Review - Human Transit: How Clearer Thinking about Public Transit can enrich our communities and our lives, by Jarrett Walker

An observation from last September’s Australasian Transport Research Forum conference was the abundance of academics versus the paucity of practitioners.

I define practitoners as those who plan and operate transport routes or closely work with those who do. They have their own work and don’t write books and attend conferences as much as academics and bureaucrats.

Consequently it was refreshing to find a book on public transport service planning by an experienced practitioner. The author, Jarrett Walker, is a service planner and consultant to transport agencies in the US, Canada, Australia and New Zealand. His work was already known to me through his blog at humantransit.org

Having read books by academics such as Mees, Newman and Kenworthy, I was particularly interested in the different perspective that a practitioner might bring.

Academics thrive on research, teaching and (sometimes) political advocacy. However the affordability, practicality and acceptance of their recommendations may not frequently be tested.

Whereas practitioners need to be more practical and less theoretical. They must often work within existing resources; something academics might find limiting. Consultants can be practitioners too but won’t get work if not seen as useful.

In Human Transit, Walker describes himself as a plumber.

We don’t expect plumbers to tell us how to run our cities and our lives; unlike (say) priests, politicians and talkback radio hosts. They may hold values but cannot choose them for their clients. But they might be able to advise on the consequences of particular decisions; for instance the cheap quick fix versus the dearer but enduring repair. And they might point out that any option chosen must respect basic laws, like water flowing from high to low, for the system to work effectively.

City planners and developers deal the hand with which transit planners must work. Some decisions, notably with respect to density, are controversial. Walker, more than Mees, sees density as conducive to higher patronage. However he adds, correctly in my view, that location and geometry are equally critical when designing neighbourhoods for direct and efficient transport routes. To Walker these factors are public transport’s equivalent of the plumber’s basic laws. If disobeyed an area may never be easily or efficiently served by good public transport.

Transport networks themselves can be optimised to favour patronage or coverage goals. Networks that optimise patronage generally have direct, fast, frequent but more widely spaced routes, whereas a coverage-oriented network sacrifices directness, speed and frequency for short walking distances. Walker insists that transport authorities set a resource allocation policy between ‘patronage’ and ‘coverage’ routes.

Only after such priorities, and thus available resources, are known can the service planner design the network.

The book is a mine of simple but often neglected insights that would make public transport better. All are applicable to Melbourne. For instance the trade-off between directness, frequency and legibility on the one hand, and requiring that people change for some trips. We’re used to road maps having different thicknesses for different types of road, from lane to freeway, so why not do the same with transit maps, highlighting our most frequent,and thus most usable routes? And, on busy roads, the contribution of the humble pedestrian crossing to improving access to bus stops and the impetus it provides to gradually increase an areas’s walkability (and thus transit access).

Human Transit does what its title suggests. It forces the reader, from layperson to transport planner, to sharpen their thoughts and ask the right questions. I highly recommend it.

Human Transit is published by Island Press and is available via the link below.

What should a good CBD periphery transport system look like? Both Chris Loader/Charting Transport and Alan Davies/Melbourne Urbanist have picked up on the fact that as soon as you get beyond a City Loop railway station the modal share for public transport halves.

This blog takes the view that while there exist ardent revheads, greenies, transit geeks and the lycra set, the majority of the population are pragmatists in their transport choices. Especially for habitual and often time-critical trips such as the journey to work.

So why isn't transit so attractive when you get much outside the Hoddle grid?

A play with the Metlink journey planner gives some answers.

Let's compare comparative travel times between similar distance middle-suburban stations and various CBD fringe workplaces. For arrival at a start time of 8:30am, the following results were had:

Laverton - Melbourne Town Hall: 33 min
Laverton - Melbourne University (Parkville): 40 min
Laverton - Epworth Hospital (Richmond): 49 min
Laverton - Alfred Hospital: 51 min
Laverton - South Melbourne Market: 45 min
Laverton - Citylink Business Park (Port Melbourne): 52 min
Laverton - Waterfront City (Docklands): 54 min

Greensborough - Melbourne Town Hall: 38 min
Greensborough - Melbourne University (Parkville): 60 min
Greensborough - Epworth Hospital (Richmond): 42 min
Greensborough - Alfred Hospital: 60 min
Greensborough - South Melbourne Market: 61 min
Greensborough - Citylink Business Park (Port Melbourne): 65 min
Greensborough - Waterfront City (Docklands): 67 min

Mitcham - Melbourne Town Hall: 40 min
Mitcham - Melbourne University (Parkville): 50 min
Mitcham - Epworth Hospital (Richmond): 43 min
Mitcham - Alfred Hospital: 47 min
Mitcham - South Melbourne Market: 52 min
Mitcham - Citylink Business Park (Port Melbourne): 50 min
Mitcham - Waterfront City (Docklands): 60 min

Cheltenham - Melbourne Town Hall: 32 min
Cheltenham - Melbourne University (Parkville): 50 min
Cheltenham - Epworth Hospital (Richmond): 43 min
Cheltenham - Alfred Hospital: 39 min
Cheltenham - South Melbourne Market: 53 min
Cheltenham - Citylink Business Park (Port Melbourne): 50 min
Cheltenham - Waterfront City (Docklands): 62 min

In every single case a trip to the CBD was faster than to a CBD fringe location.

This is even where the destination was on the same side of town as the origin (eg Cheltenham - Alfred Hospital). It could be for this type of trip that driving would be relatively more competitive than existing public transport services, and so command a higher modal share (people are pragmatic, remember). There may even be a self-reinforcing pattern; if people can't afford to buy a home in the inner suburbs near work they may choose an outer suburb on the same side of town as their work (if not always the same train line).

City fringe destinations vary in the amount of time they add compared to within Hoddle's grid.

Waterfront City is the least accessible by fast public transport - from any direction - despite its frequent trams. Compared to the CBD, travelling to it can add 20 to 30 minutes to travel time each way. For a commuter that's 40 to 60 minutes per day, or an extra week added to annual holidays. Not suprisingly its public transport modal share is just 22%, compared to double or triple that a little to the east.

Why are the variations in travel time so great?

We have an extensive suburban rail system that has metro-style operating hours but commuter-type frequencies on its fringes. We have trams, but unlike some compact European cities we don't have a dense metro in the job-dense 2-5km core (see maps by Neil Freeman) that allows for fast local travel. Instead for the 'last mile' we rely on slow surface modes, notably trams and buses, often without their own right of way.

Also the fixed rail we do have is planned and run as a suburban - CBD distributor rather than doubling as a (small) inner city metro that with its spare capacity could take pressure off trams for longer trips.

The City Loop's weekday noon reversal introduces unwelcome complexity, discourages casual travellers and prevents direct access between some city stations at certain times. Quite different from the consistency and legibility of a real metro.

For many years the possibility of getting a direct train from South Yarra to Footscray (for example) was scotched by the philosophy that everything should run via the Loop and the primacy of the CBD as a destination. The May 2011 timetable will introduce some welcome improvements for such cross-town trips, provided dwell times at CBD stations are kept down. Especially important as Flinders Street and Southern Cross are the nearest stations to Melbourne's fastest growing areas at Southbank and Docklands.

The May 2011 timetable will increase the number of trains that will run from Flinders Street through to Southern Cross. This ties in well with an important bus rerouting from last year, namely the restructuring of Port Melbourne routes (235/237/238) to operate via Southern Cross Station. This restored a rail-bus connection that vanished when the tram superstop in Flinders Street was built and buses were forced away from Flinders Street Station. Peak frequency of the Port Melbourne buses is high but further work would lift their legibility and profile.

