Thursday, July 04, 2024

UN 177: Transport for our housing targets

Last month the state government released draft housing construction targets for between now and 2051 to accommodate Melbourne's large projected population increase. 

A doubling by 2051

These targets were by local government area. Meeting them would mean about a doubling of dwellings in most inner, middle and even outer municipalities. 

The main outliers are Melton and Mitchell (a tripling or quadrupling) and Nillumbik, Yarra Ranges and Mornington Peninsula (about a 50% increase). But apart from those it's pretty close to being 'one size fits all' with accelerated densification for many historically low-growth but well-serviced established middle suburbs.   

Such growth has many implications for transport networks and services. 

Everything else being equal, more homes = more people = more cars in an area. 

How many more? 

Take, as examples, the cities of Darebin and Glen Eira in Melbourne's north and south-east respectively. Both have similar populations (around 150 000), number of existing homes (around 70 000) and housing targets (around 70 000 more by 2051). Darebin has about 1.5 cars per household while Glen Eira has about 1.6. This is a little lower than the national average of 1.8 cars per household in 2021.

Glen Eira is slightly smaller (39 square km) so is a little denser than Darebin (53 square km). They're both irregularly shaped. But half close your eyes and/or twist your head to see roughly 10km x 5km oblongs. 

How many more trips?

Australian household sizes average 2.5 people per dwelling. In densifying established areas it is likely that the composition of new homes will be different to those already there. That is townhouses and apartments replacing detached houses (which currently dominate Australian suburbs). 

Given Australian families' preference for detached houses it is likely that average people per unit will be lower than that for houses. So let's call it 2 people. If that holds and the housing targets are met then you'll see about 140 000 more residents in areas like Darebin and Glen Eira. 

It varies by age but people generally make about three trips per day across all transport modes. Thus a municipality with 140 000 more residents will need to cater for about 400 000 new daily trips. The majority of trips made are currently by private car (either as driver or passenger). That would also hold for the 400 000 being added unless something changes. Eg improved accessibility and/or attractiveness of public and active transport or a lower demand for travel.   

How many more cars?

Assuming similar car ownership rates per home to now, Darebin and Glen Eira would be accommodating 100 000 more cars each than now. Maybe that's a little high if the new homes (more likely to be townhouses and apartments) have more singles than existing detached homes. Or maybe not if detached homes are more likely to have children who (if not adults) won't have their own car. 

Anyway let's go with the 100 000, understanding that it remains a big number even if halved. Motorists expect parking in at least three places; on their property, in the street in front of their house and at their destination. The first is privately paid for so is not our concern (though surveys show that many residents use garages for non-car purposes and park on the street). The last two are predominantly at public expense so are of public concern. Especially because their space usage worsens active transport access and crowds out alternative uses for the land (including more housing). 

Most motorists probably don't particularly like driving and regret its expense, especially in these inflationary times. They'd also likely scoff at the usefulness of existing driving alternatives for their life. Which is reasonable given typically disconnected cycling paths and weekend buses only every 30 to 60 minutes in most suburbs.  

The graph below gives a snapshot of a few of the challenges that face established local government areas (like Darebin and Glen Eira) if the government's housing targets are implemented but existing patterns continue for transport (ie everything approximately doubling). 

Where will all the cars fit?

Cars are 5 metres long on average. 100 000 of them bumper to bumper forms a 500km long line. That is almost Melbourne to Mildura. Allow for backing-out space (if parked) or stopping distance (if moving) and the space requirement multiplies. That increases distances between higher value land uses, reducing active transport accessibility. Road tunnels and multi-storey parking may assist but are expensive per person served. And we're just talking about local cars; we haven't factored in those driving through from surrounding municipalities with similar housing uplifts.  

Regardless of individual attitudes towards car ownership and use, it's a physical fact that the movement and storage of private cars is the least space efficient transport mode there is. 

That's a problem because they're not making any more land. The more housing density the more intelligent land needs to be used. People value space, whether it be public parks or their own private yards or courtyards. 

Such space is the first to go when policy settings and standards 'crowd out' active and public transport by making its use unattractive and traffic engineers decree we need more space for roads and parking. 

However much of this is a consequence of planning and policy choices. For example minimum parking requirements (recently removed in Auckland), compulsorily bundling parking with property titles (desirability of unbundling discussed in this recent RMIT study), severance from over-wide roads, poor active transport links and limited public transport services all discourage space-efficient transport. Salary packaging (encouraged by federal tax rules), 'free parking' and traffic-inducing freeways also represent a substantial financial encouragement to drive, despite the social costs of people doing so.  

