Friday, November 15, 2019

Building Melbourne's Useful Network Part 28: The 904 Murray Rd Megabus we can have for almost nothing

Melbourne's inner to middle northern suburbs is fertile ground for buses. Car ownership is lower than average, they're becoming denser, the area is quite walkable and there's an excellent road grid. However there is significant competition for parking and the number of cars is increasing (knock down a house with 1.5 cars, replace with 4 townhouses with 6 cars). And not as many people can lead a low-car lifestyle as they would like.

Part of the reason for this is that the north's public transport network is incomplete, despite closely-spaced lines. Trams are frequent 7 days, trains are semi-frequent 7 days while bus service levels vary widely by day. Because only buses go east-west and these are often infrequent, transit works for trips to the city but not to many other places including local trips. 

Trips involving a connection require careful planning to avoid waits that may be longer than the entire trip might take if driving. That applies even for trips involving our top tier/premium orbital "SmartBus" services, such as the 903 between Coburg, Preston, Northland and Heidelberg. It is this corridor I'll discuss here.

The problem: A tangled network (it's the 903's fault)

Things were fairly simple until the 903 SmartBus came along about 10 years ago. 

Route 527 was (mostly) Murray Rd. 513 was Bell St. Main problems were limited operating hours, infrequent weekend service and the lack of a Northland to Heidelberg connection.  

The 903 orbital SmartBus helped a bit. However it does not consistently connect with trains and is infrequent on weekends. Being run over existing unchanged routes made the network more complex with two routes now between Coburg and Northland.  Was much unique coverage added? No. Almost all of the 903 overlaps other routes as per the map below.

As the 903 pushed west it overlapped more than just the 513 and 527. Firstly the 465 on Buckley St Essendon (see Useful Network 6). Secondly the 428 and  new 429 in Sunshine South. And thirdly Millers Rd Altona North with 232 and 411  (see Useful Network Part 5). Some of the 903's catchment is good but the overlaps hobble its patronage. Apart from McIntyre Rd Sunshine North, the main places where 903 has no overlap are either freeway sections (with no stops) or industrial areas (with low patronage). 

It's different in the east where the 903 is popular because its catchment was cleared of duplicative  routes beforehand (291 and 700). That's the correct approach, especially if you want the simplest network, best frequency and highest patronage for the least money.

What about timetables for the 513, 527 and 903 serving the Coburg - Preston - Northland - Heidelberg corridor? They're a mess. Historical patterns have meant that the most and least frequent services are different on different days of the week. For example:

* On weekdays the 903 is the most frequent at every 15 minutes. The others are every 20 minutes.
* On Saturdays the 527 is best, with a 20 minute service. The 'premium' 903 is only half-hourly. As a trade-off the 903 offers longer operating hours.
* On Sundays the 903 wins though its 30 minute frequency does not mesh with trains. Despite its more frequent Saturday service the 527 is the least frequent on Sunday, with an irregular 40 to 60 minute headway. Even 513 is better, with a regular 40 minute service.

There is no one route that provides consistently high frequency and long operating hours seven days per week. And where routes overlap incompatible frequencies like 15 and 20 minutes provide many buses per hour but an uneven service with long waits on a corridor.

A tidier less duplicative network could fix these problems with buses every 10 minutes on key corridors like Coburg - Preston - Northland - Heidelberg. And it can be done for next to nothing. Keep reading to learn more.

Untangling the network (3 to 2 routes with better frequency)

An untangled network in the north would have two rather than three routes east from Coburg. One would be predominantly on Bell St with the other on Murray Rd. Despite an overlap both would run to Heidelberg since it's a major employment hub with TAFE and hospitals on Bell St. 

As the Bell St route the 513 can remain as is, at least between Coburg and Rosanna. A separate reform (Useful Network 23) could tackle 513's confusing eastern end. Another could look at the Glenroy end (some eight year old thoughts are set down here).    

Murray Rd is harder. Route 527 is the route that most overlaps the 903 along it. Key decisions include:

(i) whether to or whether not to split the 903.
(ii) What to do with the 527 or the resources from it if it is split (eg at Coburg).

The aim should be a simple and frequent route along Murray Rd that connects well with trains and offers a shorter maximum wait time seven days a week than now. A 10 minute daytime frequency between Coburg, Preston, Northland and Heidelberg satisfies this condition. Ideally this would apply seven days per week. But if it doesn't weekend daytime service along this section should never be worse than 20 minutes.

When you look at this the choice becomes clear. The 903 does need to be split (at Heidelberg) with a new route operating to the west. And resources will be needed from a shortened 527 to fund the desired ten minute frequency from Heidelberg to Coburg.

Introducing the 904 Murray Rd Megabus

What would a split Route 903 orbital SmartBus at Heidelberg look like?

The 903 between Mordialloc and Heidelberg would remain at the current 15/30 minute service levels.

The western portion could become the 904 from Heidelberg to Altona.

904's Heidelberg - Northland - Preston - Coburg portion would operate every 10 minutes, preferably 7 days per week. This is better than a SmartBus so I've called it a Megabus.

Every second 904 trip would extend beyond Coburg to Sunshine/Altona. This 10/20 minute pattern is a better match for 20 minute trains and other key bus routes. It could also enable a higher weekend service (903 is currently only every 30 minutes).

It would be self-funding with resources freed by shortening Route 527 at Coburg to reduce overlap.

The 904 Megabus map below shows how it fits in with major destinations and the 513. 