Route 401, between North Melbourne and Mebourne University, is an example of a 'bridging the gap' frequent service that has become well accepted by those commuting from the western suburbs to Melbourne University. The trip plan above for Laverton include the 401 as its bus leg as it proved the fastest travel option. Other opportunities for 401-style service could include Melbourne University to Clifton Hill or South Yarra to Port Melbourne, although bus priority would assist in providing worthwhile savings in travel time and efficient vehicle utilisation.

It seems to me that the CBD fringe is an area ripe for future patronage growth. Unlike the CBD, where public transport modal share is near saturation, there is significant room here to grow. But this can only come about if it can get the pragmatists on board. To achieve it a fresh look at how the CBD fringe and the close job-rich inner suburbs can be better served may achieve some surprising patronage results, as demonstrated by the success of Route 401.

More Aural timetables: analysing individual route connections

It's been a busy week in Melbourne transport; trains off the rails at sidings, disruptions due to power glitches, new fares and major events in the city this weekend. Nevertheless I'm still doing the timetable beat thing, and here's the latest.

Earlier experiments emphasised headway harmonised networks. If the network was frequency harmonised it would produced a nice rhythm, repeating in as little time as possible. An unharmonised network would sound more random, with no dominant rhythm, or a longer rhythm with the same connections only repeating every couple of hours.

Harmonised headways is a necessary but not sufficient condition for good connectivity. A bus matching a 20 minute rail service isn't good if it departs as the train is pulling in.

What is really important is the waiting periods between when the train arrives and the bus departs. Too short and it means a missed connection, especially if the train is running late. Too long and the bus is no faster than walking.

This part quietens down the pulse (but it's still there if you listen hard) and emphasises the waiting periods. It does this by applying a tone for the duration of the wait (again 60 min = 1 second, 6 min = 0.1 second).

Because the human hear is more sensitive to pitch than duration, I also used different pitches. I went for three pitches only (low, middle and shrill) to aid recognition. A low pitch is a long wait (>15 min), a medium pitch a medium wait (5-15 min) and a shrill pitch is a short wait (<5 min) which also risks being a missed connection if the train is late. I'd have liked to differentiate between a near-ideal 6 min connection from a longish 14 min connection, but that would have meant another tone which slows recognition. So I stuck with three. Music afficiondos will hear that I've used notes - something I intend to keep doing with future aural timetables.

At the end I've got a comparison between the harmonisation possible between 20 min/38 min frequencies and 20 min/40 min frequencies. The tonal difference is quite stark. Even though it's a slighly lesser frequency, a well-timed 40 min service can provide predictable connectivity is scheduled carefully, whereas a 38 min service has a plethora of good, bad and just missed connections, happening at seemingly random intervals.

There haven't been many comments on this, so whether you think aural timetable have potential or are just a useless curiousity, they'd all be welcome here.

Timetables and music: An attempt at synthesis

Planning timed transfer public transport networks, composing music and designing piston-based engines are very different fields. However they are all based on repetition, based on pulses, beats or cycles.

At least in the cases of transport and music, where I think the parallels are strongest, it is quite possible for a system to have multiple beats.

Music for instance may have a series of regular pulses to form a beat (possibly played by percussion). Superimposed on these may be contributions from string or wind instruments that together form the music. Two similar instruments add a ‘body’ not discernible with one, but they must be tuned and played in time with one another and the conductor’s beat.

Good playing on its own is no guarantee of a tuneful result if the music was badly composed. An occasional bar may sound musical, mostly by accident, but most others are cacophany.

It’s the same with transport. Trains and buses are the instruments. Their drivers are the players. The timetable is the sheet music they play to. Their writer are the composers. The Department of Transport is the conductor, albeit perhaps less sensitive to harmony, rhythm and music selection than its orchestral counterparts.

Buses and trains

Trains set the pulse to pull the whole network together, while buses play the more intricate pieces to provide coverage. Removing the buses, much like withdrawing the string and wind, leaves only a sparse beat and an incomplete network or composition. And the best playing will not guarantee connectivity if buses and trains play to different rhythms composed to different specifications.

If the beat of trains is based on a service every 15 minutes, while buses come every 20 minutes, the result is an unconnected network. There is still a pulse where optimum connection times repeat, but it’s a weak and infrequent one. The pulse is the lowest common integer multiple, in this case every 60 minutes, and so is longer than either the train or the bus on its own, lowering effective frequency. Multiples are even higher with incompatible non-clockface headways such as 23 or 35 minutes, making connections more random than repetitive.

The drone machine

One of my other interests is electronics. Last week I built a drone machine. This is a bank of audio oscillators that when used together creates all sorts of interesting sounds and beat notes. It also has a feature where a low frequency oscillator can trigger a higher pitched note. This creates a beeping whose interval and note can be varied. The combined sound when several such tones are generated can be quite unusual. It is similar to the early synthesisers of the 1970s.

You can hear a demonstration below

A synthesis

Using this machine raised the possibility of creating music from bus and train timetables, possibly based on arrivals and departures at a particular interchange.

The sound produced would vary according to whether buses and trains were part of a timed transfer pulse network or random arrivals at varying times and frequencies. The difference would be so marked that a brief listen by a trained ear (especially if accompanied by a spectrogram displayed on an oscilloscope) could tell almost as much about network connectivity as poring through numerous timetables.

There is no doubt that the concept is feasible. However it would need a synthesiser more advanced than the basic model demonstrated.

As with the synthesiser above, this concept requires a low frequency oscillator beating triggering a higher frequency tone for each route to be compared. The frequency of the low frequency oscillator would be proportional to the service frequency for that route. In other words a 30 minute route might beat twice a second, while a 60 minute headway route might beat once a second. This isn’t that critical; all that matters is that the relationship between the two is exact. But you need flexibility to program this to accommodate any clockface headway, and preferably non-clockface headways as well.

This would activate a higher frequency oscillator. For musical effect I would recommend its pitch be some sort of note, eg based on a certain piano keyboard key. But again not critical.

That is a single channel. Add more identical channels for each extra route. In all cases the low frequency oscillator needs to be set to be proportionate to the service frequency, while the oscillator it triggers can be a note different from the others.

All outputs are fed to a speaker via an amplifier.

A simple interchange might have one oscillator set to a 15 minute frequency (eg a train) while another is set to 30 minutes (for a bus). The train might be assigned a lower frequency note than the bus. The sound would be four low (train) beats per second, with every second one of those accompanied by a higher (bus) beat. The overall results would be an even beat. However moving the bus to a 20 or 40 minute frequency would change the relationships between the beats. The output would still be a rhythm but be uneven, repeating every second (20 min bus frequency) or two (40 min bus frequency). Probably easier to demonstrate than describe.

The limitation of this approach is that it is based on frequency only. Hence it can only measure frequency harmonisation; it will not detect if services are harmonised but their arrival provides a constant bad connection. To overcome this the starting time of pulses needs to be variable by introducing a delay that can be user adjusted. Once this is done two 30 minute frequency bus routes, one at h:05 and the other at h:20 will be evenly spaced and not on top of one another. Having to shift the onset of certain pulses by an accurate but adjustable amount requires additional circuitry. This is complex but feasible.

However the overall result should be worth it as it is true to life. And it would truly be music composed by timetablers, or more accurately the interaction of the work of several at interchange points.

A season for pruning? Part 2: Options for the Sunshine – Melton corridor

Part 1 identified some ‘dead wood’ routes; that is bus routes that had a purpose when started but now largely parallel others. They would be the first place to start if looking for efficiencies on the bus network.

Redundant routes survived because new bus funding outstripped planning capability, which had atrophied in the 1990s ‘bus drought’ when there were few new services to plan. By the time expertise returned (often consultants hired for the bus reviews) money for their full implementation was running out, at least in established areas.

