Friday, July 03, 2020

Twelve lessons from Adelaide's attempted bus network reform (Useful Network Part 51)

Last month the South Australian state government released details of a 'better, faster and more frequent' bus network for Adelaide. Within three weeks the proposal was dead, killed by public pressure and a threatened revolt within the ruling Liberal Party.

What happened? What was the proposed network like? And can we learn from the Adelaide experience? I'll answer those questions later. First up though I'll describe how bus network reform can be good, bad and controversial.

Network reform can be good

Bus network reform is one of the most cost-effective ways to make public transport more useful for more people. Without reform networks can become a time capsule, unsuitable for modern travel needs. This condemns buses to a future of declining usefulness and patronage.

For example unreviewed networks can contain routes that stop short of new suburban centres, still serve shut-down schools and retain timetables that reflect 1970s shopping and working hours. New routes layered over unrevised existing routes duplicate service, waste money and carry fewer people than is possible with simpler revised more frequent routes. Unchanged timetables can also make buses miss trains, making them less useful for all but very local trips.

Good network reform can fix all these issues. In many established areas it can be done cheaply with the existing number of buses. The main extra cost is usually the bus and driver hours required to boost historically limited evening and weekend services.

Bus purchases are needed for coverage extensions to new areas and/or peak frequency upgrades. However their utilisation can be improved if accompanied by route reform in areas with legacy networks. Previous Useful Network items present ideas along these lines for bus networks in Melbourne. Rail Back on Track's Frequency is Freedom paper is a good summary of similar network issues in Brisbane.

The graph below shows that in areas with duplicative unreformed networks adding buses and reforming networks provides more benefits than adding buses alone. The trade-off is that some network reforms can be politically controversial, especially if the new network doesn't add service hours or buses.

Network reform can be bad 

Not all bus network reform is good. Bad network reform can be worse than none at all.

It might be driven by perverse incentives contained in operator contracts. For example paying bus companies on the basis of patronage sounds good but can lead to a tendency to favour network designs that compete with trains and trams for radial CBD travel. The result is a more duplicative network that reduces frequency and/or coverage in suburban areas. Plus local non-radial trips may be made harder.

Another limitation is if it is done in isolation. This can happen where multiple bus operators serve an area and the reform is to one company's routes only. The ill-fated 2015 Transdev Melbourne greenfields network was based on the premise that their routes were quiet in the western suburbs but busy in the east.

Transdev sought to take buses from the west to give to the east. This ignored network shortcomings in the west like their routes 223 and 903 being largely overlapped by other routes or passing through quiet industrial areas. Had these been addressed, in concert with reforms to other operators' routes, the west and north could have ended up with a generally better rather than a generally worse network.

Maps of neat grid networks on main roads can be alluring. A main road bus route may look, on paper, near homes, but continuous traffic, large roundabouts and indirect streets may preclude easy pedestrian access. Passengers would reasonably oppose bus routes being pulled off local streets, even if they are indirect. There may be substantial local flows to shopping centres that a revised network might disrupt. And areas may have demographic characteristics, such as low car ownership, low workforce participation or high numbers of commuters that affect how frequently you design a service to run and at what times of day. Ignoring such fine grained localism can prove the undoing of bus network proposals that otherwise look good on the surface.

Network reform can be controversial

Existing bus networks, by definition, best suit those who currently use them. Ask existing passengers about network changes and they will often be wary. Whereas many non-users, when asked why they don't 'use buses, will cite things that network reform can fix like indirectness, limited operating hours, low frequency or poor connectivity with trains.

A revised network that contains many 'greater good' changes that benefit several times more people than it hurts can still come under fierce and visible opposition.

Existing passengers who would lose from a proposed network change can be counted on to vocally oppose it. Supporters or those unaffected will likely be less vocal. That can skew results of passenger surveys due to higher participation by opponents of change.

Supporters also won't write to politicians or contact the media. Whereas opponents will. Opposition can include sympathy from people who don't currently use buses. They might not be across the detail but don't see why services should be cut. These factors can lead to politicians being spooked by greater opposition than actually exists.

Squabbles between different levels of government can stymie reform when both run transport services. Brisbane found that out when the state government's Translink 2013 network review failed due to disagreement with the Brisbane City Council (which, uniquely in Australia, runs many buses). Bus services remain an ungovernable tussle between Brisbane City Council and the State Government with no prospects of reform in sight.

Reform is easier (and more politically acceptable) if there is an overall increase in buses and driver hours because some can be diverted to restore some of what an initial proposal removed. However the resultant network won't be quite as simple or consistently frequent as first proposed. 

Adelaide's proposed network

Adelaide's bus network needed reform. It's horrendously complex for a start. Not only are there many route numbers but many variants starting and ending with letters. You can see the current network map here, but even this understates the complexity with some services operating only occasionally.

Like Brisbane's, Adelaide's bus network is very CBD oriented. There are few cross-suburban routes. Those that exist are typically infrequent with short operating hours. A cross-suburban trip often involves going towards the CBD, waiting and changing to an outbound bus.

Many bus routes parallel rail lines for quite long distances. Adelaide currently has a mixed suburban network with some lines operating diesel trains and other operating electric trains. Frequencies are generally every 30 minutes off-peak except for the Seaford line to the south, some stations on the Gawler line to the north and the inner section of the Outer Harbor and Grange lines. This means that if you're on an inbound bus that intersects a rail line but continues to the city you are probably better off to remain on it due to the possibility of a long wait. Adelaide's rail network is relatively large for a city of its size but the low frequency of trains, the location of stations (often in industrial areas further from houses than the main roads along which buses run) and the large number of parallel bus routes make trains less popular. 

The map below is the north-western quadrant of the existing Adelaide network. Almost all stations on the Outer Harbor line are served by the roughly hourly Route 150 bus almost duplicating the train all the way into the city. As well Route 333 operates just three times per weekday day to Port Adelaide. Further south many roads have multiple routes with confusing numbers. Examples like this illustrate why people reviewing the network might see a need for change.

The proposed Adelaide bus network didn't just come out of thin air. The Australian city nearest in size to it is Perth. Perth restored, electrified and extended its small three-line rail network and reconfigured its buses to feed it. It proved a great success with an integrated network useful for travel to a wide range of destinations. Adelaide decided to follow with rail electrification and 15 minute interpeak frequencies, starting with its Seaford line. Full adherence to the Perth formula would also require comparable reforms to its buses to maximise rail patronage, usage of the bus fleet and frequency in local suburbs.

 More recently (2018) GTA consultants wrote a report describing a network approach for Adelaide. The concepts will be familiar to anyone who's studied bus network design and reform. It  proposed a three level hierarchy comprising priority corridors, regular routes and tailored services. Priority corridors would be the main train, tram and bus corridors while regular routes would provide a 'second tier' coverage for areas between or beyond the main corridors. Tailored routes could be your special services, for instance peak only, school and shopper services serving particular concentrated needs. 

The proposed network reflected the above principles. It had an ambitious timeline for its introduction. Feedback would be in June and July with review in August. New contracts (on existing routes) would start in July with the new network starting late 2020 in most areas.  

More Go Zones (that is priority corridors with weekday service every 15 minutes or better) and simpler services were the new network's key sales messages.  

The map below shows where the new Go Zones would be. Most were in the inner to middle suburbs within about 15 km of the CBD however some extended further out. Red were existing while green were proposed.

The Go Zone increase would boost the population near a weekday service every 15 minutes or better from around 500 000 to nearly 700 000.  This is more than half Adelaide's 1.3 million population. This coverage of frequent service is rarer in the suburbs of other Australian capitals, especially Brisbane, the other mainland bus-dependent state capital.  

Most areas that did not get a Go Zone got a 30 minute interpeak weekday bus frequency. These were described as 'connector' routes occupying the middle hierarchy. This is again superior to other cities. For example 40 minute frequencies are common in Melbourne and 60 minute frequencies are common in Brisbane and Perth. Fringe areas though received hourly services, described as local routes. 