 No one along the Coburg to Northland part of the 527 would be disadvantaged. This is because the new 904 runs along the 527 alignment via more of Murray Rd. They lose the direct bus to Gowrie (hardly a bustling destination!) but gain a better bus to Heidelberg and its hospitals.

Running 904 down Elizabeth St retains the connection between Route 1 tram and Northland that the 527 currently provides with improved frequency. However if this connection is considered unimportant an option exists to route the 904 through the Pentridge development to lessen overlap with the 513 on Bell St.

Local demographics and catchment

Those familiar with northern suburb politics know about the 'Tofu Curtain' or the 'hipster proof fence'. To the south are younger high-income godless bike-riding degreed childless gourmet-eating Green voting professionals. To the north are older lower income, less degreed, Labor voting blue collar workers who are often southern European or Middle-eastern migrants or descendants of same. A fair number of the latter may be retired, unemployed or otherwise not in the labour force. Some draw the line roughly at Bell St. 

The map below shows how the 904 fits relative to soico-economic class. It serves a lot of low to lower middle areas. This is generally good for bus patronage. People may not own cars. People may also have a lower propensity to ride taxis/rideshare as they can't afford it. And if they are not in the labour force they are more likely to travel during off-peak times for shopping, medical and social purposes. 

As you can see from the map below, Route 904 is central to a lot of this catchment. The ten minute all-day frequency would provide a simple and attractive service that many would use. And because it goes to so many useful places and intersects with numerous train and tram lines it would improve connectivity even for some outside its catchment. 

Objections to splitting the orbitals (long!)

Some may object to splitting orbital SmartBus routes like the 903.

There's history and politics behind this.

Five years ago Transdev Melbourne tried it as part of their ill-fated 2015 "Greenfields" network . That took buses from the west and north to give to the east and south because they thought that the latter had higher patronage potential (and their franchise contract rewarded patronage growth).  

Transdev presented its plans to affected local governments. A few got an inkling of them by reading the minutes of a Glen Eira Council meeting (no longer online but a summary of an earlier network concept is here). Eight months later there was an online survey and public information sessions. However there was the sense that the new network was a done deal and the sessions were just window-dressing.

People in Melbourne's west were particularly upset. While it is true that some cuts were to poorly used routes not all were. And there were instances of quiet routes in the east that went unreformed. The whole thing looked like a grab from the west to provide buses for the east. A better approach would have been to keep buses in the west for nearby routes that needed a boost (eg as per Useful Network Part 5).

Another problem was that the changes were to Transdev's network only. Some changes might have made sense only if there were compensatory changes to other companies' routes in the area. Confining the changes to Transdev routes meant that some areas got a raw deal while others were left overserviced or with overlaps.

The 2014 state election was in the middle of all this. Labor gained office with a slim majority. The incoming minister (Jacinta Allan) vetoed the Transdev network proposal so it didn't go ahead. (One can only speculate how this affected business since they would have been banking on approval. It can't have been good since corner-cutting caused a major fleet maintenance crisis in 2017).

In April 2015, when the new network was due to start, the minister promised a 'more balanced' Transdev proposal with better public consultation. Clear answers were not forthcoming when asked about it in September. There appeared no progress when a similar vague answer was given in parliament in November. Like Transdev, Metro Trains suffered a similar knock-back, with its 2015 timetable proposal for higher frequencies not proceeding.

Ms Allan probably made the right decision to veto, at least with Transdev. Dragging the chain on alternative reform seems less defensible, especially since some was possible in the south and east without affecting services in the north and west. In fact changes starting this weekend go further, sensibly transferring service from overserved eastern routes to the underserved western routes.  

With hindsight one could mark the two big refusals (Metro and Transdev) as the end of ministerial support for the network reform drive that had been gathering steam since about 2011 within transport agencies. While there was much more to do, this short but productive burst delivered successful bus networks in Geelong, Werribee, Tarneit, Point Cook, Fishermans Bend, Epping North and Sunshine along with ten minute service on some train lines.

Disallowing even low or no cost network upgrades has coloured the government's record on transport. The Andrews government has proved a lion on infrastructure but a mouse on networks and services. While Sydney raced ahead with 15 minute all-day 7-day train frequencies in 2017 and promised thousands of new bus services, most of suburban Melbourne remains stuck with 20 to 60 minute train services and non-connecting, meandering and duplicative bus routes. As Melbourne grows public transport networks and services could loom as a problem for the Andrews government, just as it tripped John Brumby in 2010.

Time for another look

Why should we revive the idea of splitting the 903 SmartBus orbital despite the decision in 2015? Here's six reasons:

1. Reform in areas where routes overlap could deliver frequent and popular services connecting  trains, people and jobs. Route 903, in particular, overlaps many routes for most of the way between Heidelberg and Altona. Frequencies on individual routes are low but merging routes could provide higher and even turn-up-and-go service, like the Murray Rd Megabus, for little or no cost. A commitment to evaluate Melbourne's SmartBus network was a Labor promise in 2014.

2. The government is in a better political position now than before due to its strong results in the 2018 election. The tight margin that might have made the previous minister wary about network reform no longer applies. On the other hand its financial position is weaker due to a cooling property market (affecting stamp duty revenue) and cost blowouts on large projects. Bus network and service upgrades present opportunities by being lower risk, lower cost with sooner benefits.

