Monday, January 12, 2009

Melbourne Public Transport Standards Review

Booz & Co (who are also doing some bus reviews for DoT) have also been busy researching on service standards for Melbourne public transport overall.

Read their review here

My main observation was that though the report include a list of things that individually contribute to service (eg amount of rail infrastructure, span and frequency) it stopped short of trying to bring it all together to form an overall measure.

As a result reading the report provided a bunch of statistics that did not convey the user experience and the extent to which this varied between cities. Here's just a few questions one might ask:

* Is the public transport in my suburb good enough to go somewhere on a Sunday or stay out until 11pm?
* Can I rock up at a station or stop and expect a service in a few minutes in peak hour or should I instead plan the trip beforehand?
*Or is the train likely to be faster than driving, and if I need to catch a bus will the connection be good?
* What proportion of suburbs get a service every fifteen minutes or better?
* Are summer and public holiday service arrangements standard across the network or does each operator or route do its own thing?

These are the sorts of questions that a robust set of service standards (and study of same) would address. I have a rough idea of service levels and characteristics in some of the Australian cities mentioned. Melbourne trams good, Perth trains good, Adelaide buses good, Perth connectivity good, Brisbane trains poor off-peak, Melbourne's trains extensive, Sydney ticketing terrible etc. However I'm not sure if this report would convey this sense to the general reader (or policy-maker) so we know where Melbourne sits.

Another thing the report could usefully have done is to define an overall service standard for its client; especially since page 4 says service standards are 'largely undefined'!

The service standard method mentioned before on these pages involving coverage, frequency and span seems as good a starting point as any:

V % of the population within W minutes walk of a service that runs at least every X minutes between Yam and Zpm, seven days a week.

The figures to insert depend on what the policy maker wants the public transport network to do - ie compete with driving for many trips, or to provide a social welfare type service. For example:

* If you wanted to make it a high standard network intended to win modal share, V = 90%, W = 10 min, X = 10 minutes, Y = 6am and Z = midnight.

* If public transport is more a social service then W might be 5 minutes, X every 60 minutes and Z = 7pm.

* A network somewhere in between might have a substantial grid of primary routes where X = 15min, Y = 6am, Z=midnight supplemented by local secondary routes where X = 30 or 60 min, Y=6am and Z=9pm.

Of these three the last option could be achieved in Melbourne fairly quickly. The use of service standards to reduce the myriad variations in bus service levels to a three tier system of primary, secondary and tertiary routes with harmonised hours, frequencies and holiday arrangements would permit the sort of network legibility, information and marketing that other cities take for granted.

A good service comparison between cities might set a moderate to good level of service (such as an Adelaide Go-Zone) as a reference. The study would compare the proportion of the population within 10 minutes walk of that level of service. All Australian cities are likely to have only a minority of the population near a service of that quality, but it would be a substantial minority and comparisons would be readily made. In particular Melbourne's tram network would stand out as exceeding this level, while most of its buses (including some SmartBuses) would not exceed it. The latter would also be true of Perth for its buses, but it's train network would meet the standard, unlike Adelaide's or Brisbanes. And this sort of comparison would more closely align with the user experience mentioned above.

Robust service standards along the lines of that above can be very useful to a co-ordinating departmet. They would then flow into network design, service levels and then marketing. Adelaide excels here with it's 'Go Zone' buses where passengers are assured of a 15 min weekday/30 min evening and weekend frequency and wide operating spans and the service is marketed as a coherent network. And Brisbane has 'BUZ zones', offering services every 15 minutes until late at night. Instead of drawing from as many overseas cities, these concepts could have been proposed as practical, local and achieveable examples of effective service standards and marketing. Unfortunately, despite being raised in the body of the report, the need for genuine service standards did not make it into a conclusion, recommedation or summary of key points.

Now onto some specific points. All are fairly minor, but I can't resist mentioning them.

Page 4. Trams are ignored in the service comparison tables, presumably because other cities don't have anything matching it (although Sydney Buses could have been surveyed). In a survey this would tend to unfairly work against Melbourne bus frequencies since in other cities trams' purpose is performed by high-frequency bus routes (eg Adelaide's Go Zones').

Page 4. The report is correct in that 'Track Record' is not the place for train span and service frequency standards. However these are specified in the train operating contracts. Information from these could have been provided rather than the columns left blank. Again this make us look less favourable.

Page 7. The claim that Perth has a standard 15 minute train service might better have been qualified as only applying during the day (Monday - Sunday) with some minor exceptions. Perth's night trains run every 30 minutes - ie similar to Melbourne Monday - Saturday (Sunday some lines). Hence the report oversells Perth's night services, when they are better regarded as comparable to us.

Page 8. While Perth's 'late night' trains do run later than Melbourne, it should be noted that this is at the cost of frequency - every 60 minutes instead of ours every 30 minutes. A true comparison would also factor in trams and NightRider buses, which have seen substantial boosts lately.

Page 9. The 10-20 minute evening train frequencies quoted for Melbourne are unrepresentative since they only apply to stations between the city and Footscray, Clifton Hill, Ringwood, Caulfield or Sandringham. Claiming a 30 minute average would have led to more robust findings. Claiming a wider range of peak frequencies such as 3-20 minutes would also have helped since the quoted 5-10 minute pm peak headway is only seen at a minority of stations.

Page 10. No Australian rail system beats Perth's consistent 15 minute daytime 7-day off-peak frequency, but nevertheless the claim is made that Melbourne performs better off-peak. If we were to make such a claim for Melbourne, it would have to be on a geographic and multimode basis, with trams included.

No comments: