Thursday, April 13, 2006

Train bingo and timed transfers

I don't know if there is an official term for when up and down trains happen to pass one another at a station. However these scheduled coincidences interest me for reasons that will become obvious later.

Coincidences

The general rule is that coincidences occur at stations spaced half the service headway along the line. In other words, if trains run every 30 minutes and up and down trains have just crossed at your station, the next crossing points will be at stations 15, 30, 45, etc minutes down the line from you.

Looking at the Pakenham/Cranbourne evening timetable, we find trains cross near the following stations:

Hawksburn: ups arrive :16 & :46, downs arrive :15 & :45
Oakleigh: ups arrive :01 & :31, downs arrive :02 & :32
Noble Park: ups arrive :17, :47, downs arrive :17 & :47

On the Frankston line we have:

Hawksburn: ups arrive :16 & :46, downs arrive :00 & :45
Bentleigh: ups arrive :17 & :47, downs arrive :16 & :46
Parkdale: ups arrive :02 & :32, downs arrive :00 & :30
Seaford: ups arrive :17 & :47, downs arrive :15 & :45

In addition Frankston ups meet Dandenong downs at Caulfield and vice versa.

I haven't looked at the other lines except to note that there are passings between Blackburn and Nunawading (:13-16 & :43-46).

Application

What use is this, apart from being a curiousity for gunzels?

The answer is that stations where both up and down trains arrive within a minute or two of one another present great opportunities for bus connections. And not just for one or two trips, but for all. I have identified two main approaches to transfers. One optimises transfer speed and the other optimises service reliability.

Timed transfers: 'shortest wait' method

With the existing evening train timetable at Oakleigh, a bus pulling in at :29 (:59) and departing at :34 (:04) is able to achieve the following transfers:

bus > up train
bus > down train
up train > bus
down train > bus

Let up assume that, like the existing route 700, there are both northbound and southbound buses via Oakleigh. Furthermore, suppose that we can schedule buses from both directions to arrive at :29 (:59) and depart at :34 (:04).

The number of transfer combinations increases to the following:

northbound bus > up train
northbound bus > down train
southbound bus > up train
southbound bus > down train
up train > northbound bus
up train > southbound bus
down train > northbound bus
down train > southbound bus

In other words, there isn't just one train connecting to one bus, making one possible connection, but instead there's eight.

Imagine what this would do for bus patronage. Even though some transfer combinations might only be used by a handful of passengers, the power of careful connection planning should now be apparent and worth the extra bus dwell time at Oakleigh. Also the time saved by transferring passengers (which can approach an hour in the example here) should outweigh the extra few minutes for through passengers.

Timed transfers: 'best reliability' method

Though it delivers the shortest interchange times and fastest travel speeds, the 'shortest wait' method may not be ideal in every case, even at interchanges where up and down trains arrive simultaneously.

The 'best reliability' method schedules buses halfway between trains. At Oakleigh, buses in both directions would arrive at :16 and :46 minutes past the hour, given the trains are h:01/:02 & h:31/:32. The above eight transfers would still be possible, but the waiting times would be extended from a snappy three minutes to half the service headway, ie fifteen minutes.

To be fair this is not without its advantages. These include (i) more robust connections even if services are delayed (a train that is 10 minutes late will still allow a 5 minute connection), (ii) permit easy transfers by mobility-impaired passengers, (iii) not require buses to be held back for late-running trains, and (iv) avoid the need for buses to wait 5 minutes (or more) at Oakleigh, and so increase travel speeds for through passengers, assist on-time running and even aid bus utilisation.

Because of the desirability to keep waiting times to 10 minutes or less, the 'best reliability' method is suited to times when train headways are 15 or 20 minutes and/or there are large numbers of through passengers on the bus you don't wish to delay.

An example of where 'best reliability' scheduling would be ideal is weekend services at Oakleigh. As with evenings, the up and down trains coincide, but due to the 20 minute service, the times are h:13/:14, h:33/:34 and h:53/:54. The 'best reliabilty' method would have buses arriving at h:03, :23 and :43. This gives a constant 10 minute transfer time, which is acceptable though not 'seamless'. However the shorter dwell time at Oakleigh benefits through passengers and makes connections more robust.

Conclusion

Both timed transfer methods have their place. Fortunately both can be used on the one route at different stations or even at different times. Their common point is equal treatment of trips in all directions; there is no attempt to 'second guess' complex travel patterns. This starting point is more in accord with modern transport (and driving) patterns than old-style planning which relegated public transport to a CBD-based radial function.

The topics raised here are the sort of things that intelligent service design should be all about. To date we've seen less genuine service planning in Melbourne than in other cities like Perth, where innovations such as service co-ordination, timed transfers and headway hierachies are everyday realities. Given the benefits obtainable, the time for it to be done here is ripe.

Labels: , , ,

0 Comments:

Post a Comment

<< Home