3. A credible public engagement process (which was not done in 2015) can minimise the political risks of restructured routes. Just after an election (ie now!) is the best time to implement potentially controversial changes as there is time to modify before the next election if required.

4. Transdev's 2015 network was based on reforming only its own routes - not those of any other bus operator. This limited what could be done, especially on corridors where routes overlapped (like along the 903's western half). Integrated planning by area and corridor rather than bus operator, can deliver the best possible network, connectivity and service levels. The 904 Megabus is again an example.

5. The 2015 proposal was basically a plan to rob the west to pay the east. There is no reason why routes like Coburg to Heidelberg can't perform as well as anything in the east if service duplication is removed. Unlike in the 2015 plan, resources saved due to route changes could be reinvested locally rather than being shipped off elsewhere.

6. We could be cleverer with split locations. The section between Coburg and Heidelberg is fine grained and continuously developed. There are many intersecting train and tram lines. And trip generators, like Coburg, Preston, Northland, hospitals and Heidelberg are closely spaced. The 2015 plan to split the orbital at Northland would have made some local through trips harder. A revised approach would respect communities of interest and have split locations where there are fewer through travellers.

Paying for it (Spoiler: it's cheap!)

A good thing about the 904 Murray St Megabus is that it's basically self-funding. No more buses than now should be needed to deliver the desired ten minute frequency. And it should be possible to arrive at a version of the 904 that keeps bus kilometres and driver hours similar also.

Route 527 requires something like 7 buses to run the full route from Gowrie to Northland on its existing frequencies. Coburg, where we'd be cutting it is almost exactly half way. That frees up 3.5 (conservatively 3) buses for use on the new 904. The other 3.5 (or 4) buses allow the Gowrie - Coburg portion to retain its existing service level.

In the morning peak the 903 between Coburg and Heidelberg takes about 40-45 minutes. Peak (and interpeak) frequency is 15 minutes (or 4 buses per hour). This level of service requires 3 buses each way or a total of 6 buses, for this section.

Want a 10 minute frequency? Who doesn't? That's a lift from 4 to 6 buses per hour, or a growth of 50%. We go from 6 to 9 buses required. That's funded with the 3 buses we found before by shortening the 527.

If we simply inserted those extra two buses per hour (short Coburg - Heidelberg trips) into the existing 15 minute Route 903 schedule you'd get something like below left. The trips per hour is good but the timetable is lumpy. There are still 15 minute waits so it's not yet a turn-up-and-go service. And it doesn't harmonise with trains every 20 minutes.

On the right is the split route 904 option recommended before. This delivers a simple even 10 minute frequency between Coburg and Heidelberg. It harmonises with trains with a regular frequent service.

The main difference is that west of Coburg frequency drops from the current 15 minutes to 20 minutes. While technically a weekday frequency cut it could deliver benefits in conjunction with numerous overlapping routes.

How so? Consider that the existing 903/proposed 904 west of Coburg is almost entirely overlapped with other routes. Especially in residential areas. For example 513 and 561 on Bell St, 465 on Buckley St, and 232 on Millers Rd.

All of these routes operate every 20 minutes off-peak. Hence bus services on these corridors would change from 7 buses per hour (with the 903 every 15 minutes providing uneven spacing with the other routes) to 6 buses per hour (with the 904 every 20 minutes). The frequency harmonisation will mean more even intervals, and in some lucky cases there might be close to a 10 minute combined service.

Also, going to a 20 minute weekday frequency west of Coburg would make the 904 cheaper to run than the existing 903. You may be able to use savings to upgrade local bus networks. Or if peak usage is high you might only go down to a 20 minute service off-peak and keep peaks at every 15 minutes or close. However you would want to stagger trips to ensure that maximum waits are kept to less than 10 minutes between Coburg and Heidelberg in conjunction with the short trips.

What about weekends? 

There are several choices.

The most legible is to run the 904 west of Heidelberg every 20 minutes all the way to Altona. However this is more expensive than the current weekend service which is only every 30 minutes. It is also inefficient with likely poor patronage due to the duplication with other routes and the low usage industrial area. It might also mean that you can't afford to run the highly desirable 10 minute weekend frequency from Heidelberg to Coburg.

Another approach could be to still have the 20 minute weekend frequency but only to Sunshine. This would tie in well with a realigned 903 that no longer duplicates other routes and serves useful weekend destinations such as Highpoint. That could be the ultimate approach after the network reforms in Useful Network Part 5 and Useful Network 6.  Short trips could augment these to provide the Coburg to Heidelberg 10 minute service.

Cheapest of all could be to keep Heidelberg to Altona at the current 30 minute frequency on weekends. It could even be timed to meet 903 trips arriving at Heidelberg. Short trips (using resources from the 527) could be slotted in between to provide a higher frequency between Coburg and Heidelberg.  That would require two buses per hour for the 15 minute overall frequency (from 527 resources) and four per hour for the ideal 10 minute service (with resources to come from some of the weekday changes).

There are several approaches which can be adjusted for taste. The main point is that a Murray Rd Megabus with a service every ten minutes is very affordable for five and probably six or seven days per week.

Optional extras

What's described above are the basics. The 904 Murray Rd Megabus could be done stand-alone with no changes required outside the northern suburbs. This is to make implementation as fast and easy as possible.

However there are some other network gains that are possible. They could be done before, during or after the 904 is introduced. Some of these are:

* Upgrading service on Route 513. The new 904 overlaps a little less of Route 513 along Bell St than the existing 903. 513 is a popular route in these parts. A service upgrade might see weekends improved from every 40 to every 20 minute. Operating hours could also be extended with some after 9pm and early weekend morning service added.