The trade-off for all this is that many bus stops would close. The initial number given was about 500. However about 400 more stops would become for school bus use only, making the total number of stops cut nearer to 1000. Early warning of the reception this would get is contained in responses to the Transport Ministers's tweet promoting the network (some below).

Designers of frequent networks work on the basis that most people are willing to walk a little further to a better service. However difficult street layouts or over-severe cuts can make the extra distance involved more than 'a little'. You can see the effect of the sparser network on this proposed network map for north-west Adelaide.  Routes are a mix of 15 minute Go Zones and 30 minute Connector routes. There is less duplication particularly near the line to Outer Harbor. Those not willing or able to walk would get a flexible route bus similar to the phone one hour ahead service in Gawler.

Below shows the whole proposed network. The detail is too fine to see. However the colours of routes, with the red Go Zone, blue Connector and green Local are visible.

For more detail you can download the proposed network map pdf here (large file). You may wish to compare it with the existing (very complex) network map here.

A point of interest is that like Canberra's new network, Adelaide kept its stark weekday/weekend divide. Under both the new and old network Go Zone corridors had a basic 15 minute weekday/30 minute weekend frequency. This weekday-centric pattern is the same as Melbourne's SmartBuses, which in our case has caused weekend overcrowding. The middle-level connector routes do similar with service every 30 minutes on weekdays and 60 minutes on weekends.

New networks in Houston and Auckland, in contrast, operate their main routes every 15 minutes seven days per week. The same is true for the premium BUZ services in Brisbane and the 900-series routes in Perth (sparser but more frequent than Adelaide's Go Zones). This tram-like frequent all week service is desirable as it can promote low-car lifestyles and thus encourage higher bus usage at all times.

Also notable is that Adelaide's proposed network remained very CBD-centric. Part of this is because Adelaide is long and narrow. Its trains are less frequent than Perth's and stations are closer together, making them slower. Adelaide's one city centre station is on its northern fringe and road congestion may also be less. This could make the time savings for bus + train trips versus bus only trips less compelling than in larger cities with worse traffic and more stations. Even in the new network many routes still operate to the CBD. Some of the few cross-suburban routes went from every 60 to every 30 minutes but there are no frequent circumferential routes such as operate in Melbourne and Perth.

New network abandoned

The promised 'better, faster and more frequent' bus network was not to be. Two weeks into the consultation period the SA premier announced that the proposed network would not go ahead. Media report are here and here.

Information on the new network was immediately stripped from most (not all) SA government websites. Visitors to the Adelaide Metro site got this:

With the premier rather than the minister doing the announcing it couldn't have been a sharper repudiation of the proposed network and, indirectly, his transport minister Stephan Knoll.  Immediate prospects for bus network reform in SA seem grim with it regarded as a 'hot potato' that no politician will wish to touch for at least a few years.

12 lessons from Adelaide

What can we learn from this experience?

Quite a lot.

The mix of state a government with a reputation for cuts to public transport (eg 2018 budget and plans to privatise), poor communication from blinkered technocrats in the bureaucracy, a public wary of change and nervous politicians proved fatal. 

Although the theory seemed good, there were issues with the design of the proposed network. Opposition to it focused on the number of stops deleted. And some trips undeniably became more difficult.

Information was pulled so soon after the premier's announcement to abandon that I didn't see the changes in detail. However I did notice how they were communicated. Overall insufficient area-specific detail was provided, especially for those who would lose stops or have their journey altered.

Here's some possible lessons from the Adelaide experience. If they had been adhered to this post might have to introduce rather than farewell the bold proposed network.

1. Failure to explain the new network's impact by suburb Before it was taken down the website notes on the new network was divided into a few broad regions. You could select an existing route number and get short notes on its replacement route. However many suburbs are served by multiple routes. And people wanted more fine grained information on what was happening to their stop and any nearby alternatives. 

Avoiding suburb-specific information probably made the new network look worse than it was. It would also have made it easier to spread fears about it. Each suburb needed its own write-up on what was changing. Users should have been able to compare suburb network maps side by side. A feature where people can click on their stop to see how the new network affects it (including nearby substitute stops) would have helped. It can help public acceptance if people see that there's swings and roundabouts, not just straight cuts.  

