Tuesday, January 11, 2022

TT #148: Planning & scheduling efficient bus networks

I often talk about networks and timetables as finished products but not enough about the structures, processes and cultures that lead to the creation of good ones. I want to partly remedy that today. 

I first recommend reading this item by Matthew Rosenbloom-Jones, a US-based transport planner and scheduler, as background. 


Mr Rosenbloom-Jones wrote about the decline of schedulers and the rise of planners in how timetables happen. In particular, how (more remote) planners may lack the intricate operations knowledge that schedulers (who can smell the bus workshops from their offices and may eat lunch in the same room as drivers) pride themselves in. 

There's likely class and formal education differences too. Degreed planners might never have driven a bus. In Melbourne's case, partly because so many live in tram-rich inner suburbs convenient to their CBD DoT office, even riding one might be a foreign experience. Whereas schedulers possibly started as drivers and got to their position via a driver, maintenance or depot management roles. Thus they are rich in operational experience. But both might be weak in seeing things from a passenger view and be too set in their ways due to bureaucratic or organisational inertia. 

The article is American but many of the same forces apply here. And not just in transport.

For example the decline in prestige of narrow industry and employer-based crafts and the rise of more general and portable degree-based credentialism to soak up the larger number of university graduates now compared to 50 or 60 years ago.

We have also seen the importation of a cadre of transient international manager-overseers and, at the bottom end, an also imported and insecure 'Coolie class' to do outsourced grunt work in areas like cleaning and security. The replication of late 1800s colonial class structures nearly 150 years later in modern franchised rail was discussed here. This is less apparent for our buses, but schedulers fit in to the 'skilled artisan' class category described in that article with planners being dependent functionaries constrained by their DoT management superiors. 

Transport has always had some degree of interest from government and thus regulation. One can go back to concessions being granted for railways and tramways. Not to mention the 'Octopus Acts'. Then when buses started there were rules about them not competing with (now) government railways or even themselves. 

When public transport operations could no longer pay their way (from about the early 1970s) the government went from being regulator to payer. Further involvement came when fares were integrated  across modes in the early 1980s.

The emergence of subsidies should have implied a degree of network planning capability as inefficient routes became a matter of public money cost-effectiveness rather than a private matter for the bus operator. In practice this was only inconsistently applied. The early 1990s service cuts were particularly blunt with popular routes cut but some duplicative services surviving. And in the decades following it has taken an inordinately long time for us to be scrapping or combining duplicative routes with much work still to be done (although the pace is picking up with the very minor 673 to be deleted this month). 

Public concern about the lack of network coordination rose about 12 years ago. The Department of Transport were basically contract managers shovelling money to transport operators regardless of whether bus routes were efficient or connected with trains or not. Besides efficiency could be measured in different ways. 

For example a bus route that's 22 minutes long running every 25 minutes with two buses might be considered to be efficiently scheduled. Its operating hours may also support efficient driver rosters with a minimum of overtime. However it is not an effective service if part of its role is to connect with trains every 20 minutes from early until late. 

The PTUA found little timetable coordination between transport services in Melbourne. 2010's important Lavanas / Stone ATRF paper found no one was responsible for it. The momentum generated led to The Greens and then the Coalition to propose a public transport planning body, though Labor saw nothing wrong with current arrangements. 

Labor lost government to the Coalition which established Public Transport Victoria. This included the contract managers and planners from DoT, the marketers and data people from Metlink plus people from the Transport Ticketing Authority. Hence it simplified institutional arrangements regarding public transport, although it wasn't quite the promised 'one stop shop'.

While not much new funding was provided for bus services, the pace of network reform quickened. 2014, 2015 and 2016 saw major changes in some areas. Some happened after Labor regained office but all were planned during the Coalition's time.

From mid 2015, early in its term, Labor started to wind back potentially controversial service reform in favour of its 'Big Build' agenda. A couple of years later it folded PTV back into a larger Department of Transport that de-emphasised network planning. The term 'bus reform' was hardly heard. But last year it set up a bus reform team to revive it again.

Below is the 'no more excuses' speech from the late Paul Mees delivered before the 2010 election. It proposed significant change to the way we planned public transport in Melbourne. 

The complications of designing and specifying service

Roughly speaking if an existing bus route is upgraded or slightly extended the incumbent bus operator gets to run it. If it's a new route it's subject to an open tender process. Both changes require a service specification document prepared by planners in the Department of Transport. 