Taking the housing increase as given, it's effectively a trade-off between more space for people or more space for cars. A choice exists but only if policy settings encourage rather than discourage more space-efficient transport.  

What if there isn't? Accommodating more cars pushes destinations further apart, reducing the usefulness of active transport. The result is suburbs that have all the problems of housing density but few of its benefits. 

Existing link between housing and public transport service

People, including housing density advocates, sometimes assume that better public transport will somehow automatically follow anything done in housing. For example YIMBY's 'Missing Middle' plan glosses over the need for a frequent multidirectional transport network in areas where intensification is proposed. However the evidence is otherwise; dense but poorly served housing areas are a growing problem in Melbourne. 

Densifying areas like Moonee Ponds are one example, with off-peak train frequencies basically unchanged for decades. Neither have the tram nor SmartBus networks been significantly extended in the last 13 years. 

The pace of 7 day upgrades on local bus routes is dramatically slower now than 15 years ago. Victoria's Bus Plan, now in its 4th year, has made little headway redressing bus service inequities. At best there may be a belated response after about 10-15 years and even then weekend frequencies might not be particularly high. For example the Route 505 and 546 upgrades starting later this month.   

More broadly, recent RMIT research found that public transport service was lagging apartment builds. Our busiest public transport modes have seen a per capita service decline. This includes trams, which although generally still more frequent than trains and buses have not recovered from a 70 year decline and then stagnation in their frequency. 

This slow progress on service has limited the practicality of public transport for many trips where it has potential to be useful. 

Six directions for Melbourne public transport to accommodate housing growth

The housing construction targets need an accompanying transport infrastructure and service agenda to get the transport mode shift needed to make denser areas more rather than less liveable. Otherwise some of the criticisms of Melbourne 2030 (a land use plan unbacked by transport initiatives) may hold for these targets. Advocating density in established areas is fraught with political risk, especially if there are no gains for existing residents, such as better transport connections and broader public realm improvements such as parks, libraries and community centres.  

Double Service Frequency on Everything, a five word public transport plan, would transform public transport across Melbourne. A 2 minute video summary is below:   

Double Service Frequency on Everything would be a rough network average. In practice some routes would retain their current frequency while others might run 3 or 4 times more frequently at certain times. Priority, at least in its early phases, would go to off-peak service to minimise new rolling stock and infrastructure requirements. It would massively transform the existing network from a peak-oriented radial suitable for a minority of trips to a much more versatile all-week frequent grid suitable for many more.    

There is a great deal of flexibility in the implementation of such a frequent network. The multimodal Future Frequent Network is a coherent example, drawn from updated Principal Public Transport Network concepts. If you wanted to tie it to housing provision I've previously outlined priorities for social housing growth areas (2020) and housing priority areas (2023).    

Would this in itself be enough given the ambitious housing and thus population targets in across all of Melbourne? Probably not. Something even bigger is needed. And not just for frequency either. 

1. A fast and frequent network

A network vision that truly complements the housing agenda would likely include not just the frequent network mentioned above but also a fast and frequent network. This would comprise not just the radial rail system going into the CBD but routes in other directions such as popular rail feeders, circumferential routes and direct frequent corridors though suburban centres.

A fast and frequent network would multiply the number of suburban connection points across Melbourne. You could argue about modes, with light metro, trams or premium buses being possibilities. 

But if you want to get the most done soonest it's hard to go past Toronto-style two tier express/stopping buses running every 5-10 minutes until 1am (as I suggested as an SRL precursor). Think of it as a Future Frequent Network but with fast express buses overlaid over key circumferential routes (such as  existing SmartBus orbitals and the SRL precursor) and some new radial and L-shaped corridors to growth areas without local rail like Wollert and along Taylors Rd west of Caroline Springs. Bus wormholes should exist at pinch points to deliver a busway like travel experience. 

2. Five minute frequencies on the network's core 

Because so much of Melbourne's public transport runs only every 30 to 60 minutes for much of the week, the main emphasis of Double Service Frequency on Everything has been to get to a 10 minute service on the more frequent routes with most of the rest running every 20 minutes. 

While a large improvement on current service levels, even 10 minute service isn't great for complex trips involving two or more transfers (as more of them will need to be if public transport is to have a higher mode share). Even assuming all services run, such trips could have up to 27 minutes total waiting, which adds excessive variability for all but long distance trips. 