Scrapping redundant bus routes in favour of upgrading others a further round of improvements, but this time without large additional funding. Unfortunately there is only so much you could do with simply axing ‘dead wood’ routes as they amount to no more than a few percent of total service kilometres.

Larger change requires defter hands in which the skills of the artisan, composer and economist are at least as important as those of the axeman. Artisan because the hand requiress a careful eye and trained mind that sees the network as a whole. Composer because it is within the beat set by train arrivals that buses are best planned. And economist because they value efficiency and understand opportunity cost.

A bus corridor makeover

I will present an example where rethinking a pair of routes may improve overall service levels and connectivity. The routes chosen include the western portion of Route 216 and all of Route 456. Together they form the main bus coridor between Sunshine, Caroline Springs and Melton. Two alternatives to the current timetables are discussed. Each has its own pros and cons. Hence there is always a need to consider the ‘greater good’ since it’s difficult to make any change without disadvantaging somebody.

View Larger Map

The area and services

Sunshine is an established rail-based centre populated largely by low-income migrants who use buses heavily. Its shopping centre and timetabled crossing of trains in both directions make it an ideal interchange hub. Caroline Springs is a 1990s housing estate built remote from the railway with only a basic bus service. Melton is served by country trains and has an older core surrounded by new housing. Unlike Sunshine, Melton is polycentric, with the railway station, main street shopping strip and enclosed shopping centre beyond easy walking distance from one another. Suburban trains operate every 20 minutes to Sunshine. Two bus routes, run by different companies, leave there for Caroline Springs; Route 216 (which has come from the city) and Route 456 (which extends to Melton). Both bus routes run approximately every 30 minutes, so do not connect with trains every 20 minutes. And neither do they mesh with one another, so they do not form an even 15-minute service. In addition Route 216 terminates just short of the Caroline Springs Town Centre so its coverage is very limited.

How many buses?

It’s useful to know how many buses are used on current routes so that changes can be costed. The number of buses depends on route length and frequency. It can be estimated from the timetable, either numerically or graphically.

During weekday off-peaks Route 216 takes 20 minutes to get from Sunshine to Caroline Springs. The 30 minute service provided would need 2 buses if run as an independent route. Due to the efficiencies gained by continuing this service beyond Sunshine, the actual requirement may be slightly less – let’s assume 1.8 buses to be safe.

Route 456 takes 55 minutes to do the trip to Melton. Hence a single bus can go there and back every 2 hours, meaning that the existing 25-30 minute service would need 4 buses.

Any resource-neutral service improvement would need to use no more than 6 buses. If it could be done with 5 buses it may be possible to free a bus for a route that needs it more.

Summary of the existing service

The departure list below shows existing weekday off-peak services and their connectivity with ex-City trains arriving at Sunshine.

The above shows that each train is followed by at least one bus. However connection times can be up to 20 minutes. Being significant compared to travel times (approximately 25 minutes to Caroline Springs and similar by train to the city) connection times have a large influence on end-to-end travel speeds and the relative attractiveness of public transport.

I did not examine connections for other travel patterns (eg city-bound train to bus & bus to train, but the uneven bus service intervals indicate likely similar results.

Here’s some key numbers:

Estimated number of buses used: 5.8 approx
Connectivity with train: no
Services per hour (Sph): 4.25 (measured over 4-hour period)
Average Bus frequency (Favg): 14 min (=60/Sph)
Maximum Bus wait (Wmax): 21 min
Maximum Train-Bus connection: 20 min
Service efficiency (ie Favg/Wmax) = 67%

Although the average bus frequency is 14 minutes (between Sunshine and Caroline Springs), maximum waits can be up to 21 minutes as the routes are not scheduled together. Worst case train connections are a similar figure, although most are better.

In the sample time period there are six more buses than required to consistently connect with each (off-peak) train. Connection quality varies due to unharmonised bus headways. Just missed connections are common; on six occasions the bus departs within two minutes of the train arriving. And effective frequency is lower than the 14-15 minutes it would be if services were evenly spaced.

This represents both potential inefficiency and opportunity. For it may be possible to deliver more consistent connections and/or higher effective frequencies with the same or fewer buses.

Scope for improvement?

Some ‘back of the envelope’ attempts to improve schedules are presented below. Not being a professional bus scheduler there will be aspects missed, but the general concepts, borrowed from timed transfer networks worldwide, should still be sound. The complications posed by multiple operators serving the corridor are not discussed. There may be scope to improve efficiencies by swapping routes between operators, or routes could be shared between operators, as already sometimes occurs (eg Route 400 between Sunshine and Laverton).

Option 1: One frequent route to Melton, meeting every train.

Achieved by deleting Route 216 past Sunshine and using saved resources to boost Route 456’s frequency to 20 minutes all the way to Melton.

Approximate bus times are shown on the graph below.

Graphs like these are commonly used by train schedulers. They show how many vehicles are required to run a service and indicate connectivity at interchange points.

Time is shown across the horizontal (x) axis and location on the vertical (y) axis. The steeper the slope of the line, the faster the travel. A horizontal line means the bus is stationary, for instance waiting at an interchange. Each coloured line shows the path of a bus; in this case six are required. The dots near the top signify train times in both directions.

Some numbers below:

Connectivity with train: yes (3 to 7 min in all directions)
Services per hour: 3 (entire corridor)
Average Bus frequency (Favg): 20 min
Maximum Bus wait (Wmax): 20 min
Service efficiency (ie Favg/Wmax) = 100%
Estimated number of buses required: 6

Several points are worth noting:

* The timetable aimed to create a timed-transfer network at Sunshine. Buses arrive, trains arrive, trains depart and buses depart. Passengers can simultaneously transfer in any direction with minimal waiting.

* Sunshine station presents special opportunities for service co-ordination since trains both to and from Melbourne arrive within a few minutes of one another. This doubles the number of good bus-train connections possible. Hence a bus there and back can connect with no less than 4 trains.

* Connection times are between 3 and 7 minutes. Given the standard for on-time train running is within 4:59 minutes, anything under 5 minutes is arguably too tight. It may be possible to adjust bus times by a minute or two to increase the probability of successful connections in peak travel directions.

* Recovery time is nearly all spent at Sunshine rather than Melton (where it is only 1 minute). Spending it at Melton would mean that the bus would arrive at Sunshine too late to make train connections.

* A nine minute recovery time at Sunshine has been scheduled. Buses arrive at :57 and depart at :06, to exchange passengers with trains in both directions around the top of the hour, repeating every 20 minutes. Such dwell times heavily use interchange space, with the bay being occupied nine minutes out of every twenty. However the best measure of bus bay efficiency is passenger throughput, and it is possible that despite the dwell time this could be quite high given the large number of connection combinations possible (4 per 9 minute occupancy, not including bus-bus transfers).

* Measures that could free space at bus interchanges and/or reduce dwell times while preserving connectivity have not been explored. One example is a layover area. Here the bus empties at a drop-off only bay near the station entrance, moves to a nearby layover area and returns to the designated pick up stop in time to leave on time. Another possible tool is ‘active transfer management’ which would also speed buses, especially routes carrying through passengers, such as orbital routes. Here the bus would be scheduled to leave just after the train is scheduled to arrive (say h:04 instead of h:06), but would be held back to at least h:06 if the train is late.

* The tightness of the some bus/train connections and the very short layover at Melton may mean that this timetable might look good on paper but be fragile in practice. It is in these marginal cases where cutting running time by 2 to 4 minutes per direction can greatly improve reliability, connectivity and bus utlisation. Removal of close stops, more direct routes and bus priority at intersections are the sort of cheap projects that could make the difference here.

