Wednesday, September 29, 2010

Headway harmonisation hits the news

Today's Age carries an article making the point that no one is in charge of Melbourne's public transport, especially when it comes to service co-ordination.

It contained several quotes on who is responsible for service co-ordination. The confusion evident only demonstrated that different parties have different ideas of what this means.

The calls for a transport planning agency have largely come about due to the gap between the integration that passengers see in fares and information and the disintegration experienced when timetables do not connect, nor are even planned to connect.

Although the Altona example given in the Age article was not a good one, the point made about the non-harmonisation of bus with train timetables is easily able to be tested.

Earlier this year I checked timetable harmonisation at selected bus/train interchanges in Melbourne and Perth.

The method for assessing harmonisation was as follows:

* If the train ran every 20 minutes and buses ran every 20, 40 or 60 minutes, this was counted as harmonised. However if the bus was every 15, 38 or 55 minutes this was not considered harmonised.

* Public holiday arrangements for buses can vary. Tthe figure given is the percentage that follow train holiday patterns (eg if train runs to a Saturday timetable the bus does as well).

* Routes planned as a combined service (eg 216/219 or 827/828) were counted as one.

* Only daytime off-peak services were examined.

* Of course the above does not guarantee connectivity; a bus every 15 minutes could consistently miss a train every 15 minutes by 14 minutes. However unless services are very frequent headway harmonisation is a pre-requisite for high connectivity. And if frequencies are harmonised it should be easier to optimise times for connectivity at the busiest locations.

The extent of bus/train headway harmonisation was found to be as follows:

Percentage of bus routes harmonised with trains by interchange (Mon-Fri/Sat/Sun/Pub Hol)

Perth - Bassendean: 67/33/100/100
Perth - Canning Bridge: 80/100/100/100
Perth - Murdoch: 100/100/100/100
Perth - Whitfords: 100/100/100/100

Melbourne - Cheltenham: 75/75/75/75
Melbourne - Clayton: 80/33/67/100
Melbourne - Craigieburn: 0/0/0/100
Melbourne - Cranbourne: 43/50/50/80
Melbourne - Dandenong: 78/86/80/56
Melbourne - Frankston: 64/67/70/77
Melbourne - Hoppers Crossing: 100/100/100/80
Melbourne - Huntingdale: 75/0/50/100
Melbourne - Pakenham: 100/0/0/80
Melbourne - Reservoir: 17/17/25/67
Melbourne - Sunshine: 30/20/40/80
Melbourne - Watergardens: 83/83/100/83

The above shows that bus/train frequency harmonisation is almost universal in Perth but variable in Melbourne. Similar comments apply for public holidays, with a network-wide standard existing in Perth but not in Melbourne, where holiday practices vary between bus operators and routes (despite substantial recent progress). Both reinforce public impressions of bus routes as complex and unreliable, contrary to the Minister's recent call to embrace bus travel.

The ability to achieve widespread connectivity in one city and patchy connectivity in another indicate that although we have authorities clearly responsible for policy and contract management (Department of Transport), ticketing (Transport Ticketing Authority), information and marketing (Metlink), the equally important function of service planning remains on the outer, despite its centrality to network usability.

My guess is that Stone's paper will conclude along similar lines. While the Age article doesn't mention this, the conference in Canberra is likely to be the Australian Transport Research Forum. If previous years are any guide, the paper presented today should later become available through the ATRF paper archive which is well worth a read for anyone with a transport interest.

4 comments:

Alan Davies said...

Peter, have you ever written a post on what the technical difficulties and challenges are in achieving coordination between modes?

Peter Parker said...

Alan - thanks for the idea for a post. Hope this answer will suffice for now, but it could be worth fleshing out in more detail later.

Although I would add that historical and organisational factors are a more important determinant of co-ordination than technical limitations in most cases.

The main exception where technical constraints are most severe is our rail network. These are the now well-known factors that limit service frequencies, reliability and recovery times when trains are delayed.