The cure to this is even higher frequency on the core parts of the network. A 5 minute all-day frequency on the busier lines would bring our service levels near to those which currently run on Vancouver's Skytrain or Toronto's subway. Melbourne is not a stranger to high frequencies in the past. For example our cable trams ran frequently. Today's frequent 401 and 601 university shuttle bus routes are highly productive. And 5 minute frequencies represent mainstream thinking, as exemplified by 2012's Network Development Plan - Metropolitan Rail which had a multimodal service frequency hierarchy starting at 5 minutes for its core routes. 

Examples of routes for which a 5 minute service could be suitable all day (and even at night) include Metro lines to destinations like Newport, Sunshine, Clifton Hill, Ringwood and Dandenong. This could be considered in conjunction with all-day express services where line capacity permits. The benefits of separating trains from intersecting traffic through level crossing removals will be found indispensable at frequencies like every 5 minutes. Similar high frequenies could apply for key tram lines serving densifying areas, such as Route 19, 59, 86 and 96 to name a few with maximum waits never exceeding 10 minutes late at night. 

10 minute frequency lines and routes that might have had this finish in the early evening might have this service extended to midnight or later, such as is general practice in Toronto and Vancouver.  A 10 minute frequency operating over about 18 or 19 hours of the day would be a particular gain for Melbourne, which even on its existing frequent lines like Frankston, only runs its 10 minute service for about 9 hours on weekends. Both Sydney's metro and regular train lines far lead Melbourne's with 10-15 minute maximum waits, even from outer suburban stations, at nearly all the time services run. 

7 day 24 hour public transport is something that big cities have. So should Melbourne. While much was made of Night Network operating across all modes, its hourly train and bus frequency makes it a dealbreaker for much travel. And it only runs weekends. The next stage could entail boosting all night service to all week with much higher frequencies similar to Toronto's network. Of all the modes trains are by far the dearest to run all night so you'd start it with buses. 

With an 'always there for you' philosophy, an approach to service design like the above would encourage the sort of low car living that is essential for the proposed housing densities to effectively work. 
3. More capacity (but only after frequency)

When a service is already frequent then adding capacity is another way to cater for higher ridership. We're talking about bigger trains, bigger trams and articulated buses. Or, if you've got a capacity crisis on your hands, just ripping out seats on existing rolling stock to allow more standing room. Higher capacity vehicles are good for handling high peak loadings and improving passengers carried per driver productivity metrics. 

Melbourne has generally been better at adding capacity than adding frequency, especially on its tram network, where small non-accessible trams are steadily being replaced by larger accessible trams. Ditto for the HCMTs that will be running through the Metro Tunnel. 

Adding capacity is good for catering for existing growth on crowded routes and improves comfort for existing passengers. Which may induce some further patronage growth. However, unless frequency is already very high, you should always insist on a 1 for 1 replacement or better to ensure frequencies can be maintained and if possible increased.   

4. A war on delays to maximise throughput

A common objection to public transport is that it is several times slower than private car travel. The most extreme differences, especially for trips involving a change, can be reduced by providing a frequent network. Timed transfer networks can also help but are more suited to low density fringe and regional areas like we are not discussing here. 

Once frequency is good other steps are desirable to increase speed and reduce travel time variability.  Depending on mode these may include extra tracks for passing, signalling upgrades, simpler stopping patterns, tram priority, bus lanes and priority treatment at signals.

Delay must be attacked at all sources with slicker operating practices and targeted infrastructure upgrades. Some things that cause delay are not immediately obvious. For example providing shelter along a rail platform at a popular station can help spread passengers more evenly and thus reduce boarding delays on wet or hot days. Do this at multiple stations on a busy line and you might have saved a couple of minutes. A similar story (along with accessibility gains) can be told for enabling level boarding without excessive gaps. Adding extra entrances at busy stations could probably induce as much extra patronage as some existing stations already have, all without delaying existing passengers. Short, wide, sheltered and direct walkways must also exist at major interchange points to speed intermodal interchange. 

All these can create a virtuous spiral of improvement. For example bus and tram priority speeds travel, creating a speed benefit for passengers. The reduced run time can enable higher frequencies with the existing fleet and drivers, which further improves capacity and reduces waiting. This encourages further patronage uplifts, with an improved farebox recovery ratio. 