Houston recently transformed its bus network along lines similar to that proposed for Adelaide. Its public information included dual trip planners on the old and new networks. People could plan test rides. This personalised information helped reassure people that the new network would be better, or at least no worse, for them.

2. Old network maps can exaggerate current service levels A new network map with fewer lines on it can invite unfavourable comparisons with existing map with more routes. The first instinct is for people to find their house and panic if service has been removed from their nearest stop. However old maps with complex networks can give an inflated impression of current service if they show buses that run only occasionally as prominently as more useful services.  

It may have helped consultation if occasional routes were made less prominent on the existing network map so that comparisons would be fairer. Also line thickness as well as colour might have made the frequent corridors stand out more in both the before and after networks.  

3. Operating hours are important The current network contains many part-time routes with occasional trips or limited operating hours. Some routes run Monday to Friday, some Monday to Saturday and some seven days. This might not be grasped by passengers until they try to plan a trip on a route only to find its timetable is unsuitable. One of the thrusts of a revised network is (or should be) to standardise operating hours so that more routes on the simpler network operate 7 days over longer hours. Unfortunately the information on the new routes was sketchy with regards to operating hours, potentially underselling the revised network.  

4. Did they really need to cut late night routes at the same time? Adelaide has a few routes that operate after midnight on Saturday nights. Their occupancy might be lower than daytime routes but they are a high profile part of the network, cheaper than trains to run and help some get to and from jobs. Night routes are also justified as a safer alternative to drink driving. The new network planned to cancel most of these services, increasing opposition from another set of people you don't want opposed. 

5. Prepare the ground by briefing key people beforehand A lot of work is needed before a new network proposal goes public. For a start it should be run past important stakeholders like bus operators and local council planners to comment on its workability. Local politicians should be briefed on the network's impact on their constituency with area-specific answers to likely questions. Such background knowledge would help them sort genuine from non-genuine concerns. Apparently this did not happen, with even government MPs claiming no prior information or warning. The threat of a revolt caused the premier to intervene over the head of his transport minister. This is about the most damning thing a leader can do short of dismissal. 

6. People like timetables with specific times (even if buses don't keep to them) One feature of the new network was that some routes would operate at defined headways rather than at specific times. There's a lot of operational benefits to this especially on corridors where traffic levels vary. For example trips can be sped by not having buses stop at time points at quiet times. That improves bus utilisation and potentially enables more routes to have frequent service.

However it's only appropriate when frequency is very high (a headway based 20 minute timetable is not acceptable). Resource-optimising planners love the concept but politically it's easy to ridicule. Hence it needs to be implemented in a 'high trust' or at least 'low controversy' environment. Impending privatisations and stop closures are red flags that make passengers wary of change, even if good. Many will also remember the Light City Bus fiasco caused by the government accepting a cut-price contract before dumping the operator due to continued lateness. More recently, in 2018, there were budget cuts to services. Now might not have been the best time to introduce it.

7. Don't hide the bad stuff The initial word was that 500 bus stops would shut. However this proved to be an understatement. It  turned out that hundreds more stops would also lose public routes (though they would keep school services). Understatement of losses is particularly damaging if it's uncovered by the opposition, as occurred here. And it overshadows the more important question as to how many of the closed stops had useful alternatives nearby.

8. Reassuring people is important Even though people are often more fearful of what they might lose than what they might gain from the new network, the information stressed the opposite, for instance the number of people who would gain Go Zone coverage. This was instead of reassurance like the number of people who would retain coverage with the new network. Better suburb specific information on alternatives would also have helped lessen concern.  

Reassurance and promotion messages have different roles. Stress the former when trying to win acceptance of a controversial network. Then once the network is operating and you want to get more people to discover its benefits then ramp up the marketing by selling the advantages. 

9. Too much at once? Adelaide might have tried to do too much too soon. Auckland, a similar city, split their reforms into eight regions done over three years. Perth is even more gradual, but the cumulative result is transformative with reviews being routine business. It's true that Houston successfully did all its reforms in one go. However, because it has a stronger CBD and a weak rail network, Adelaide's bus ridership includes many middle class commuters coming from politically marginal seats. It also has an older population, with a median age of 39 versus Houston's 33.  