To summarise, Option 1 is the highest service choice, providing effectively a TrainLink service between Sunshine and Melton. It scores well for frequency, consistency and connectivity. However it does not save buses (compared to the present arrangements) and poorly serves Caroline Springs. It also removes the ability for Caroline Springs residents to catch a direct bus to the city.

I would also wonder whether the service is overkill, and if there are more deserving routes for the resources. Much of the route is sparsely populated and may not justify a 20 minute frequency (this being transit hostile development where the greatest population densities are off the highway). While it is some distance away, the corridor is also served by diesel trains, which offer a much faster (but currently less frequent) service between Sunshine and Melton.

Melton’s population growth is likely to support either a more frequent diesel train or electrification, both of which would further lessen the need for a 20 minute bus service to Melton, since changing to a local bus at Melton Station may be faster than the direct bus from Sunshine.

Option 2: Connection with every train to Caroline Springs, with every second trip extending to Melton.

This option reduces Route 456 to a 40 minute frequency, carefully timed to meet every second train. Route 216 west of Sunshine Station becomes an independent route every 40 minutes to Caroline Springs, timed to provide a 20 minute frequency over the combined portion between Sunshine and Caroline Springs (to meet every train).

The graph below shows how this service might run: (click to enlarge)

It’s messier than Option 1 since two types of service are offered. Three buses provide a 40 minute frequency to Melton with similar times to every second trip of Option 1. In between are Sunshine to Caroline Springs trips, again every 40 minutes, arranged to provide a 20 minute frequency along the common portion. The existing timetable allocated about 25 minutes to do this trip, so two buses are required to operate this service.

Some numbers:

Connectivity with train: yes
Services per hour: 3 (to Caroline Springs), (1.5 to Melton)
Average Bus frequency (Favg): 20 min (40 min Melton) Maximum Bus wait (Wmax): 20 min (40 min Melton)
Service efficiency (ie Favg/Wmax) = 100%
Estimated number of buses used = 5 (Melton 3, Caroline Springs 2)

There are several points worth noting:

* This is a lower service option for Melton, with the non-clockface 25-30 minute Route 456 service reduced to 40 minutes (also non-clockface). However unlike the current arrangement it provides a consistent connection with every second train. And if introduced in conjunction with an improved Melton train timetable the frequency cut may not be missed.

* The Sunshine – Caroline Springs portion (216) would have fewer buses and lose its direct link to the city. However it would gain better train connectivity and a less lumpy timetable. Reliabilty should also improve as delays affecting the Sunshine-City portion of Route 216/219 would be quarantined. Travel to Footscray and the City would remain possible through better train connections or changing to Route 216/219/220 bus (a combined 7.5 minute frequency) at Sunshine.

An initial draft had evenly spaced departures from Sunshine towards Caroline Springs and Melton. However return trips had two arrivals every 40 minutes at Sunshine. This is an effective frequency of 40 minutes or a 50% service efficiency – both unsatisfactory for a corridor with a current 14 minute average frequency.

A way around this is to hold each Route 216 return journey at Caroline Springs by 20 minutes. This gives us a 20 minute effective frequency and a 100% service efficiency, which are both good. However having buses lay over for about as long as they are moving is inefficient.

How could the idle time at Caroline Springs be better utilised? A look at the existing Route 216 terminus indicates it terminates in housing, distant from local trip generators. A short (2km) extension to Caroline Springs Town Centre would lessen idle time, serve a local trip generator and connect with other routes.

This arrangements may save nearly one bus compared to the current arrangements. Possible uses for this resource in the Sunshine area could include: Boosting frequency of Route 471 from 25 to 20 minutes to connect with trains, extending Route 451 north to Watergardens Station, extending Route 454 southwards to the industrial area, extending Route 417 northward to Sunshine to better cover industrial areas, later trips to Derrimut (Route 400), or extended coverage in Sunshine North are all contenders. I will not pick which is the highest priority, but some are likely to attract more patronage than economising parts of 216 and 456 would lose.

Option 3. Melton to Caroline Springs via Sunshine.

This can provide a service outcome similar to Option 2, depending on the treatment of buses at Caroline Springs. Instead of the routes being independent they are interlined, ie buses from Melton do a short trip to Caroline Springs via Sunshine and then return to Melton via Sunshine.

Due to the backtracking and dwell time at Sunshine there are no advantages in this for the passenger. It would only be considered if it allows the service to be run with fewer buses.

Conclusion

The above examples indicate the potential for service planning to improve bus network legibility and connectivity. Two options for the Sunshine – Melton corridor have been presented.

A season for pruning?

It’s been a wet winter, the flowers are out and plants have grown fast. Gardeners are out with the shears, trimming undergrowth and unwanted branches.

Our bus network has had a similar windfall. A 15-year drought ended with many new routes being added in the last five years, a growth rate without recent historical precedent. Operating hours broadened, coverage spread and patronage surged.

Students of public administration will be aware that being required to spend large sums in a short time does not guarantee top value for money or high accountability. Careful planning and effective targeting of expenditure can be a casualty of blanket 'one size fits all' funding increases.

In public transport this could occur by boosting services on a route without regard to its network importance, relationship with other services or opportunity cost. Hence some corridors got more buses per hour but ‘lumpy’ timetables with long waits. Roads with overlapping routes were sometimes overserviced while adjacent high employment areas missed out. And buses are not all headway harmonised with trains like they are in Perth, despite our sometimes lower train frequencies making connectivity even more crucial.

The plant analogy can go further. If the bus network was a tree, parts would have strengthened greatly but others would still be weak, bushy and hardly alive. There would be many twigs, some well-placed and others with odd kinks formed years ago for reasons long forgotten. Not all supporting branches would robustly connect with trains or directly link suburbs.

Too much undergrowth can retard the funding, marketing and visibility of direct high-service routes most likely to attract patronage. Tangled scrub can also appear impenetrable, especially for commuters who drive past it on the way to the station.

If the political and economic climate favours service cuts, there can be coverage gaps. If it favours service increases there can be duplication. Our current network has both, reflecting past feast and famine. The latest feast has vastly improved buses and driven patronage to forty year highs. Nevertheless there remains great opportunity to carry even more through an efficient network that neither over or under services.

The patronage potential of buses is clearest when comparing their ridership with trams. Buses ‘should’ carry more than trams as their catchment population is higher, they exclusively serve the largest suburban shopping centres and run in the suburbs where most people live. But the actual figures are very different; 170 million for trams and barely 100 million for buses, indicating much higher service and patronage intensity on the smaller tram network.

This large disparity can’t all be due to trams’ CBD running, population density or a passenger preference for rails in tar. For there exist bus corridors that already have timetables and patronage that approach those of trams. The fact that buses don’t run on rails and overhead wires is no reason for low average patronage. However trams’ directness, legibility and frequency show aspects of successful service planning that can effectively apply to buses (as confirmed by usage of high-service ex-Met and SmartBus routes).

Both directness and legibility are a result of good planning, which is relatively cheap. Frequency costs scarce recurrent expenditure. But some can be raised internally from restructuring less efficient routes. However a growing city with a service backlog will also require increased external funding to develop the network. Such funding requests are more likely to succeed if there is evidence that the existing network is already planned efficiently and additional resources would boost rather than duplicate services.

Reviewing, untangling and making legible the bus network is key to making it more useful and boosting patronage. A good first step is to examine the current network for corridors that could be simplified. The multimode frequent service maps as presented above are useful tools for this purpose.

My examination of these maps has produced the following ‘dead wood’ portions of routes that may warrant pruning if 'greater good' gains can be obtained elsewhere:

219 (portion): Part west of Sunshine duplicates 903. A weekend variation serves areas covered by Route 471.