* Continuing Route 904 (ie existing 903) via Highpoint Shopping Centre along the Route 468 and 408 alignments. This would remove duplication with the Route 465 along Buckley St and provide improved access to Highpoint from two directions. More detail in Useful Network Part 6.

* Terminating Route 904 (ie existing 903) at Sunshine with a new dedicated industrial area route connecting to Altona. This would enable a new network for Altona North including an Altona - Millers Rd - Footscray Route 411 SmartBus. More detail in Useful Network Part 5.

* Potential realignment of Route 904 (ie existing 903) between Coburg and Essendon via Strathmore. This could provide a better connection between train lines and improved directness to Essendon (further aiding access to Highpoint). An alternative route or route extension would be needed to maintain service to Bulla Rd and DFO. Eventual reform might see the ten minute service proposed for the 904 extended from Coburg to Highpoint or even Sunshine.

* Potential extension of Route 903 from its new Heidelberg terminus to La Trobe University. This would replace part or all of Route 548 and/or 551 in a local network restructure. Such a direct connection has been widely requested. Its merit would need to be compared against other ideas for connecting Heidelberg with La Trobe, such as described for Route 550 in Useful Network Part 15.


Described is a new way to get across Melbourne's middle-northern suburbs. It serves all major transit, jobs and shopping destinations between Coburg and Heidelberg. The frequent service assures good connections with all intersecting modes. And its cost is low. 

What do you think? Do you have other ideas for the area? If so please leave your comments below.  

PS: An index to other useful networks is here.

PPS: Back in 2011 I favourably reviewed Human Transit by Jarrett Walker. It talks a lot about public transport network design topics like we cover here. You can buy it via the link below. The small commission I receive from purchases helps support Melbourne on Transit

Tuesday, November 12, 2019

Timetable Tuesday #48: Route 580 - The seven day bus just a few use

A month ago we looked over the Greensborough local bus network and suggested some improvements. A distinguishing thing was the large number of quiet bus routes in the area. 

Unlike new growth areas which have lots of people with few buses, some established outer non-growth areas have few people with many buses. Examples include areas around Belgrave, Warrandyte and Eltham. Such areas are identified by having having many routes with low passenger boardings per bus hour, and thus poor productivity. 

One of those quieter routes in the Eltham area is the 580. On school days it has 14 passenger boardings per service hour. This is below the 20 boardings per hour that Infrastructure Victoria regards as being a cost-effective service. One might tolerate this low number given the area's lower than average population density. 

However on non-school days the number drops to 9 boardings per hour. Hence a significant proportion of Route 580's passengers are school students. Weekends are lower again with 5 boardings per service hour on Saturday and just 4 on Sunday. In fact Route 580's sparse weekend usage put it in the bottom 12 performers on both days. 

With such low patronage for the services run it's worth a closer look. 

What factors make the 580 so unloved?  

Here's the route. 

Route 580 basically does two things. 

It's a connection from Eltham to Diamond Creek. These are stations on the same train line but the bus covers areas the train doesn't. Eltham is a larger centre and has more trains than Diamond Creek. Secondly it provides local coverage around Diamond Creek with some unidirectional loops which, unfortunately, are not understandable at first sight. 

You can see where it fits in with other routes below. 580's main unique coverage is along Wattletree Rd. Other parts are overlapped by other routes such as the 343 (itself a quiet route) and 381 (even quieter). 

Its current length is a far cry from the 580 of 1992 where it was just a tiny run going from Diamond Creek to about about a kilometre east. In fact the while area had far fewer routes then than now. Those that ran might have been indirect (eg the 517) but overlaps were rare. Subsequent routes got layered over existing routes with boardings per kilometre on each one being close to Melbourne's lowest. 

The experience showed that there are certain places where no matter how many bus routes you layer on, people won't use them. Eltham/Diamond Creek appears to be one of them. In contrast other places can have a complex and infrequent network (eg around Monash Clayton, Springvale and Dandenong) but people will use buses at above average levels. 


With one odd exception, Route 580 meets minimum service standards for local buses in Melbourne.  That is it runs every hour or better until 9pm Monday to Sunday.

However other aspects of its timetable are unusual.

Weekday and Saturday operating hours, frequencies and timetables are almost identical. Both feature half-hourly service from first to last trip.

Like on weekdays the 580's Saturday service starts before 6am. This is more generous than most other bus routes which start around 7 or 8 am on Saturdays. Route 580's Saturday service is so good that despite being one of our quietest routes it offers the same start time and frequency as most of our busy SmartBus routes up until its 9pm finish.

Route 580 offers poor connections to trains. This is because its 30 minute Monday to Saturday frequency does not evenly mesh with either the 20 minute frequency at Eltham or the 40 minute service at Diamond Creek.

Timetables below (click for clearer view)

What about Sundays? Here the 580 runs hourly. Unlike the Saturday start time, which is grossly earlier than minimum standards, the Sunday start time of 8:30 - 9am is the same as minimum standards. 

Another 580 quirk happens on public holidays. One part of minimum standards is that a Saturday timetable runs on all public holidays except for Christmas and Good Friday. The latter two days get a Sunday timetable. PTV website timetables that comply with this pattern have notes to this effect.