A more measured pace with review regions like south, central, north and hills might have allowed a more consultative approach. If there were problems with proposals in the first area lessons could have been learned for other regions. And because issues would have been smaller (rather than city-wide) they would likely have been able to be addressed by the minister without the premier intervening and cancelling the lot.

10. Would a staged approach have been less scary? As well as dividing changes by area, planners may, as a first cut, make their initial proposals less radical than what they might see as the ultimate network. For example rather than have all routes in an area operating frequently on a coarse grid they might skew the balance a little away from frequency to coverage. 

If a particular coverage gap looks a bit too stark or roads are too hard to cross they might introduce an interpeak-only shopper style route for in between coverage. Especially if this route does not increase the network's peak bus requirement its operating cost should be fairly low while also providing a service for people who are less mobile. 

That would have been a compromise but may have defused some opposition. Instead of the complete abandonment that happened it might have enabled a revised new network that removed most complexity while delivering some frequency gains. The option always exists of modifying or deleting it in a subsequent mini-review if usage proves disappointing. Those who follow what Transperth does sometimes see this approach where occasional deviations and duplicative routes fade away over several timetable changes.  

11. Don't assume universal technology use, familiarity or ownership This is another of your 'technocrat vs traditionalist' battles. Departmental bureaucrats and the private sector consultants who sometimes advise them are overwhelmingly tertiary educated, middle class and comfortable with technology. Many bus passengers, especially in a city like Adelaide with an older population skew, are not. 

Technology dependence has problems on two grounds. Firstly online-only consultation can exclude people from participating. Secondly the network's proposal for 'on-demand' buses in some areas may disenfranchise some users. The need to maintain two lots of credit (mobile phone and transport smartcard) may be an issue. The one hour prior notice requirement is another barrier that makes 'on demand' services less flexible than their promoters claim. While favoured by geeky technocrat types, on-demand buses are rarely operationally cost-effective unless service levels are cut.  

12. Where's the personal touch? What was proposed was a major network change that would have changed how people get around, and, even, in some cases, whether they have service nearby at all. It's standard for cities planning changes of this magnitude to have drop-in sessions where passengers can discuss concerns with transit planners. Consultation for Adelaide appears to be online only, apparently with no face-to-face meetings.

COVID-19 has undoubtedly made public engagement harder. Adelaide has an older age skew. Seniors are likely to be in COVID-19 high risk groups, less familiar with online engagement methods and are more likely to be disadvantaged if bus stops are closed. But when angry their letters, petitions and phone calls can spook marginal seat politicians. Especially Liberal politicans whose most loyal supporters are middle aged and older people. Now you know why there was a party backlash and the premier over-ruled his minister.

More than usual thought should have been given for those less comfortable with digital technology. Maybe they were, for instance notices at bus stops and surveys on buses? However this would not have been fully effective with fewer people travelling. Particularly amongst older high risk groups. Mail-out surveys might have been another approach. 

I offer no firm solutions. But some should have been found given the changes were intended to be long term. Though timelines were already tight so a deferral couldn't have happened without putting off the new network's introduction.

Even after the COVID-19 threat passes transit agencies need to think about consultation methods that maximise community representation and participation. Approaches that seem procedurally fair can end up skewed in ways that either downplay or magnify discontent out of proportion to its size.

For example town hall feedback sessions at off-network locations exclude many working people if held during the day and people without cars if held at night. The majority of attendees can be privileged local council or advocacy group representatives (who don't necessarily ride buses themselves) instead of the 'real' public that needs to be heard.

A tiny group (not all of whom are bus riders) can magnify a gripe out of all proportion to its importance or the number of people affected. This group should be heard but not so they drown out those who a new network might benefit.  Politicians and other decision-makers sometimes visit these sessions and may get an unfair impression of a network's merit.