280/282 (portion): A local route that duplicates the 901 SmartBus along a low-density residential area (Foote St/Reynolds Rd Templestowe).

246 (Latrobe Uni extension): Overlaps with other routes between Clifton Hill and Latrobe Uni.

286 (entire route): Largely duplicated by two SmartBus routes (901 and 906) along Blackburn Rd

293 (part): Duplicated by new SmartBus Route between Doncaster Shoppingtown and Eltham (Main Rd roundabout).

295: Duplicates 903 along Station St between Box Hill and Doncaster Shoppingtown.

340/350: Overlaps with 250 between Ivanhoe and Latrobe Uni.

445 (part route): Duplicates other routes between Werribee Plaza and Hoppers Crossing Station. Truncation could allow removal of stopping restrictions (including to a local shopping centre) which lessen legibility.

478/479 (part route): Duplicates Route 477 and tram between Moonee Ponds and Airport West.

479 (City – Moonee Ponds portion): A weekend extension that duplicated by a frequent tram service.

483 (entire route): Freeway service for Sunbury. Will become less necessary after rail electrification.

500 (entire route): Duplicates 901 between Broadmeadows and Melbourne Airport. Duplicates 479 between Melbourne Airport and Sunbury.

544 (part route): Duplicates 901 between Roxburgh Park and Broadmeadows

563 (part route): Duplicates tram along a large section of Plenty Rd and 901 between Greensborough & Plenty Valley SC

623 (part route): Duplicates 626 and 900 between Chadstone SC and Carnegie. Scope for rerouting along Neerim Rd to replace portion of 624.

673 (entire route): An hourly service entirely duplicated by parallel longer routes

691 (part route): Monash Uni extension – replaced by direct high-service 900 SmartBus

694 (entire route): Largely parallels 663 & the extended 688 for all but a few stops.

745 (entire route): Four occasional routes with very low patronage

777 (entire route): A short route with few services and trip generators

Apart from removing redundant routes (relatively easy since no coverage is sacrificed), there are other things that could further simplify the network and increase effective service frequencies, preferably using resources saved from route rationalisation. The most obvious is more evenly scheduling overlapping routes that share a corridor. Other gains could come from rerouting that lessens duplication and improves coverage. Such changes won't please everyone, but not considering them may result in higher gains foregone. Some examples of this type will appear in a future post.

New Frankston line timetable and bus connectivity

The major feature of the October 2010 Metro train timetable was the upgrade of the weekday Frankston interpeak frequency from fifteen to ten minutes. As frequencies increase dependence on timetables lessen. And many passengers would be treating ten minutes as a ‘turn up and go’ service.

Intersecting tram routes would also be considered ‘turn up and go’, at least on weekdays. However for buses reference to a timetable is still important to ensure short journey times.

Connectivity and effective service frequency are maximised if train and bus headways are harmonised. The extent of frequency harmonisation varies across Melbourne as each mode is often independently scheduled.

This survey shows the effect of the Frankston line train frequency boost on bus headway harmonisation for each bus route serving stations between Caulfield and Frankston.

Station – Route – pre Oct 2010 – post Oct 2010

Caulfield – 624 – yes – yes
Caulfield – 900 – yes – no

Ormond – 630 – no – yes
Ormond – 625 – yes - yes

McKinnon – 626 – yes – yes

Bentleigh – 701 – yes – yes
Bentleigh – 703 – no – no

Moorabbin – 811/812 – yes – yes
Moorabbin – 823 – yes - yes
Moorabbin – 824 – no – yes
Moorabbin – 825 – no – yes

Highett – 827/828 – no – yes
Highett – 708 – yes – yes

Cheltenham – 600/922/923 – yes – no
Cheltenham - 811/812 – yes - yes
Cheltenham – 822 – yes - yes
Cheltenham - 827/828 – no - yes

Mentone – 708 – yes - yes
Mentone – 811/812 – yes - yes
Mentone – 825 – no – yes
Mentone – 903 – yes - no

Parkdale – 708 – yes – yes

Mordialloc – 705 - n/a - n/a (peak service only)
Mordialloc – 708 – yes – yes
Mordialloc – 709 – yes – yes
Mordialloc – 903 – yes – no

Aspendale -

Edithvale – 858 – yes – yes
Edithvale – 902 – yes – no

Chelsea – 857 – yes – yes
Chelsea – 858 – yes – yes
Chelsea – 902 – yes – no

Bonbeach –

Carrum – 708 – yes – yes
Carrum – 780 – yes - yes
Carrum – 857 – yes – yes

Seaford – 780 – yes – yes

Kananook – 778 – yes – yes
Kananook – 779 – yes – yes
Kananook – 832 – yes – yes
Kananook – 901 – yes – no

Frankston – 770 – no – no
Frankston – 771 – no - yes
Frankston – 772 – yes - yes
Frankston – 773 – yes - yes
Frankston – 774 – yes - yes
Frankston – 775 – no - no
Frankston – 776 – no - no
Frankston – 779/780 – yes – yes
Frankston – 781/784/785 – no – yes
Frankston – 782/783 – yes – yes
Frankston – 788 – yes - no
Frankston – 789/790/791 – no - yes
Frankston – 832/833 – yes - yes
Frankston – 901 – yes – no

Total before: Yes 39 – No 13
Total after: Yes 39 – No 13

Notes: Routes with irregular headways (eg 703, 775, 776) are counted as non-harmonised. Only routes with off-peak weekday services are counted.

Analysis

While the move to a ten minute frequency has not changed the number of bus routes that can claim to be harmonised with Frankston line trains, to conclude that there has been no change oversimplifies the effect on individual routes.

The most significant harmonisation gain was that restored to 20 minute frequency bus routes. I say ‘restored’ because they were previously harmonised when Frankston interpeak trains ran every 20 minutes (before the 1990s upgrade to 15 minutes). Some of these routes (eg 630 and 824) were the busier services that escaped the bus service reductions in the 1980s and 1990s (which saw some weekday frequencies reduced from 20 to 30 minutes). The other harmonisation gain has been in the Frankston area, where routes to Karingal (789/790/791) and Mornington (781/784/785) now combine to provide an even connection to every second train.

The main loss to harmonisation has been to the SmartBus and near-SmartBus routes (600/922/923, 900, 901, 902 and 903). Their 15 minute frequency matched the previous interpeak train frequency on the Frankston line. The effective frequency (ie the interval between when optimum connections repeat) has increased from 15 to 30 minutes for trips involving a transfer. However many passengers are likely to treat this as a ‘turn up and go’ connection, especially if changing from the bus to the train.

Metro’s Andrew Lezala favours the introduction of ten minute interpeak train service frequency on more lines, along with similar frequencies on tram and major bus routes. The latter obviously involves the orbital SmartBus routes, which currently run every fifteen minutes.

This pattern provides a legible ‘one-size fits all’ pattern where almost every interpeak orbital service runs the full route, despite greatly varying patronage along it. However if bus patronage continues to rise and more train lines are upgraded to run every 10 minutes interpeak, a future change to a 10/20 minute frequency pattern (lower service on the quieter portions) could provide needed gains in capacity and connectivity. Efficiency would also rise as services more closely match demand. There would be some decline in legibility, as alternate services would terminate short, but this may be outweighed by the consistent connectivity achieved by matching bus and train frequencies and the higher service on the busier portions.

Conclusion

While the numbers show no change to the proportion of bus routes that are frequency harmonised with trains during the weekday interpeak, the higher train frequency has improved overally connectivity by reducing maximum waiting times. It is for this reason that the gain of a 20 minute headway service becoming harmonised is greater than a 15 minute frequency service losing it.