The specification gives service-related details like the route required, times of the first and last service, days of operation and frequencies. There may also be requirements such as coordinating with trains at certain stations served in directions that favour peak travel at various times. 

Sometimes the requirements are so complicated that the Department basically has to write a timetable prove the specification can work with reasonable efficiency. Thus their role might be more like the hybrid 'planner-scheduler' that Rosenbloom-Jones talks about. Also to get an idea on costs other factors like run times, the number of buses and the number of stops required need to be known.   

Even that is inadequate if you want to be absolutely sure you are getting value for money. There may be a case where a particular specification causes an inefficiency with bus usage or driver rostering. Or, conversely, it may be possible to combine a new route with an operationally inefficient existing route (eg a bus with a long layover time) to deliver efficient services on both old and new routes at lower cost. In other cases a very small increase in cost might enable a disproportionately better service. 

Some bus depots are at bursting capacity and need adding to before extra buses can be purchased. Another operator may have poorly used school buses that could be run to provide a better service to the public at other times. In another instance a reformed public bus route may be so similar to an existing school route that the special school route is no longer needed and an extra bus is available to boost frequency on a public route.  

All these matters are very bus operator specific. Senior drivers and schedulers in the bus company might be aware of them. But planners in a large transport department several degrees removed might not be, especially if their brief is simply to add a new route over an existing unreformed network. Without knowing it the specification they draft risks being expensive to deliver relative to the service provided or miss opportunities for low-cost improvements. 

Bus operators may regard their scheduling methods and fleet utilisation profiles as their 'intellectual property' to be jealously guarded. This is because it's part of their scheduler's 'craft' and is tied to their cost base which affects how low they can bid for business that is still viable for them. 

Even if bus companies were to share opportunities for low cost service upgrades (sort of like 'unsolicited bids' but for service rather than infrastructure) with the Department of Transport, the chance of the latter being able to find (say) $50 - 500k per year to upgrade one or a small cluster of bus routes outside the convoluted budget bid process is slim. This is while the Department of Transport is embarking on other adventures like new fixed or flexible roues over an existing unchanged network. 

One can see here how there is a risk of antipathy between planners (mainly with the DoT) and schedulers (mainly with the bus operators). And the planners themselves might not have enough influence in the Department to shout down some of the crazier and less efficient proposals their superiors may advocate.   
Planning successful bus network reform

What are the ingredients of our most efficient and wide-ranging bus network reforms, such as happened in Brimbank in 2014 and Wyndham in 2015? It was certainly not the administratively clean process of the Department specifying the services and telling the bus operator that's what they need to run. 

Instead it was an iterative back-and-forth process between the Department of Transport and the bus operators who contributed significant local knowledge. Although covering fewer routes the Brimbank reforms were more complicated in that they involved multiple bus companies who each needed to get a fair deal. Wyndham was easier because there was only one operator running all routes in the area. On the other hand Wyndham routes often had difficult coordination constraints due to the need to connect with (sometimes infrequent) trains on multiple lines. Sometimes specification requirements had to be compromised if 'greater good' improvements were possible. 

Hammering out the best solution required a cooperative approach between DoT planners, the bus company involved and a scheduling consultant skilled in the use of advanced scheduling optimising software. The value of this became apparent when some routes were able to be run more frequently than initially planned with the same number of buses.   

In both Brimbank and Wyndham there was some interaction between school and public bus routes. This was critical to deliver efficiencies needed to maximise frequency on as many routes as possible. Leaving school bus reform off a public bus revamp may be simpler but means that less can be done with a given budget, especially with regard to peak service frequencies.    

The smallest reforms, which should be routine business, involve minor timetable tweaks that fix run time issues or recoordinate with trains when their timetables have been adjusted. These have become more common over the last ten years in  Melbourne. 

However larger changes, that other cities (eg Perth) would also regard as routine business, involve the redistribution of service kilometres between routes to suit changing demand patterns are still quite rare. Fortunately this is starting to change, with the best recent example being low cost but very worthwhile timetable changes to routes operated by Transdev last year. Part of the reason for the success of this is that Transdev is big enough to have a mix of routes whose services could be said to be insufficient or excessive. Thus bus hours resources could be shifted from the latter to the former.  If the operator was smaller and/or it didn't have excessive service that could reasonably be pruned then the scope for reform is less without additional funding.  Transport authorities could encourage bus operators to submit ideas for these types of reforms, although this should not replace its own patronage monitoring and network review functions.

Adelaide showed how bus network reform can fail if too much is attempted at once. Transdev Melbourne's 2015 'greenfields network' attempt also failed following a change of government, a more risk averse minister and genuinely bad parts in its plan (that I'd have rejected too). 

