Tuesday, November 3, 2015

Airport rail link study - analysis

So I read the official report of the Melbourne Airport Rail Link Study, which doesn't seem to be available any more but I can send you a copy if you like.

Well, I won't beat around the bush - it's not a good study. I started doing a page by page critique but in the end I found there were three main categories of failing:

  1. It suffers from being a study in isolation, rather than a part of a full scale integrated urban and transport plan. Of course we didn't have a full scale integrated plan back then (and still don't really) but it causes all sorts of problems for the study.
  2. It shows indecision about the purpose of the airport rail link - whether it's to be an express shuttle or a suburban stopping service
  3. The methodology is flawed with the result that some worthwhile options were not even considered.

So let's look at some examples. I'll try to avoid repetition because that's dead boring, but take it from me that these failings are liberally scattered throughout the report.

A study in isolation

So as I said up there (I could make you scroll up and read it again but I'm Mr Nice Guy today), the study was made in isolation - it examined options where a project started, delivered an airport rail link, and then finished. It couldn't examine any options where a project started, delivered all sorts of improvements over a number of years, and allowed an airport rail link to grow from the seeds it dropped by mistake.

Intuitively, an airport shuttle has high commonality with an intercity high speed rail link. They use similar technology, and in Melbourne they'd most likely run along a similar corridor. Unfortunately at the time this airport link study was released the Federal government's high speed rail study was incomplete. This meant that only basic tests could be run to see if the two were compatible, or to realise any economies of scale between the two. In an integrated long-term transport network study, each project exists as a stage of a rollout, rather than being designed in isolation. That would not only ensure compatibility between the two, but the studies into each project would be cheaper too.

Surprisingly, in this study, as the team went through the various options for an airport rail link they only rated integration with the PTV and Melbourne Airport plans as one of the criteria in the assessment. Some options were shortlisted even though they didn't fit with the other plans, which means they would need to be modified, at some cost.

There are several examples of options being rejected by the study due to capacity bottlenecks in the rail system today. For instance:

  • On p113 the study team found that "Melbourne Metro is expected to reach capacity at the time when MARL services are introduced", and therefore decided not to run airport services via the Metro. But lack of capacity could be solved by introducing technologies from overseas which allow frequencies of up to 48 trains per hour, using systems similar to our proposed High Capacity Signalling.
  • On p119 the study team were measuring running times via various routes, to see whether it was possible to meet the requirement for overall journey time to the airport. The team assumed current line speeds and timetables, even though these could be improved with track upgrades - but the upgrades were outside the scope of the study, which made certain routes unviable.
  • On p125 the study team came up against a critical lack of capacity at North Melbourne station, because airport trains would be added to the already overcrowded tracks used by Sydenham services, and more capacity would be required on the Craigieburn line as well. The station has space for two more platforms and the junction could be redesigned, which would benefit the entire system, but again this was outside the scope of the study.

In each of these cases it would be possible for the extra capacity to be added, but the cost of doing so would be out of proportion to the value to airport travellers. An integrated transport plan could recognize that other services on the system would be improved by the works, giving a benefit to passengers on non-airport services. These options couldn't be considered by a plan for an airport line in isolation.

Possibly the most telling example in this category was on p133. The effectiveness of various station locations was being examined, but the study had to eliminate both feeder buses and Park & Ride from consideration, because those two factors have almost infinite variability and would be impossible to study within the timeframe. But they can change the dynamics of a transport network dramatically, and affect the viability of a project. An integrated plan would, by definition, include all of that.

Indecision of purpose

The case for an airport line was defined by the study team in terms of air passengers (p10-11 and throughout the document). If travellers are the target market, the ideal service is an express stopping at major transport interchanges only. However, throughout the study there are indications that the thinking behind the plan is for a metro service. For example:

  • In the Functional Requirements section express running is described as "desirable" but stopping services are acceptable if the travel time requirement is met. Air travellers don't want to stop - particularly at stations where airport trains will be acting as a Metro service because there's nothing else available, as it would overcrowd the trains.
  • It was assumed that airport trains would run either the Melbourne Metro tunnel or the City Loop, and stop all stations through the suburban system (stated explicitly on p151 and p130 respectively but hinted at several times prior). This is consistent with a metro service but not with an express shuttle.
  • The orbital option, with airport trains running from Broadmeadows to Sunshine, was eliminated from consideration because it "will not align with future metropolitan rail strategic operational planning, due to them being in two different sectors of the future rail network". An express shuttle should be part of a separate network again, on separate tracks.
  • On p133-4 we are told Skybus will continue to operate after the rail line opens - this would be unnecessary if the rail service was an express shuttle.

But then on p133-4 we are told that the line will operate with flat rate fares, separate from the Met zone/fare structure, and at rates similar to Skybus tickets today. That will make the airport line unattractive as a suburban service.

An unstated assumption in many recent rail studies is that all rail lines are assumed to be equal - if there's rails there should be stations. There's no concept of express shuttle services (airport), regional services skipping suburban stops (RRL) or bypass lines serving existing stations only (Melbourne Metro). New stations are good for the people who live within the "pedshed", but it means that new rail lines only increase the catchment of the system - they don't reduce overcrowding, either of passengers on trains or trains on tracks.

Flawed methodology

The supposedly exhaustive examination of available corridors, with multiple sub-options and three vertical alignment options for each, took so much of the study team's time and effort that other available corridors, eg the Upfield line, were not considered. A good study doesn't need to test everything - it should be clear from the outset that certain combinations of horizontal and vertical alignment are badly sub-optimal, which will leave the study team free to target their work more precisely.

Possibly because of this drain on the team's resources, some variables were eliminated from the study and converted to assumptions - eg the service frequency of ten minutes (p133). By not testing the model with different combinations of variables the study team has left the project open to the possibility of vicious cost cutting - for example, if a future government decided to build the line but only allow for hourly services, the study would have no hard evidence to prove that the line would fail to provide an effective airport link under those circumstances.

In both the Strategic Merit (high level) and Rapid Appraisal (low level) sections of the study, equal weighting was given to every criterion - and each criterion was given only two possible values, there was no graduation of "more worse" and "less worse". This resulted in a relatively crude method of scoring. It should be the role of PTV to establish a standardised method of appraisal (a set of criteria, with weighting of each) based on an extended form of triple bottom line, formulated by customer surveys.

There were also internal contradictions in the text. The question of through-running with services in the Dandenong corridor (via the Melbourne Metro tunnel) was specifically avoided due to uncertainty of the way the rail system will be run when the rail link opens. However the benefits of single-seat journeys from the south-eastern suburbs to the airport were noted at several points. Even if the operational layout had been settled, the study made no reference to the requirement for extra rollingstock to service the longer route, or to track capacity on the Caulfield lines for the extra services. Surprisingly, near the end of the study on p145 the cost savings from through-running of trains are quantified, even though the possibility of through-running had not been confirmed.

Possibly the most disturbing inconsistency was that at several places the study showed signs of an incomplete understanding of track design - for instance, assuming the use of a flat junction next to a tunnel portal, instead of using a split portal arrangement to form a completely grade separated junction (which would have a negligible marginal cost). Also, on p103 the line is specified to cross the Maribyrning River after flying over a local road - which would require gradients too steep for trains to climb.


As a study, this shows severe flaws and its findings should be considered indicative at best. Before any moves are made towards building a rail link to the airport, the above three issues should be addressed.