Nevertheless to enhance their potential as strong feeder services, the principle that SmartBus services should be headway-harmonised with trains is worth keeping, and an approach to enhance this for our orbital routes is suggested.

DART’s first day

Without the speeches, balloons and sausage sizzles that normally accompany a railway extension, today marked Day One of DART, or Doncaster Area Rapid Transit.

DART is a four-route SmartBus network serving Manningham (Routes 905, 906, 907 & 908). It’s the government’s response to calls for a railway, which would have been dearer to build and directly served a smaller catchment area.

Poster in bus

Demographics

Manningham is predominantly middle to upper class. There is a high representation of professional two-income families. Housing is predominantely detached with parking for two cars. Nevertheless new townhouses are also popular in the areas nearer the CBD (being larger than a flat or villa in the nearby blue-chip suburbs of Kew and Balwyn).

Like other north-eastern suburbs Manningham lacks a large body of low income earners. These mostly settle in other parts of Melbourne which are better served by cheap rental properties and (up to now) better public transport. It does however have a substantial immigrant population (particularly from Asia) who tend to share (if not exceed) the middle class incomes and educational aspirations of the general population.

Manningham has no substantial industrial base, with most local employment being retail and services. There is also significant commuting towards the CBD, made possible by a freeway and express buses.

It also has no university campuses, though education levels and aspirations are high. This would tend to indicate significant commuting for education puposes.

Like many other parts of Melbourne, the baby boomers who settled Manningham are ageing. However average life expectancies and health standards are above average, so the majority are likely to be active and mobile well in to their seventies and eighties.

Doncaster Park & Ride

History

In the late 19th century a short-lived tramway ran between Box Hill and Doncaster, and the road between them retains that name. Market gardening was a dominant land use and the area did not become heavily urbanised until after WWII. However comparative closeness to the CBD, proximity to the Yarra, elevated views and the rise of the motor car (which made the lack of a railway less of a hindrance) contributed to Manningham’s growth in the 1960s and 1970s. It was part of a large belt of emergnig eastern suburbs between the railway lines and beyond the tram tracks including suburbs such as Forest Hill, Glen Waverley and Wheelers Hill. Traces of Templestowe’s rural past are evident today in its acre lots and lack of kerbing on some roads.

The Doncaster railway was part of the 1969 Metropolitan Transport Plan, which set the agenda for road building over the ensuing 40 years. Recent rail extensions have all been electrifications of existing rail corridors rather than new lines. And where new lines are proposed, they are either for growth corridor outer suburbs or underneath dense city land uses. And so all of Manningham’s public transport task is handled by buses, operating either to the city via the freeway, local shopping centres or railway stations in adjacent corridors.

Before SmartBus and DART, major bus routes in Manningham typically operated every 30 minutes on weekdays and roughly hourly on weekends (though some routes had two hourly or no service). This network was augmented by peak hour freeway express services and a fairly new circular shopper route.

DART represents a doubling of service frequency and later finishes on major routes. The four DART routes and two orbital routes (which started earlier in 2010) offer a SmartBus level of service (15 minutes weekdays/30 minutes evenings and weekends). This frequency is similar to suburban trains on weekdays and less on weekends. DART spans are broadly similar to trains except for the early finish on Sunday. The upgraded routes also had their numbers changed (from 300 series to 90x series) to reflect SmartBus route numbering.

Some existing routes were changed to reflect DART’s coming, but others were not. Hence there are some overlaps between regular routes and the upgraded SmartBus services.

City terminus near Southern Cross Station

The DART Test

The introduction of DART (and intersecting orbital routes) present opportunities for examining what happens when you (effectively) double transit service, especially in an area that lacks some of the demographics conducive to usage and starts with some handicaps for this.

Manningham’s limited public transport, low population density (especially in Templestowe) and high car ownership has historically resulted in a modal share lower than suburbs served by rail.

Demographics such as high workforce participation and lack of a local university tend to lower average occupancies in between the peaks. This produces a commuter pattern more like the Sandringham line (quiet interpeak) than the Dandenong or Sydenham line (busy day and night).

Secondly, unless regional centres such as around Shoppingtown and The Pines are built up patronage is likely to be unidirectional, lessening efficiencies, even if services in the peak direction are well used. This is different to (say) an orbital or cross-suburban bus between two railway lines which will attract patronage in both directions at any time.

My hunch is that DART will bring a substantial patronage increase, as it deserves to do so. However due to less favourable demographics (ie wealthier with more choice), I suspect that patronage elasticity per service kilometre added won’t be quite as high as we've seen on the three orbital routes and certain local upgrades.

Doncaster area frequent service map

The Yellow Orbital's first day

The first trip

Today I was part of a small group who rode the maiden full-length journey of Route 901 Yellow Orbital SmartBus. We took NightRider and Skybus to reach the airport by 0600.

Our aim to catch the first bus at 0630 nearly didn't happen since the bus did not pass the location indicated as its stop by a temporary sign. Luckily one of our group had the foresight to wait elsewhere and could hail the driver who stopped around the corner so the rest could board. This appeared to be an early glitch; subsequent trips were observed by others to use the marked stop.

Not suprisingly we had the bus to ourselves for the first hour or so; it was still too early for most Sunday engagements and many facilities along the way were under construction. Patronage increased greatly on the established Ringwood to Frankston portion, with the section south of Dandenong carrying a fully seated load nearly to Frankston.

The driver paced his speed well and the bus ran to time for the entire route. There was a short recovery time at most timepoints, but this should lessen once patronage builds and extra time due to ticket sales is factored in (travel is free for the first two weeks).

Along the route

Scenery varied between light industrial (most of the route until Epping and around Dandenong), bulky goods retail (around Nunawading) to residential (around Gladstone Park, Epping, Templestowe, Knox and Frankston North). There were also forested and semi-rural pockets around Yan Yean, which by any standard now has an extremely generous bus service for its population density.

Patronage potential

The existing Ringwood - Dandenong - Frankston section of Route 901 is now established as a high patronage section. It's everything a successful bus route should be - a direct line between trip generators of regional and even metropolitan significance.

Though its environment is pedestrian hostile, the sheer size of retail trade (and workforce) along Nunawading's 'Golden mile' along Whitehorse Road should assure reasonable patronage along this section as well.

My guess is that the north-eastern portion of the route won't be so popular; its feeder role to Blackburn Station will be somewhat lessened by the DART routes which will soon offer SmartBus-standard freeway express city services to much of Manningham. It also passes through areas of suburban (but low) population density at Templestowe and rural levels of density around Yan Yean.

South Morang to Epping should be moderately used as a rail feeder, though this task will be performed by Route 571 which will operate until suburban rail reaches South Morang.

There is some industrial jobs between Epping and Roxburgh Park (especially after the Market opens) that should attract some patronage. It is also from Epping (or even Greensborough) that 901 becomes a faster service to the airport than catching a train to the city plus a Skybus.

Route 901 duplicates local routes between Roxburgh Park and Broadmeadows and Route 902 between Broadmeadows and Gladstone Park. Hence in under a year Gladstone Park has gone from having very low to high levels of transit service.

I would expect the airport portion to be extremely successful - provided that the stop is in a handy location (for both travellers and airport workers) and people know about the service. The airport precinct is a huge employment area and the route will be a big lift for the area, which until today had extremely low service levels.