Part of the latter was due to the limitation of single operator thinking and planning. Whereas Brimbank and Wyndham's networks were developed within the DoT (with input from bus operators and contracted specialist schedulers), the 2015 network was more Transdev's work more in line with the  outsourced franchising ideology of its contract and its payment method (which included incentives for increased patronage). 

Transdev's single-operator proposal was considered politically unacceptable as its consultation process was poor and it took resources from the west leaving them with a generally inferior service. While it was true that some Transdev routes in the western suburbs were poor patronage performers, the west also has popular routes run by other operators that justified more service. A genuine network approach would have kept bus resources in the west through an integrated multi-operator approach like taken in Brimbank while still scaling back its quieter or duplicative routes. 

This tale shows the limitations of the single operator approach for large network changes or shifts in service levels. Though administratively and contractually less convenient, a multi-operator place-based network reform would have delivered wider benefits and (likely) less of a public backlash. It is significant that unlike the 2012 metropolitan bus franchise contract, the 2020 version does not require the franchise operator to develop its own greenfield network (although it must still cooperate with other reform).    

Maintaining reform's benefits

It's not good enough to introduce a reform and hope it endures. Because it won't necessarily if there isn't the institutional memory to recognise what is good about a network and to retain it through successive timetable changes.

I have been a fan of multi-route timetables where two or three routes served a common corridor before branching out. Provided trips were evenly spaced they could provide a combined frequent service along the overlap section before branching out to provide local type coverage services to less dense or purely residential areas. Provided passenger information highlighted the frequent service you would have a larger yet still legible frequent network that passengers would use. 

The 2014 Brimbank network (or its follow-up echoes) specified such offset scheduling on at least two corridors from Sunshine. These included 400/427 along Forrest St and (later) 426 / 456 along Ballarat Rd. Each route would operate every 40 minutes but their times, staggered by 20 minutes, would evenly feed interpeak and weekend trains (every 20 minutes). Thus those on the combined corridors would get an efficient feeder service mostly operating at the same frequency as trains. 

These benefits are fragile if there isn't institutional knowledge and memory to preserve a network's benefits. In a transport department those who manage relationships with bus operators are not necessarily planners nor appreciate the intricacies of network design. A re-coordination might be approved without it being adequately assessed for suitability against an original specification. Or there might be a little tweak (such as a desire to add a deviation or extension) that adds so much run time that an even and well-coordinated timetable for the other 90% of passengers is harder to implement and falls by the wayside. An example of the atrophy that can occur to a formerly harmonised and coordinated network is this story about the aforementioned 426 and 456 in Brimbank

On the other side of town the 802/804/862 group is another example of uneven times. That didn't used to be the case; go back 20 or so years and it was an even 15 minute frequency. While Dandenong line trains have since changed from being every 15 to every 10 minutes, this is likely one instance where they would have been better off keeping the non-harmonised 15 minute frequency than to tolerate gaps of close to 20 minutes as now. Moving back west, the Route 180 to Werribee (nominally every 20 minutes) got a recoordinated timetable that opened up a 28 minute gap during morning peak. 

To summarise, reformed bus networks can atrophy. Original planning aims can be forgotten and neglected. For reason of lack of knowledge or expediency the Victorian Department of Transport has  not always been particularly good at preserving a network's service benefits over the longer term.  


The development of new bus networks in Brimbank and Wyndham were an example of a network planning vision meeting a scheduling craft. And, largely unseen by departmental planners, rosterers able to allocate driver shifts to run the scheduled services. 

The same results may not have been possible if a more hands-off approach, of the DoT simply putting a series of specifications out to tender, or giving them to an incumbent bus operator without detailed feedback and follow-up optimisation, been taken. Even if the routes were the similar frequencies would likely have been less due to less efficient bus usage. 

At the other extreme simply letting individual bus operators develop networks has also not been successful, especially in areas with connecting services and/or where they are not the sole operator. 

Hence a collaborative approach, as has been shown to work in Melbourne for some network reforms, needs to be taken to develop the best possible network from available resources. And once that's been done vigilance is needed to ensure that future timetable changes do not whittle away a network's design features such as good coordination and high frequency.      

1 comment:

Liam said...

An occurance I encounter too often is a bus service scheduled to coincide with a rail service, for multiple services. This ensures rail passengers see their bus depart while allighting the train, while bus passengers rush to a myki validator to meet a closed train door.