However PTV (who sometimes obfuscates instead of informs) deems bus routes with non-standard holiday patterns too hard to explain on its website. Hence it leaves readers with the cryptic note below. 

It is possible (but not certain) that the 580 does not follow the standard. Instead it may be the last remaining Melbourne bus route that runs a Sunday service on all public holidays.

Route 582, run by the same bus company, also from Eltham, is the opposite to the 580. It conforms to the standard public holiday pattern, even though, with its ~7:30pm finishes it doesn't conform to the operating hours aspects of minimum standards.  


I've already said what should be done with the underperforming 580; I reckon it should be scrapped as part of network reforms that could make local buses more efficient and better used. 

But what do you think? Is there still a place for it? Would simply harmonising it with train frequencies help? Or are there other routes (eg 343 and/or 381) that should be removed instead of 580? 

Please leave your comments below if you have any thoughts. 

PS: Back in 2011 I favourably reviewed Human Transit by Jarrett Walker. It talks a lot about public transport network design topics like we cover here. You can buy it via the link below. The small commission I receive from purchases helps support Melbourne on Transit

Monday, November 11, 2019

Fixing Melbourne buses: 3CR Community Radio Interview

Last Wednesday (6 November) I spoke to John McPherson and Kevin Healey from Radio 3CR's City Limits program on improving our bus network. The interview goes for 25 minutes. There's a fair amount of detail on local bus network history, service planning and the politics of network reform. 

Have a listen here (from about 31 minutes in).  

Thanks to John and Kevin for the interview. And don't forget to check out other 'City Limits' episodes on the link above as I'm sure you'll enjoy them.

PS: Back in 2011 I favourably reviewed Human Transit by Jarrett Walker. It talks a lot about public transport network design topics like we cover here. You can buy it via the link below. The small commission I receive from purchases helps support Melbourne on Transit

Sunday, November 10, 2019

Melbourne's 10 most productive weekend bus routes

Big universities are major weekday patronage generators for many of our most productive bus weekday routes.  Usage is so high that short university shuttles have been added to relieve regular routes.  These have been very successful where service is every 10 minutes or better.

Weekends are different. Big shopping centres are where the people are but the buses aren't. They're more likely to be stashed in depots, out of revenue service. This is because frequencies on many routes (including SmartBuses) drop by 50% on weekends despite continued (and sometimes heightened) demand for travel. Unlike with the university shuttles there hasn't been much of a service response to this except for the recent and temporary Route 700 Chadstone Shuttle

Another, largely historical, pattern are routes with a moderately frequent Monday to Saturday service but a sparse Sunday timetable. These are prevalent along main roads in Melbourne's inner-west and north. Typical examples run every 15 to 20 minutes Monday to Saturday with a drop to every 40 to 60 minutes on Sundays. .

Examples include 406, 408, 410, 465, 472 and 527. Saturday service may be three times Sunday's. This is because these routes typically had no Sunday service until the 2006-2010 minimum standards upgrade program. When it was added it was typically to the minimum standard of a neighbourhood bus rather than what you'd expect on a popular major route.

This variation in service levels (far bigger than variation in demand whose true value is only known if Saturday and Sunday service levels were equal) is why some routes have higher boardings per bus hour on Sundays than Saturdays, despite lower overall trip volumes on Sunday. Another factor that causes differences in rankings between the days is that not all productive Saturday routes run on Sundays. For these reasons it is better, notwithstanding the title, to look at Saturday and Sunday numbers individually.

The most productive Saturday routes

What are the ten most productive Saturday bus routes in Melbourne? Regular readers will not be surprised that 900 from Caulfield to Rowville via Chadstone tops the list with over 80 boardings per Saturday bus service hour. 733 (Box Hill - Oakleigh), 279 (Box Hill - Doncaster) and 907 (City - Mitcham) take the next three places with at least 50 boardings per hour. Rounding out the top 10 are 630 (Elwood - Monash University), 508 (Moonee Ponds - Alphington), 767 (Southland - Box Hill), 800 (Dandenong - Chadstone), 903 (Altona - Mordialloc) and 150 (Tarneit - Williams Landing) with at least 40 boardings per hour.

These routes typically run every 30 to 40 minutes during the day on Saturday. 907 is better at every 20 minute frequency while 800 is substantially inferior with buses only every 60 minutes (morning) and 120 minutes (afternoon).

Seven of the ten (shown in red) service major shopping centres. Most commonly Chadstone, Box Hill and Doncaster. The three that don't (150, 508 and 630) serve densely populated areas with favourable catchments (eg many students or people without cars). 

The most productive Sunday routes

Sunday's list is different. Seven of the top ten Saturday routes feature in it. 900, 733 and 279 again occupy the top three positions. 900 (88 boardings per hour) and 733 (62 boardings per hour) are again a little ahead of the pack. The rest (279, 630, 767, 907, 631, 623, 903 and 302) all attract between 40 and 51 boardings per bus hour.

Some top scorers only got their ranking because they serve a favourable catchment with infrequent buses. Examples are 733, 279, 631, 623 and 302 which run hourly only. Route 900 is better, operating every half hour. However its very high boardings per hour figure indicates that it too is grossly underserviced.

Similar to Saturday eight of the ten top performers serve major shopping centres. The remaining two (630 and 631) serve catchments with high bus using populations (eg Monash University students). 

Routes more productive on weekends than weekdays

It's quite rare for bus routes to carry more passengers on weekends than weekdays. More common are routes that attract more boardings per hour on weekends than weekdays. This is due to historical tendencies to run buses more frequently on weekdays than weekends (especially Sunday).

