and 30 minute frequency. Unlike in the east largely duplicative routes, like the 527, were not merged into the new service when the SmartBus started. Scope exists to do this to allow a low-cost turn-up-and-go service on this high patronage potential corridor as discussed here.   

Coburg to Essendon: Patronage here drops compared to other parts of the 903. However it provides a cross-suburban service connecting into DFO during shopping hours. 

Essendon to Milleara SC: Here's another case of unproductive network duplication. The 2000s-era decision to route 903 via Buckley St (where it overlaps the 465) to miss the large Highpoint shopping centre is one of the oddest made on the whole SmartBus network. Whereas the 465 operates every 20 minutes (better in peak) to meet trains at Essendon, the 15 and 30 minutely 903 cannot given the 20 minute train service. Its main benefit appears to be to improve operating hours on Buckley St, something that could have been cheaper by adding 465 trips. And a westward connection from Milleara could have been done by extending another route, eg the 406, rather than to bring the 903 over. Transdev in 2015 wanted to reduce services on this segment but not amend the general poor network structure. While undesirable this is understandable since their change was confined to their routes only and serious reform (like discussed here) would have involved other operators' routes.

Milleara SC to Sunshine:  The 903's catchment here is a mix of not much, residential and industrial in about equal proportions. One might argue that service here is excessive but you wouldn't now remove it without providing a useful alternative. 

Sunshine to Altona: The first part of this segment is residential in Sunshine South, where it overlaps other routes. Then its catchment becomes low density industrial. Afterwards land use on Millers Rd changes to commercial and residential. Finally the 903 parallels the train line before terminating at Altona Station. Bus services on Millers Rd are numerous and complex. The number of buses per hour here is excessive (over 8 buses per hour off-peak weekdays) but intervals between them are uneven, increasing maximum waits to more than is necessary. For example it is served by the 903 from Sunshine (4 buses/hour) and the mostly quiet 232 from the CBD (2 buses/hour). Probably the highest patronage potential route is the 411 from Footscray but this only has one bus every 40 minutes. While Millers Rd justifies a SmartBus service (including buses operating until midnight and a frequent weekend service) the industrial area in Brooklyn does not. In contrast the 411 would justify an upgrade. Because it could only concern itself with its own routes, the only instrument Transdev had in 2015 was cutting its own quiet services in the west rather than wider network planning that a proper area review would tackle. A better approach might have been to finish the 903 at Sunshine and use its Altona North area resources to boost 411 to SmartBus standard instead. More on this here.


SmartBus is a 'one-size fits all' service. With minor exceptions it provides the same level of service in dense areas with lots of students, universities and shopping centres as it does through sparse bush or industrial areas. Its headway is almost always incompatible with train frequencies, especially in Melbourne's north and west. Slavishly sticking to the orbitals is leading to overcrowding on some segments and excessive service on others. On the other hand there's a certain simplicity of service that seems attractive. And it's undeniable that the orbitals coming marked a step change in bus services available in Melbourne's middle suburbs. 

What are your views? Should the simple orbitals be kept or is splitting beneficial? Maybe there are other places they could be split not suggested here. Please leave any comments below.

PS: An index to all Useful Networks is here.

Melbourne on Transit bookshop

Favourably reviewed books about transport and cities. Purchases via these links support this blog and its independent reporting (at no extra cost to you).   


Ricky said...

I think some of the splits are going to be rather tough here Peter, so I’ll go through them all in order of what I would do with them:

901; Split in 5.
Obviously Ringwood-Dandenong-Frankston as a seperate route. Patronage justified. Probably a frequency boost.
Blackburn-Ringwood. I’d probably cast this off onto the former 367, it much better suits being a link to Maroondah Hospital, and patronage wise, might do better out of it, drop the frequency back to a harmonised 30 min.
The Pines-Blackburn. If anything I see this having potential for a split yet extended 703 as a straight north-south link from The Pines to Clayton. It’s tough to know what to do with this one, but I can see it being a better use of resources, especially giving access to Monash from the north and Doncaster region.
Greensborough-The Pines. See 902 Greensborough-Doncaster.
South Morang-Greensborough. If anything this should head down Plenty Rd, replacing the 386/7, then follow the metro ring round to Greensborough. Whilst yes, admittedly, losing a connection to Mernda’s residential region would be painful for some, having a frequent smartbus replace it, and provide far better connections than running through cows and kangaroos, speaks volumes and would probably quash some of the want of an 86 extension. Service Broady-Greensborough.
Broadmeadows-South Morang. No change, Service Broady-Greensborough
Airport West-Broadmeadows. See 902 Airport-Broady

902; Split in 3.
Doncaster-Chelsea. Obvious split off route, increase frequency.
Greensborough-Doncaster. Convolutedly, I’d extend 281 back to Eltham, have 582 pickup the old 901 route from Fitzsimmons Lane roundabout on, and run the 902 replacement over the top of the 293. I’m 50/50 about the 901/293 split, which ever has the higher patronage might be better to become the smartbus, but at least the saved resources from the 293 however could go into upgrading the 281. The Greensborough-Eltham segment might be better off being pumped into other local routes, or an extension of the 580, but it’s a much of a muchness.
Broadmeadows-Greensborough. No change, service Airport-Doncaster.
Airport-Broady. This’ll be controversial. I would duplicate this segment, hook it onto your proposed 500, and give it a frequency boost. North-West links are something that is lacking, let alone directly to patronage generators. So having a connection that’s reliable across the North-West, and links the two major transit hubs, would be far better than little ol Airport West.

903; Split in 3.
Northland/Latrobe Uni-Mordialloc. I’m undecided whether this should go to Latrobe (Replacing the 350) or Northland. On one hand, Northland has good patronage to Heidelberg, but not many passengers beyond, on the other hand, a link from Box Hill to Latrobe has been talked about for many many years. Whilst yes, duplication is bad, I’m pro it for this situation.
I would also consider whether running it via Elgar and Burwood Highway over Riversdale/Warrigal to serve Deakin, might be a better choice, with the 732 taking a slight diversion over the old routing. I’ve always felt that 903 not serving Deakin was a lost opportunity, and would be somewhat reasonably easy to rectify.
Sunshine-Heidelberg. Definitely run via Highpoint as has been commentated by you Peter. Other than that, no changes.
Altona-Sunshine. If anything, I’d take the resources out of this section and put them into 471/411. Somerville Rd isn’t a patronage generator at the best of times, so carrying fresh air through the industrial west doesn’t help that portions situation. A lower frequency service might be justified, or an extension of the 417, but even then, if same stop transfer at Altona Gate was provided, it somewhat defeats the purpose.

All in all, they’re a confusing mess that need a lot of work.

Peter Parker said...

Thanks Ricky, Some ambitious splits there that make sense. My thoughts are:

901: Frankston - Ringwood leave alone, boost weekend to 20, peak to 10 min, interpeak to 10 min probably in that priority order

901: Ringwood - Greensborough. Delete. Replace with a. Ringwood - Box Hill route along Whitehorse Rd (every 20 min, based on trains at Box Hill being every 10 min). b. 703 extended north to The Pines SC and boosted to every 10 min. c. New Heidelberg - The Pines route (modified 280/2 probably every 20 min.

901: South Morang - Melbourne Airport. Reroute to Airport West (swap with 902). Upgrade weekend to every 20 min. Needs another route number due to split (909?)

902: Chelsea - Doncaster unchanged. Need large frequency increase in Springvale South - Nunawading portion (run short trips?)

902: Doncaster - Greenborough. Reroute to form more direct corridor not via Eltham eg as per 293. Boost to every 10 min weekday, 20 min weekend. Reform other local routes eg 582 extension to Greensborough. Some other changes in Eltham area to support SmartBus removal.