Rail speeds are mostly controlled by on rather than off-line factors, most notably our closely spaced suburban stations. Average 35 to 40 km/h speeds can be excruciatingly slow on longer lines like Frankston, even if its frequency is good. Some lines have third tracks to enable peak expressing but consideration should be given to fourth tracks (and/or passing loops) to enable frequent all-day two tier service that includes expressing. Alternatively it may be more cost-effective to build an entirely new rail alignment (with new stations and interchange points) to cater for high demand or to enable 'knock-on' benefits for the existing train or tram network. It is hard today to see how the Werribee line would cope without the Regional Rail Link, for example. 

5. Network reform across all modes

Higher population density and a desire for mode shift requires a new emphasis on efficiency and making the network more useful for more trips. 

For buses this needs network reform, including making existing routes more frequent and direct, as has been discussed amply here. There also needs to be a large bus component for the 'fast and frequent' network, weekend frequency boosts and a completion of MOTC minimum standards 7 day service to all but the quietest local routes (upgrades to about 60 routes needed all up). Responsiveness should be much faster, with bus routes rolled out far earlier in a new estate's development and less restrictions for GAIC-funded bus networks. 

For trams this means a faster, more accessible and more connective network. You don't need to go far from the CBD for the tram network to cease being a grid and for it to be radial. The tram network has basically stagnated with the very slow pace of accessible stop roll-outs, few upgrades to timetables, little network reform or extensions past dead-end termini that date from when trams competed with rather than fed trains. Also a concern is its declining efficiency due to vehicle traffic. Something more than the 2023 'plan for a plan' tram plan is needed to make the tram network fit for purpose with many more multimodal suburban connection points

For trains we need to review network complexities including an excessive number of peak stopping patterns on some lines, unfathomable City Loop operations and poor frequencies, especially early mornings and nights. The stalled greenfields timetable roll-out needs to be restarted, and not just on the  lines directly served by the Metro Tunnel. 

The pandemic, working from home and the uplift from the Metro Tunnel next year should relieve CBD capacity issues for a while. But if Melbourne grows sufficiently it may recur as a problem. Suggested solutions could include the disruptive, controversial but cost-effective City Loop split (as supported by Infrastructure Victoria) and/or additional cross-city Metro lines (eg MM2 via Fishermans Bend with capacity gains for the west). Both of these would convert more of the network to a simpler, more through-routed system which is far better than provincial-style stub termini or confusing CBD-centric loops.  

6. Tram-like local networks for suburban centres

Suburban centres will likely be too large and busy for a single interchange point at a station to be useful for all but a few destinations. That has been the assumption to date, with centres like Box Hill, Dandenong, Frankston, Broadmeadows and Sunshine having many routes terminating there but a handful of (if any) through routes. 

This network model has outlived its usefulness and needs to be replaced with something more akin to frequent CBD trams operating frequently on defined corridors. Termini can be on the far side of the centre but are never in the middle. That just forces time-wasting transfers and bus turnbacks in space-starved core areas. As well as permitting easier movement within a centre this also has benefits including better central interchange space efficiency due to shorter dwell and loading times. A concept for Frankston is shown below.  

Each line could be two or three less frequent routes that are co-scheduled to provide a frequent combined service where they come together. Location-specific passenger information is required to make such multi-route corridors legible. This is something that PTV hasn't historically been very good with but is cheap to improve. 

Similar opportunities should be taken for Suburban Rail Loop centres like Southland, Clayton, Monash, Glen Waverley, Burwood and Box Hill to have similar through-routing. Again the aim would be to provide a tram-like service within its heightened density 1.6km radius centre including to major destinations like Box Hill Hospital. In many cases this will be most economically done by reforming and upgrading existing feeder buses rather than putting in dedicated shuttles (which are prone to having only short operating hours and adding network complexity). 


Denser housing in our suburbs needs commensurate investment in transport services and infrastructure to get people to where they wish to go. The need for a strong public and active transport component means a need to plan for far higher per capita usage than currently exists. Six directions to help enable this have been presented. Failure to keep public transport service development up with housing risks resulting in suburbs that are less prosperous, less liveable and less green. 

1 comment:

Ian Jones said...

Point 7 - Importance of defining and implementing supporting road space allocation
Perennial issue, yet continuing un-defined, remains defining parameters around allocating road space. Increased headways, enhanced transfer points, and faster PT routes, particularly in built-up areas, inherently require re-allocation of road space from car-oriented requirements, such as parking, and travel, along with signal priority and enhanced intersection designs to reduce bus/tram and car shared spaces. But great stuff Peter!