Service costs and spreading the peak - 2

As well as fares and frequency, another way to spread the peak is to provide express services during shoulder peak. Generally expresses are even more 'peaky' than train frequency. This is good for inner-suburban passengers (as it gives inner suburban stations a broad peak of frequent services) but less favourable for outer suburban commuters who have increased travel times relative to if they travelled during the height of the peak.

The following times indicate the first and last peak direction express trains passing through Caulfield (or Camberwell). 'Express service' is defined as one not stopping at five or more stations*.

AM Peak

Belgrave/Lilydale: 5:46am - 10:20am
Cranbourne/Pakenham: 6:12am - 8:36am
Frankston: 7:43am - 8:47am

PM Peak

Belgrave/Lilydale: 3:40pm - 7:37pm
Cranbourne/Pakenham: 4:41pm - 6:06pm
Frankston: 5:16pm - 6:21pm

(*) Based on 30/9/07 timetables, which are out of date. Will be updated with new data.

The main limitation of the above spans is that they change according to where you draw the line on the number of stations the expresses miss. For example, there are many trains that express through three stations. This will broaden the span - to most of the day in the extreme case of the Frankston line where it becomes meaningless.

Because it's better to spread the peak and this requires both service and behavioural changes, I opted only to include the faster express trains. This is in line with passenger behaviour - passengers able to will prefer the faster expresses, even if the travel time saving is small. Shoulder travellers on some lines pay a penalty in reduced service frequencies and/or fewer expresses, and tackling this is helping to spread the peak and increase efficiency.

What do the above lists show? The Belgrave/Lilydale lines have wide peak express spans, with express services running well into the shoulders. The Frankston line has very narrow peak express windows (literally 'peak hour'), although it has many lesser expresses outside those times and during the day. Cranbourne/Pakenham has expresses opearating during most peak times. The most conspicuous gaps where there are no or slow expresses are on the Frankston line before 5pm and on both the Cranbourne/Pakenham and Frankston lines from about 8:30am. These early finishes do not encourage 9:30am starting, though if this was encouraged the after 6pm services also need attention.

The narrow express peaks on the Frankston line warrant further attention. Planning for it is complicated by the interplay with the other Caulfield group lines. Service priorities are elsewhere and nowhere is the competition fiercer but on the Caulfield group. The number of Frankston expresses appear to be held back by a need to provide a service to inner suburban stations (Malvern - Hawksburn) that cannot be satisfied by existing Carrum, Mordialloc, Cheltenham, Westall and Oakleigh local trains alone.

Current arrangements have the advantage of relieving the burden on the Cranbourne/Pakenham line (which has heavier patronage but less infrastructure). However Frankston is also a long line with its own distinct needs. These are a wider selection of expresses to serve areas beyond about Moorabbin and more frequent service closer in.

This makes the concept of a two-tiered service based on frequent local and express trains running over a longer span attractive with better use of the existing third track to Moorabbin. However it could also mean that some zone 1 stations (eg Glenhuntly or Carnegie) may lose express trains, and more services would miss the loop, running direct to Flinders Street.

To summarise, spreading the peak by making shoulder peak service more attractive is a viable way to improve capacity and network efficiency. While it can't cope with meeting long-term requirements, it's a worthwhile short and medium term strategy to relieve pressure on the network and has an exceptional return per dollar spent.

Service costs and spreading the peak - 1

What gives best 'bang for the buck' when it's decided to use service levels as a lever to boost patronage or change patronage patterns (eg spreading the peak)?

Whether you look at it from the customer side (better services) or the supply side (costs and resources), the answer is rather 'lumpy' and not linear.

As an example, adding an extra three trains per hour to a subway service that's already running every three minutes isn't going to make the service that much better, assuming the existing system is already reliable and is not overcrowded. On the other hand, using them to increase an hourly service to every fifteen minutes is a major boost and makes the service much more attractive.

On the supply side, the costs in boosting services escalate as follows:

Stage 1: Existing infrastructure sufficient, existing rolling stock sufficient, need to train/roster more drivers.

Stage 2: Existing infrastructure sufficient, need more rolling stock, need to train/roster more drivers.

Stage 3: Need more, infrastructure, need more rolling stock, need to train/roster more drivers.

Stage 1 is the fastest and cheapest, while Stages 2 and 3 are the dearest and take longest.

Additional peak capacity is likely to require Stage 2 or 3. In contrast, major improvements to off-peak services only require Stage 1. This explains why the marginal costs of increasing off-peak patronage is lower than increasing peak riders. If some of the latter can be induced to shift their travel times to shoulder peak periods, total patronage grows and the large fixed costs of rail are spread across more passengers, increasing efficiency.

Against this is a question as to transport's role in society. Should transport adapt itself to people's lifestyles, or should we expect people to change their living according to what suits the transport system? The former will obviously cost more than the latter, since more Stage 2 and 3 improvements will be needed to cater for a sharp peak. However public transport economics can't afford to impose itself too heavily on passengers, since more will find driving more attractive, train patronage will stagnate and modal share will fall.

To date the main measures that have been taken to spread the peak include:

1. Local bus improvements. Many routes now finish at 9pm instead of 7pm. This may lengthen the evening peak, although there is still a frequency penalty for those leaving work after about 6pm.

2. An 'early bird' ticket. A special ticket providing free travel for train travellers (only) who reach their destination by 7am. Intended to spread the morning peak.

3. Some increases in shoulder period train services.

The following graphs compare current service provision across the AM and PM peaks on several popular lines. For those not from Melbourne, the lines can roughly be described as follows: Belgrave/Lilydale: 3 tracks for better peak expresses, high service levels and low catchment population growth; Cranbourne/Pakenham: 2 tracks, moderate to high service levels and high catchment population growth; Frankston; 3 tracks for better peak expresses, moderate service levels and moderate catchment population growth. All are at least 40km long and have stations every 1 to 3 kilometres. Suburban services on the latter two lines also have to share with goods and/or country trains.

All trains that pass through the nominated points are counted; even when they do not stop. Using a moving time period (rather than fixed 30 minute blocks) would be better, but is not needed for broad comparisons.

Apart from the intensive services on the Belgrave/Lilydale line, the major distinguishing feature is its extended morning peak. This compares to the Dandenong line (especially) where service falls to off-peak levels by 9am. In this regard the Frankston line does better, though its frequency is less during the 'early bird' period.

During the evening, the broad peak on the Belgrave/Lilydale line is again apparent. Frankston again has the shortest peak, with the period of significantly enhanced service lasting for about two hours. Possibly due to its third track and relaxed capacity constraints the general pattern is that the Frankston line runs faster expresses (but fewer of them) than the Cranbourne/Pakenham lines, which has more stopping trains, but better frequency.

Of the lines surveyed, the Cranbourne/Pakenham lines have the least infrastructure, the most non-suburban trains to share with, the highest overall patronage and the greatest patronage growth. This causes its peak services to be under a degree of pressure unseen on the Belgrave/Lilydale line, as reflected in the reliability statistics. This makes it the most attractive candidate for peak spreading, particularly around 9am, 4pm and 7-9pm.

My Grandma owned a car

Transport academics and leaders talk about the future for Melbourne public transport (2 parts).

Discussion at the Future Melbourne wiki.