The best way to get a reasonable sample is to use a variety of feedback means including online, social media and in-person. All methods need as large and representative number of 'not organised' passengers as possible including people who are 'time poor' and 'don't do meetings'. Along with non-passengers who might use buses if the network was better.

The in-person components of this should not include off-network locations where people must make a special trip. Instead engagement should be at places where people are at anyway (eg train stations, bus interchanges and large shopping centres at both weekday and weekend times). Weekdays are good for seniors while weekends suit many working families. Shopping centres attracts a good sample of potential as well as existing passengers.

Drop-in sessions, where people can stay as short or long as they like, seem to work better than a set time meeting. Surveys can be effective. These should be available in both printed and online formats. Social media and direct email can also be effective to encourage participation, especially if you have a ticketing system that has registered users that have email addresses. Communications people from transit agencies tend to engage with city-wide media more than local media. With the closure of many local newspapers alternative means of reaching people are needed, for example through social media, area-based Facebook groups, local community groups and even shopping centre noticeboards.


The failure of bus network reform in Adelaide is bad news. Passengers will be saddled with complex and confusing routes for years to come.  Politicians will be scared to support even less radical reforms due to the risk of a public backlash despite its benefits around the world (including in parts of Melbourne).  

Adelaide's sobering experience has given things other cities like Melbourne can learn from when trying to improve bus networks. I've mentioned a few. If you have any more please leave them in the comments below. 

PS: An index to all Useful Networks is here.


Michael said...

To much to quickly to many stops gone numbering to confusing would of thought they learned much from the overlays from the 1969 mats plan small increments would have worked much better through routing of inner North south East and West routes is where I would of started Adelaides bus upgrades with an easy numbering system better go Zones and back to the very fast T routes system then the hills routes if all went well O-Bahn and north east then outer south and finally outer north

Malcolm M said...

The public consultation was woeful. I put a question by email and on the Facebook page about how buses from the Aberfoyle Park area would "feed" into a new train extension to Flinders when the bus stop was a 300 m walk from the station. There was no attempt to answer the question either by email nor on the Facebook page. It wasn't just me, but there was no attempt to address anyone else's question on Facebook either, only a message to their feedback formally. This is a playbook for failure of the whole project, because it allows rumours to start simply because of lack of information about what is being proposed. It would have been a wise investment to have a member of the project team answer such questions on Facebook, perhaps providing details of what was proposed and that the team welcomed the feedback to modify the final plans.

It would be good to have had a journey planner for the current and proposed system.

There were questions on the Facebook page about which bus stops would be scrapped, and it seemed the development team hadn't planned to that level of detail. There was nothing about the benefits of scrapping some stops through better journey times. On many of the major roads in Adelaide, each stop a bus makes causes it to miss a green wave in the traffic light system.

There was no background technical report on the website. This could have put the case for a budget-neutral bus reform, and listed the patronage of each route per hour of service. Such a report is bound to have been prepared, but was kept hidden from the public.

Brendan K said...

I hate to say it but I actually have some sympathy for the SA Libs and public servant technocrats over the poor/non-existent public consultation that they're being reamed about. The feeling would have been any attempt to reform the bus network, however beneficial for most, would result in angry pushback and opposition that they just didn't want to deal with and suffer the political repercussions of. So they gave it a shot deciding to go hard and go early, trying to do it all at once and with as little public input as possible, hoping that it would wash over most people (as has happened with bus route reform in other smaller cities) but alas this did not happen in Adelaide. I can't help but wonder if all along they had an inkling that the reform wouldn't succeed first time and that any other watered down "reform-lite" version of the changes would be easier to swallow when they inevitably come along. Either way it's a shame that Adelaide has missed an opportunity at genuine network reform for its main mode of public transport for what may be years to come.

Unknown said...

What a thoughtful, lucid and wise reflection. Hits the nail on the head.

GMatt said...

Interesting comment on the weekend divide for the Canberra service. The change in frequency between weekend services and weekday services is considerably large.
On weekends all services operate every two hours. The result is that multi service journeys are not viable.
I have been digging into the recent ACT timetable working out the process as I go.
Would be interested to know of any published material that can assist with analysis.