902: Greensborough - Airport West. Swap with 901 to Melbourne Airport. Boost weekend service to 20 min, weekday peak to 10 min, maybe weekday interpeak to 10 min.

903: Mordialloc - Heidelberg. Unchanged. Frequency boosts particularly weekend to every 10 - 20 min at least between Mentone and Doncaster. Potentially extend to La Trobe Uni.

903: Heidelberg - Sunshine. Renumber as 904 due to split. Every 10 min 7 days from Heidelberg to at least Coburg funded by merger with part of 527. Operate via Highpoint (replacing 468 & part of 408) and boost 465 operating hours and Sunday service to compensate Buckley St. Boost weekend service to every 20 min and weekday to 10 min all the way to Sunshine if Highpoint is done.

903: Sunshine - Altona. Replaced with rerouted 412 (Altona - Sunshine as per blog post). Upgrade 411 to SmartBus with 10 - 20 min weekday service, 20 min weekend with long hours between Footscray and at least Altona (Laverton desirable).

Cameron Coy said...

I do think it's a good idea to split up the orbitals. This is what I would do:

901: I'd split it into multiple sections. Keep Ringwood-Frankston as is (increased frequency). Replace the Whitehorse rd section with local routes (such as an upgraded and extended 370 to Box Hill and 907 to Ringwood). Extend the 703 to The Pines. Delete The Pines-Greensborough section. Keep the Greensborough-Broadmeadows section but re route via RMIT and the Ring Road and possibly also route via Greenvale. Extend the 577 to Greensborough to service Plenty. Terminate at Broadmeadows.

902: I'd Split it into 2 section. Chelsea-Doncaster (increased frequency) and Box hill- Melbourne Airport. I'd possible run direct to Greensborough and have a local route run Greensborough-Eltham-The Pines.

903: I'd split it into multiple sections. Keep Mordialloc-Northland (increased frequency). I'd replace Heidelberg/Northland-Coburg with a modified 513(split at Heidelberg and run Heidelberg-Coburg at 10 min off-peak/ 6 min peak) and 527(swap at Coburg end with 526 and extend to Heidelberg via Southern Rd and Waoria/Upper Heidelberg Rd at 10 min freq). Delete/replace Sunshine-Altona. Run Coburg-Sunshine Via Highpoint (replaces 408/468) (increased frequency). I'd divert the 465 to Sunshine whist extending the 406 to Keilor Park (and onto Melbourne Airport).

Heihachi_73 said...

Any chance on running the southern half of the 901 to Box Hill instead of Blackburn if it ever got split? Blackburn has absolutely nothing going for it and no quick way between there and Box Hill if you've missed the train (ignoring peak). Even when the 286 to Box Hill was a thing, it was timetabled that the 286 would leave just before the 901 so there was never a connection unless the former was late (or the latter was early, which almost never happened).

If the 901 absolutely must terminate at Blackburn, maybe someone can convince Ventura to extend the 703 to Box Hill instead. With no level crossing to contend with it should be a piece of cake to go straight along Whitehorse Rd, if not via Laburnum station instead (the bridge has a 4.2m clearance), which would avoid the Whitehorse Rd traffic lights that the 901 always gets stuck at.

If the 901 terminated at Ringwood I would have absolutely no use for it. As long as Metro's services are to remain complete rubbish east of Blackburn (I really want to say Burnley) I will keep getting the 901 when possible (obviously it's a different story on the weekend when the trains are 10 minutes and the bus is 30).

That said, extending the 907 to Ringwood might have some merit as well, although it would leave the rest of Whitehorse Rd (particularly Nunawading) without anything useful in the east-west direction.

If the Whitehorse Rd/Ringwood section replacement meant reinstating the old 270 via X, Y and Z, taking 45 minutes to get to Mitcham and finishing at 5PM on weekdays (the last one at my old bus stop used to be 5:45PM towards Ringwood), count me out, I had enough of that 17 years ago when I lived at the top of Ringwood Street! The uselessness of the 271 and 370 is bad enough.