Photos taken on the first trip

Promotional billboard on entrance to the airport

The old stop for the regular routes (478/479/500)

The new (temporary) bus stop sign

Broadmeadows

Roxburgh Park

Greensborough

The Pines Shopping Centre (Doncaster East)

Finishing touches at Blackburn

Arriving Frankston

Timetable at Frankston

One passenger who won't be needing a service to the airport

Departing Frankston for Melbourne Airport

Metro's Caulfield - Camberwell shuttle: the bigger picture - Part II

Yesterday I mentioned the Metro proposal for a Caulfield - Camberwell bus shuttle between major rail hubs. I felt that it filled a gap in the network. And given the high patronage of similar north-south routes further east (903, 703, 902) even in car-dominated postwar suburbia, its chance of success would be high.

A 72 tram extension might be desirable long-term, but a lot can be done with buses. And there have been recent examples of strengtened shuttles between major trip generators, especially universities.

Exisisting shuttles

The earliest was Route 401, a new route running every 3 to 6 minutes between North Melbourne Station and Melbourne University. This was followed by additional services between Huntingdale Station and Monash University, firstly the 900 SmartBus and then additional short services on Route 630. Deakin University was next, with the extended 281 and new 768 routes supplemeting the existing well-used Route 767.

The extra capacity of these routes has been welcome. In some cases the use of multiple stops and unharmonised timetables do not necessarily exploit combined frequencies and patronage potential. Also there are risks of 'grafting on' inefficiencies when a new route is added without reviewing others nearby.

Nevertheless the principle of a frequent shuttle between major nodes has been tried, politically accepted and found successful. If more such routes were to be added, Caulfield to the Camberwell (or Glenferrie) appears to be a prime candidate, especially given that beneficiaries would be spread far and wide, as far out as Dandenong or Frankston.

A dedicated high-frequency limited stop shuttle could well work, just like 401 did. Especially if marketed well which is helped by good service design (eg a Route 401 every three minutes is easier to sell than three different routes each every 15-20 minutes serving different stops).

New route or enhanced existing?

However the proposal for a stand-alone route would be dearer than trying to form something out of the existing local network. Such a low-cost approach could be a useful early test of the shuttle concept. Enhancements such as additional services and express running could be added later if successful.

What to do with the 624?

The centrepiece of this thinking is Route 624, currently a complex route serving destinations as diverse as Chadstone, Oakleigh, Caulfield, Auburn and Kew, with a split in the middle. Its Auburn station stop is within easy walking distance of the Swinburne campus, and only slightly further to Camberwell.

The Route 624 timetable shows considerable peak/off-peak variations. Off-peak the trip from Caulfield to Auburn takes 15 minutes. During peak this can extend to over 30 minutes. Hence this route saves most time for off-peak travel (which universities can attract).

Current service frequency is 30 minutes, ie too low to be a reliable train connector in a busy and dense area. The aim will be to try to increase this to 15 minutes, though as Route 401 has shown higher would be even more attractive.

It is here that some shears are needed. Like a garden a bus network needs periodic tending, and undergrowth can get in the way of or inhibit the growth of strong branches. And Route 624 is an excellent candidate for this since it tries to do too many things in too many areas.

The eastern part of the route, which takes 25 minutes, runs between Oakleigh and Chadstone. It serves a low income postwar housing commission area as its local service. This part should be kept, with two buses maintaining the current frequency.

The central part (Chadstone - Caulfield) comprises of a frayed section with differnet sections of the route diverging and meeting again. Weekday buses alternate between the Carnegie and East Malvern sections, though on weekends and evenings only the Carnegie portion is served. Travel time between Caulfield and Chadstone ranges from 15 to 25 minutes.

The Murrumbeena/Carnegie portion of Route 624 shouldn't be too hard to delete. All that is needed is switching Route 623 from Dandenong Road to Neerim Road (as recommended in the recent local bus review). A rerouted 623 would serve existing Route 624 stops at doubled weekday frequency. Dandenong Road remains served by the Route 900 SmartBus, a service that did not exist when the current local network was planned.

Dealing with the East Malvern portion of the 624 is harder. Most parts are within the 800 metre pedshed of railway stations and tram routes. However two some parts are not and changes to adjacent routes would be needed to preserve coverage if this portion of 624 is deleted.

One possibility is to deviate Route 612 via Waverley Rd and part of Darling Rd to provide some substitute coverage (and a new link between the No 3 tram and Chadstone). This introduces a kink in an already circuitous local route. However the 'greater good' test is worth applying; if the change allows a more frequent high-patronage route to run nearby the overall benefit may indicate it is still worth doing. And in any event Route 612 has doubled frequency (30 minutes) and Saturday running so the trade-off would probably be accepted by residents.

Deleting 624 east of Caulfield Station would also mean no service on Burke and Wattletree Roads. A possible substitute, recommended in the Bus Review, was to extend Route 734 southwards from Glen Iris to Caulfield. This would generate patronage on the 734 (Caulfield being a stronger trip generator than Glen Iris), provide a handy link to the Alamein line and provide more frequent 7-day buses in this part of East Malvern. As 734's running time is not much less than 30 minutes for its 30 minute off-peak frequency, at least one if not two more buses would be required.

Options for an enhanced 624 between Caulfield and Auburn/Camberwell

This leaves the remaining western portion of 624, the boosting of which was the main object of the above trimming. As mentioned before, off-peak travel time between Caulfield and Auburn is 15 minutes, a figure that allows zero layover and driver break time if a 30 minute service is run between the existing trips (providing a combined 15 minute schedule to Auburn). Hence if the above pruning has released enough resources to run short services between Caulfield and Auburn it will be very tight, and not be conducive to reliability. The fact that the main route extends to Kew adds further challenges compared to if all buses terminated at Auburn.

A possibility could be to truncate all Route 624s at Camberwell instead of Kew. This would provide the requested Caulfield to Camberwell link but at the cost of poorer access to Swinburne University. However it isolates the Camberwell - Caulfield route, which could present operational and marketing advantages. If run via Tooronga Rd (instead of Burke Rd) the route would be slightly less direct but cheaper to implement as it retains 624's current coverage, so requires no substitute route to be funded.

There is then the problem with what to do with the remaining portion of the 624 to Kew if it was terminated at Camberwell. Again our friend the 612 could come to the rescue. If the 612 was split at Camberwell the portion from Chadstone could be extended to Kew via (say) Camberwell Rd, Auburn Road and the current 624 route. The benefit of this is that Kew residents get a direct bus to the major centre of Camberwell rather than just Auburn.

Like all ideas there are trade-offs and in this case these include forced transfers at Camberwell for 612 passengers (including many students attending schools around Canterbury) and the extra running time caused by the diverting routes through Camberwell (instead of the current straight running via Auburn).

Straight or bent grid network?

The latter raises a valid question about what our network should look like. A pure grid would treat all stations as equal interchange points. However in practice some nodes (eg Camberwell or Glenferrie) are bigger than others (eg Auburn) and even have webs of their own. Hence there is a temptation to use the gravitational pull of these larger centres to pull routes in, much like you see around Box Hill or Bentleigh (buses don't serve Laburnum or Patterson stations for instance). The benefit of this is reduced transfers for those going to the larger centres, but at the cost of slighly less directness and some increased running times for other passengers.

Conclusion

As many questions as answers have been raised as any option chosen will have disadvantages. However as the bus reviews have found our current network is not necessarily optimum. Network thinking along the above lines, although applied to a small area in response to an operator's suggestion for a new route, could be helpful in trying to design a network more attractive to passenges.

Metro's Caulfield - Camberwell shuttle: the bigger picture

Today's Herald Sun carries an article about a proposal from Metro Trains for a north-south bus route between Caulfield and Camberwell.

Metro's motivation may well be to relieve loading on two of Melbourne's busiest rail corridors. Nevertheless from a total network perspective it has a lot going for it, but only if planned in conjunction with existing nearby routes, which have just been reviewed.