Trams and (to a lesser extent) metropolitan trains have more of a tradition of all-week frequent service. Unfortunately the big regional shopping centres, which are most active on weekends, usually only have infrequent buses from most directions. This mismatch between service and demand explains why weekend buses to such centres can attract higher boardings per hour than weekday services.

Saturday vs weekday productivity

The most extreme example is Route 900 from Caulfield to Rowville via Chadstone, with 84 boardings per hour on Saturday versus 51 on weekdays. That is a Saturday:weekday ratio of more than 1.6:1.

Other examples where Saturday productivity exceeds weekday productivity AND Saturday productivity is at least 20 boardings per bus hour are (in descending order): 279, 284, 555, 790, 903, 789, 843, 215, 556, 862, 293, 885, 626, 800, 402, 625, 527, 767, 504, 559, 513, 251, 685, 566, 506, 907, 439, 683, 411, 552, 623 and 804.

You can again see the shopping centre influence with routes shown in red. Routes shown in blue go to major weekend attractions (eg zoos) or serve a main peri-urban corridor. A surprising proportion of these good performing Saturday routes lack Sunday service. These include routes 284, 885, 800, 559, 506 and 804. It is not surprising that these feature highly in this list of six day bus routes that most justify having Sunday service added. In addition some others, while they have 7 day service, have limited Saturday operating hours, eg 279, 843, 215, 293 and 251.

Sunday vs weekday productivity

What about Sunday versus weekday productivity? The leading example is Route 843 in Endeavour Hills (whose network we discussed here). This has an extraordinary 34 boardings per hour on Sunday versus 19 on weekdays or a 1.8:1 ratio. Saturday is in between at 24 boardings per bus hour. The reason for 843's number is its Sunday underservicing with buses every two hours. Yes, high productivity can wreck a bus route (or at least make it underperform in the patronage stakes).

Other routes where Sunday productivity exceeds weekday productivity AND Sunday productivity exceeds 20 boarding per bus hour are (in descending order): 900, 849, 862, 220, 279, 903, 767, 845, 631, 251, 626, 216, 623, 223, 670, 219, 293, 624, 555 and 732.

What places do these high-productivity weekend routes serve? Routes to regional shopping centres, such as Chadstone, Doncaster Shoppingtown, Northland, Box Hill stand out. As do some to other major destinations such as Melbourne CBD, Footscray and Dandenong. Popular weekend attractions or day trip routes also feature (eg 683, 685 and 439) especially on the Saturday list.

Routes like those discussed are where you'd concentrate resources on if you wanted to improve weekend bus services. Even if you had to reduce weekday frequencies of poorly performing routes or, better still, revamp networks to reduce service overlap and duplication. 

Does high Sunday productivity correlate with high Saturday productivity?

The short answer is yes. Routes with 7 day service were graphed by productivity (below). A route with a high boardings per hour ratio on Saturday is also likely to be popular on Sunday.

The main places where Sunday usage is lower are traditional working to lower middle class suburbs in many areas across the north and outer south-east (eg around Hallam with routes like 892 and 893). Sunday trading appears less prevalent there than affluent inner and bayside suburbs. However in lower income outer suburbs (eg Deer Park and Craigieburn) Sunday productivity remains above average (eg Routes 424 and 529). They're only below the 1:1 line because Saturday usage is so good.

The very underserviced (733 and 900) have high numbers on both days. Those only grossly underserviced on a Sunday (eg the 843, 845 and 849 cluster around Endeavour Hills) are well above the line. I discussed the Endeavour Hills network in Useful Network Part 27. Other examples of Sunday underservicing include the 302/304 pair (covered here) and routes around Highpoint (discussed here and here).

What about routes with no Sunday service? (a hypothetical)

The Sunday vs Saturday graph above can only have routes with Sunday service. We saw the strong link between Saturday and Sunday productivity. The graph below includes the ten most productive Saturday routes without Sunday service (800, 506, 536, 885, 281, 804, 814, 284, 559, 612). Not surprisingly most feature in Thirteen Melbourne bus routes that most deserve Sunday service. I took the Saturday boardings per hour and subtracted 20% to reach a conservative Sunday estimate.

All these routes, if they got Sunday service, would be more productive than most other routes and boost the overall network average. And they would connect people to jobs and services they need seven days a week. So if you wanted to bring Sunday bus service to areas where you're sure people would appreciate it and heavily use it then you'd start with these routes. 

Summary and implementation

This has been our look at our most productive weekend bus routes. A simple doubling of frequencies on our busiest weekend routes using the existing bus fleet would make a huge difference to the network's usefulness.

And given the close link between Saturday and Sunday bus route productivity, adding Sunday service to our most productive Saturday routes currently without it would also have a large effect.

Where routes have sections that are poor performers and others that are popular you might add extra short trips to provide doubled frequency only on the section that needs it. Examples include sections of orbital SmartBuses like the 903 and the Box Hill - Clayton section of Route 733.

Network reform can be even more economical, especially where routes have overlapping catchments. For example simplifying 802/802/862 into two routes might save resources compared to retaining the existing three routes. Weekend service might involve boosting one extra route (instead of two) so that two out of two routes have the upgrade service (instead of three out of three, possibly at a lower frequency). 