Prime examples would include single line sections, infrastructure and rolling stock reliability, delays due to loadings, etc.

High train frequencies lessen the complexity of planning buses to connect (and the consequences of a missed connection).

Our extensive (but sometimes low frequency) suburban rail network is somewhat harder to connect buses to than a less extensive but higher frequency network.

As an example, trains on the furthest fringes of our rail network (Pakenham @ 55km from the CBD) run every 30 or 40 minutes during the day. In Perth the frequency for Mandurah @ 70km out is 15 minutes daily - much easier to plan buses for.

Perth's trains are also frequency harmonised with one another - whereas due to historical factors (mainly some lines getting frequency upgrades while others did not) Melbourne's are often not.

High train reliability ensures that a co-ordinated network works as well as it is planned. The latter is a particular concern for the pm peak since outer suburban buses typically meet only every second or third train.

In addition to train reliability, information systems and procedures such as active connection management may also help reduce the consequences of delays (though anything to do with holding buses may also have consequences for bus-only passengers and other connections).

Leaving aside the complexities introduced by train late-running, it would seem that a reasonable first step would be to attain optimum connectivity (at least for popular interchange points and passenger flows) as against the timetables.

Luckily the flexibility of buses (including their ability to take multiple routes (given a suitably permeable road network) and pass one another) lessens the constraints compared to trains.

Peter Parker said...

Part 2 in reply to Alan...


Those constraints that remain are overwhelmingly not what I would call technical or as hard and fast as with trains and single tracks.

At one extreme very high bus frequencies without a lot of service planning would solve them but at a price.

Much cheaper would be a timed-transfer network able to deliver a reasonably well connected network for many trips at a lower cost (but note the problems caused by train delays, although it helps if the pm peak is more spread than the am peak, which is fortunately it is in the outer suburbs).

I mentioned organisational factors earlier. If bus routes are planned seperately to timetables it may be difficult to get an interplay between the two that favours connectivity.

For instance a bus route might take 39 minutes to run between two stations. Two buses might be able to provide a local service operating every 45 minutes. There's a bit of 'fat' for recovery time but the buses are moving most of the time, so on the face of it this looks like an efficient use of buses.

However if the buses are running in an area where the trains are every 30 minutes, so the best connections will repeat only every 90 minutes. Which suddenly makes a service that's efficient on paper quite ineffective when it comes to serving passengers.

A lift from a 45 to a 30 minute service (which would enable uniform connectivity and triple effective frequency for trips involving a transfer) would need more buses.

A drop to 60 minutes would improve connectivity for those catching the train and provide a memory timetable. But it's still a frequency drop for local travellers. And it wouldn't be very efficient operationally as the bus would be sitting around for 21 minutes out of every hour.

But supposing you could vary the route. Either making it shorter to 25 minutes, or longer to 55 minutes (maybe there's a nearby route that it could be amalgamated with). Then you'd get train-friendly 30 or 60 minute headways and good bus utilisation.

If there are stations at both ends of the route you still might not get good train connectivity at both in all directions, but it should be possible to optimise for peak flows. And the connections would be reliable, with an effective frequency equal to the bus frequency (every train or every second train).

Even the objection about the reduced frequency may be able to overcome if there's a similar nearby route overlapping a busy common section. Staggering times could allow two 60 minute routes to form a combined 30 minute service, potentially connecting with every train. This is something that two 45 minute routes would not be able to achieve.

I suspect that the seperation between route planning and timetable planning could be one of those factors preventing these sorts of co-ordination opportunities being exploited.

A challenge? Yes. But an insoluble technical difficulty? Probably not - since other places might not have particularly high service frequencies, but what they do have has been carefully optimised to connect.

Alan Davies said...

Thanks Peter, much appreciated. Thoughtful and insightful. Can I suggest you might want to put this up as a separate post so others can read it too (bit buried away here)?