Maps

View Larger Map

Also see Metlink local area map for local network

Notes about the area and transport services

- Approximately 10km from Melbourne CBD.
- Major universities adjacent to Caulfield (Monash) and Glenferrie (Swinburne) stations.
- A high international student population (who are heavy users of public transport - day and night).
- A major activity centre at Camberwell. Caulfield has similar potential if its underutilised racecourse is redeveloped.
- Both Caulfield and Camberwell are major junctions on Melbourne's busiest railway lines. All passing trains stop at these stations. Trains run approximately every 3 minutes during weekday peak, 7.5 to 15 minutes during off-peak and 15 to 30 minutes evening.
- No strong north-south public transport links for the six kilometre stretch between Glenferrie Rd and Warrigal Rd in the area south of the Glen Waverley railway (trains and trams are all east-west). In contrast the suburbs to the east (around Warrigal, Blackburn and Springvale Roads) have partial or full north-south SmartBus routes that are amongst Melbourne's busiest.

Existing services

Current options for travel between Caulfield and Glenferrie/Camberwell include:

- Train via a transfer at Richmond. Involves some backtracking but may be faster than the direct options that involve tram or infrequent bus.
- Route 72 tram between Toorak, Gardiner and Camberwell. Also involves some backtracking if coming from Caulfield as the line turns west instead of continuing south. Only a minority of daytime trains on the Caulfield group stop at Toorak, forcing an extra transfer at Caulfield.
- Route 16 tram between Malvern, Kooyong and Glenferrie. Provides a direct link to Swinburne University but can be slow. Not all trains stop at Malvern.
- Route 624 bus between Caulfield, Tooronga and Auburn. Serves the quieter station in between Glenferrie and Camberwell which is not served by all trains. The route is direct but its 30 minute frequency is less attractive than train or tram.

Overall

A Caulfield - Camberwell bus route that is more attractive than existing options could make a lot of sense. It would fill a major gap and be at least as well used as portions of some SmartBus routes.

Strong circumfrencial bus links can ensure that those who don't need to enter the CBD are not forced to do so. This frees up space for those who must (eg instead of staying on until Richmond, passengers at Caulfield may opt to change to a bus to Swinburne or Camberwell instead). And it's good economics too; the marginal costs of carrying passengers increase in congested areas, so frequent cross-suburban buses could pay for themselves in rail infrastructure not needed, and attract patronage in their own right.

My only proviso is that if implemented the route would not become an example of grafted-on planning which is inefficient and detracts from the capacity to fund local bus review recommendations. A way to avoid this is not to evaluate the proposal in isolation, but to considering adjacent existing bus routes (eg 624), especially if these could be modified to more economically form the new service.

Flinders Street: Melbourne’s rail hub

Flinders Street is the core of the Mebourne suburban train network. All trains, whether direct or via the City Loop go there, except for some suburban off-peak shuttles. Driver shift changeovers also occur at Flinders Street, which is why you often see many drivers there awaiting the next shift.

Trains towards Flinders Street are termed ‘up’ while all those from it are termed ‘down’. Train identifiers, or TD numbers, used for internal purposes, change to form the next service at Flinders Street.

This next service could be either (i) the train returning to its origin group of lines via the City Loop, (ii) the train returning to its origin group of lines having previously operated via the City Loop, (iii) the train continuing as a through service to the other side of the city via Southern Cross and North Melbourne, or (iv) the use of Flinders Street as a stub terminus, where drivers change ends and trains return to the origin line.

Of these the first two are most common. Through-running is less widespread and less publicised, frequently being subject to change. Compared to through-running, stub operation is less economical of track and platform space so is confined to the less intensively served lines (eg Sandringham, with a 9 minute peak frequency).

It is also at Flinders Street where decisions are made to alter train patterns, so that up trains form a different down train to that specified in the working timetable. These short-notice alterations, or transposals, assist recovery from late running and are extremely common. Less commonly, late trains may be altered to operate direct rather than via the loop, providing Flinders Street with a somewhat more reliable service than loop stations.

However changing drivers and transposing trains at at Flinders Street can cause difficulties for staff and passenger information. There is often uncertainty as to what service an incoming train will form until the last minute. And display screens do not have real-time information, unlike the ‘minutes to’ indicators at other stations.

Information in walkways

Platform information displays at Flinders St

The above images indicate the absence of real-time information at Flinders Street, despite its importance for efficient mobility on a complex system prone to delays. Without information passengers cannot make the best choice of which platform to change to. Passengers for stations served by trains on multiple platforms might go ‘platform shopping’, unnecessarily clutter walkways and slow those alighting. Conflicting uneccessary pedestrian movements (eg to/from platforms 4 & 5 on the diagram below) can reduce people throughput and the station's efficiency.

In contrast passengers armed with real-time information when they alight can proceed directly to the required platform (sometimes but not always indicated by announcements), without unnecessary deviations. This improves walkway throughput and platform efficiency. These are both important considerations given increased pedestrian flows (from a busier network) and revised timetables (that promise more through services or rely more heavily on passengers changing trains). For similar reasons the larger information displays need to be plainly visible from (but not directly in the line of) busy walkways (where lookers block those who know where they’re going), and a ‘keep left’ rule, encouraged by unidirectional ticket barriers, could be encouraged.

It was mentioned earlier that trains are frequently transposed. Some transposals do not affect passengers. But where city-bound trains are listed in the timetable as through services but are cut short transposals matter. On other occasions trains that have served loop stations before arriving at Flinders Street may have their destinations changed from that timetabled. Whose job is it to inform the passenger of these changes?

If the driver was continuing straight through, it would be reasonable to expect them to provide an announcement. However as mentioned above, Flinders Street is the major point where drivers change over. The driver alighting at Flinders Street is only concerned about his next service and may not know the fate of the train he has just left. Conversely his replacement is only concerned with where the train is going, not from whence it came or where the timetable states the train should go.

Platform staff have little information beyond printed timetables and even if they did this may not help those already on the train. Hence information is often limited and the full benefit of though services is not always realised due to the high chance that they are terminated short (and other services – typically late running but earlier scheduled trains - are extended instead).

Could altered driver rostering change this? Changed procedures so that more drivers stayed with the same train as it entered, stopped and left Flinders Street Station would provide a continuity often currently missing. If backed by announcements as to what the train will form before it stops at Flinders Street this could provide improved information for passengers.

However moving driver changovers from Flinders Street to suburban stations, although desirable for short dwell times and information continuity, poses other challenges. Transposals will still be required, and these are likely to be easier to arrange if drivers were pooled in the same location that decisions are made (rather than dispersed across several suburban stations).

A possible solution is to schedule trains so that through-services constitute the regular operating pattern. Even if trains continued to be transposed (as they will be) transpositions would affect fewer passengers. Eg if a late-running Frankston train formed a later Laverton (or Williamstown) service rather than its scheduled Werribee train, the bulk of passengers making cross-city trips (eg South Yarra to Footscray or Caulfield to North Melbourne) would be unaffected. This is because few passengers on the Frankston line would be travelling further than Newport, but a fair number would want Southern Cross, North Melbourne or Footscray.

This arrangement may come about if Victorian Transport Plan proposals to link the Frankston and Werribee lines via Flinders Street and Southern Cross come about. While potentially controversial, this operating pattern would need to apply at all times, not just peak periods.

In the interim, greater attention to information and announcements could make passenger throughput more efficient, improve legibility and speed end-to-end travel. When patronage was static or falling these sorts of matters tended to be pushed into the background as unimportant. However with different operating patterns, more trains, higher patronage and crowded walkways, the ‘scientific management’ of pedestrian flow, based on appropriate information given at the point of need, is going to have to be rediscovered to maximise passenger throughput and speed travel.