The same could apply where there is significant overlap (often involving SmartBus orbitals) and there are alternative alignments that could serve popular destinations better (eg routing 903 via Highpoint). See the Friday Useful Network posts for more discussion on streamlining local networks. 

Over to you. What do you think about weekend bus services and where can they be improved? If you have thoughts please leave them in the comments below. 

PS: Back in 2011 I favourably reviewed Human Transit by Jarrett Walker. It talks a lot about public transport network design topics like we cover here. You can buy it via the link below. The small commission I receive from purchases helps support Melbourne on Transit

Friday, November 08, 2019

Building Melbourne's Useful Network: Part 27 - Endeavour Hills and Doveton

Last week we covered Hampton Park to Berwick in the City of Casey. It proposed some frequent, Useful Network routes along main roads serving key destinations like Fountain Gate Shopping Centre, Dandenong Hospital and various TAFE and university campuses. I didn't say much about local routes. But these too should be subject to review, with attention paid to simplicity, coverage, strong termini and timetables that connect with trains.

The northern part of last week’s expanded Useful Network is below. The main change for Endeavour Hills is the extended 841 along Heatherton Rd operating every 20 minutes. This replaces the thrice-daily 842 from Fountain Gate and provides a new, direct connection to Dandenong via the hospital and Chisholm TAFE. This removes the need for Endeavour Hills passengers to change buses to make a short distance (but likely popular) trip. And, as you’ll see later the extended 841 permits a simpler Endeavour Hills local network. I'll discuss that later.  

Endeavour Hills community profile

Endeavour Hills is a large and established 1970s-80s era suburb comprising mostly stand-alone houses. Leafier and with bigger homes than Doveton to the south, it is one of the higher income parts of the Dandenong area. Its street layout reflects the fashions of 1970s traffic engineers. 

Their doctrine favoured a road hierarchy comprising a coarse limited access main road grid good for car traffic throughput (but impermeable for pedestrians) and curvy local streets good for slowing cars (but poor for simple, fast and direct walking and bus routes). A medium sized drive-in shopping centre, not walkable from most areas, services the whole suburb. There are no main street shopping strips and few local milkbars. There is good access to the freeway but the suburb’s separation from other areas makes it poor for active transport. 

Endeavour Hills is an interesting contrast to neighbouring Doveton, planned in the 1950s to house automotive and allied industrial workers. This has straighter streets, is more contiguous with surrounding land uses (including industry jobs) and has several small local shopping strips. Homes might have one car but rarely two. 

A parallel to Endeavour Hills in terms of layout and era is Weston Creek in Canberra, with Woden performing a similar role to Dandenong as a major suburban centre and transit hub. Again its peripheral location and wandering internal streets have impeded efficient and direct bus services. 

Existing network and services

Partly due to the local street layout Endeavour Hills local bus routes are indirect and complex. They have not had a serious revamp for years. What changes reflect ‘graftings on’ with one recent addition (863) and a small extension of one other route (843) to the shopping centre. 

Four main routes (843, 845, 849 and 861) extend from Dandenong with weak termini. Each operates at the same frequency; 40 minutes on weekdays, 60 minutes on Saturdays and 120 minutes on Sundays.  Because they go the same way to Dandenong the routes combine to provide a more frequent service where they overlap in northern Doveton. 

A 2006 – 2010 program rolled out ‘minimum service standards’ to bus routes across Melbourne, with service at least hourly until 9pm seven days per week. Despite high social needs and good patronage the Dandenong area got dudded, with only a few routes getting upgrades. This is why the premier’s seat of Mulgrave (which is close to Dandenong) has amongst the least bus service in Melbourne, particularly on Sundays.  Endeavour Hills does get daily service but operating hours are short and the two-hour gaps between buses are double the minimum standard.

Endeavour Hills is split between two electoral districts. These are Dandenong (Gabrielle Williams MP) and Narre Warren North (Luke Donnellan MP).  

The limited Sunday timetables aren't the only quirks with Endeavour Hills buses. Whereas trains, trams and some buses have later services on Friday and Saturday nights than on other nights, in Endeavour Hills the late services are on Thursday and Friday nights.  This possibly dates from when Thursday nights was late night shopping in the suburbs. 

Late weeknight finishes like these are now rare on Melbourne buses. They used to be common in the days before minimum standards when most routes finished at 7pm but some had trips added for late night shopping. And they were even more common before the 1990/91 bus service cuts. More recently minimum standards upgrades effectively scheduled what were late shopper trips seven nights per week on the routes that got them with widespread 9pm finishes. 

“You can go anywhere you want as long as it’s Dandenong”. That’s only a slight exaggeration if we’re talking about buses in Endeavour Hills.  Yes, Endeavour Hills has two non-Dandenong routes. The 843 goes to Fountain Gate so could potentially be useful. Except its timetable has just three trips on weekdays and none on weekends. One can also go to Hallam Station and Hampton Park Shopping Centre via the 863. This is a new route. Unlike all other Endeavour Hills routes it runs to minimum standards. Even on Sundays with buses every 40 minutes. Unfortunately it doesn’t go to places local people want to go to and patronage is very poor. In contrast, even though Dandenong Hospital, Chisholm TAFE and the centre of Endeavour Hills are on Heatherton Rd there is no direct route linking them.  

The extended 841 addresses this issue by providing a new way into Endeavour Hills from Dandenong via key trip generators.  A large part of the funding for it can come from what we decide to do with the existing Endeavour Hills bus network, which I’ll discuss later.

Heatherton Rd SmartBus proposed in 2010 Bus Review

Further confirmation of the worth of a Heatherton Rd connection from Dandenong to Endeavour Hills (and Fountain Gate) came in the 2010 government-commissioned Cardinia Casey Greater Dandenong Bus Service Review. It pointed out that there were 5000 jobs and 15000 students in the education and health precinct just north of Dandenong. It proposed a realignment of Route 861 via north Dandenong, extension to Fountain Gate and an upgrade to a Principal Network (ie SmartBus) service level. The eastern part of the route is different but it's conceptually similar to the 841 Heatherton Rd/Dandenong extension mentioned above.

The review found that Route 844 in Doveton had good patronage performance but a dead end terminus. It proposed a south and east extension to Hallam and Fountain Gate. Reforms were proposed to local routes 843, 845 and 849 - see the link above for more information.

Revamped Endeavour Hills bus network

Endeavour Hills’ bus and pedestrian hostile street layout makes it difficult to economically provide fast and direct bus routes that everyone can walk to. The map below is my best attempt. It retains service at nearly all existing stops while reducing the number of routes by two. Some local trips are less direct but everyone retains one bus access to Endeavour Hills Shopping Centre and, in most cases, Dandenong. The network is also simpler than that proposed in the Bus Review. 

The centrepiece of the new Endeavour Hills network is the extended Route 841 mentioned last week. This provides an efficient alternative connection to Dandenong (via Chisholm TAFE and Dandenong Hospital) and seven day service to Fountain Gate (and beyond).

Much of Endeavour Hills is remote from Heatherton Rd and the abovementioned 841. Hence neighbourhood routes, serving areas north and south, are needed. A pair comprising Route 842 (a reused number) and an altered 843 is suggested. Both operate from Dandenong to Endeavour Hills Shopping Centre. They operate as a through-routed pair, similar to 494 and 495 in Point Cook. This means that 842 trips arriving at Endeavour Hills Shopping Centre have only a short dwell time before forming Route 843 trips back to Dandenong. And vice versa. On every trip.

This means that those in James Cook Drive wishing to go to Dandenong get a faster ride if they get the westbound 843, remain on the bus at the shopping centre as it forms a (new) 842 and then get a direct trip to Dandenong.  A similar arrangement already happens with the 845 and 849 so residents won’t be unfamiliar with it. 

Overall frequency is also improved. Departures of 842 and 843 at Dandenong would be evenly staggered so that those going home can board either bus, benefiting from a doubled frequency. The use of consecutive numbers and departure from the same bay at Dandenong Station would make communicating this simple. 

How frequently would the 842 and 843 operate? A base 40 minute frequency (Monday to Sunday) is similar to current weekday service levels and a major boost to weekend services. Staggering would give a 20 minute combined frequency from Dandenong.  Peak services on each route might rise to 20 or 30 minutes. Evening frequency might be hourly (or 30 min combined) with operating hours extended where they don’t currently meet the 9pm finish minimum standard.  

Because the 842/843 pair replaces four existing routes (843, 845, 849, 861) this arrangement should be cheaper than now. The freed resources would go towards (but not entirely fund) the 841 extension (with wider benefits as far south as Cranbourne). If this is successful additional short trips could run from Endeavour Hills to Dandenong to provide a combined 10 minute service during peak times. 

Every bus network reform has some downsides. This is no exception. A price for merging 4 routes into 2 (or more strictly 5 into 3 with the 841 extension playing a major role and replacing the current 842) is that some streets could lose buses.  The unhelpful road pattern extends walking distances longer than they should be.

A modified 863 (numbered as 845 above) is suggested to provide fill-in coverage, connecting people to trains at Hallam Station and shops at Endeavour Hills. Its northern terminus is weak so is not something I would normally do. However there is no logical place to economically extend it. Due to 863’s currently low patronage a 60 minute 7-day frequency is suggested though a 40 minute service would connect better to local routes like the 842 and 843. 

Not much change is suggested for Doveton. The main network change is extending Route 844 south to the highway. This provides connections to two east-west routes and more local shops. 844 would also gain 7 day service and longer hours, as is appropriate for a popular route in a high social needs area. I considered an eastern extension to Hallam but this would overlap other routes and cost money that I'd prefer spent on the 841 Dandenong extension and service upgrade. 

Route 828 would become part of the 830, starting at Dandenong instead of Hampton and extending east to Pakenham. Its weekend service could be every 20 or 30 minutes rather than the current 40 to 60 minutes. Either 20 or 30 minutes on weekends would suit trains at Dandenong but the more expensive 20 minute option would be better for bus/train connections at Berwick Station. 40 minutes would be the cheap option, still delivering an upgrade for all of 926 and a Sunday upgrade for Doveton.

Both the 844 and 828/830 changes are self-contained and can be done independently of the other reforms discussed here. 


Described is a local bus network for Endeavour Hills. In conjunction with the Heatherton Rd route it should provide improved access to attractions such as Fountain Gate Shopping Centre along with health and education services at Dandenong. Also local routes would provide a higher combined frequency from Dandenong than operate now, especially on weekends. 

What do you think? Please leave your comments below. 

PS: An index to other useful networks is here.

PPS: Back in 2011 I favourably reviewed Human Transit by Jarrett Walker. It talks a lot about public transport network design topics like we cover here. You can buy it via the link below. The small commission I receive from purchases helps support Melbourne on Transit