$100 million to save The Overland

In light of the impending closure of The Overland rail service between Adelaide and Melbourne, a few days ago ABC Radio Adelaide issued a challenge to its listeners to come up with ideas to make The Overland viable, for a budget of $100 million and within a timeframe of 10 years. Host Jules Schiller invited me on to the show to present a potential solution; this post is a more detailed look at the solution I suggested, and a look at how the performance and business plan would stack up.

First off, $100 million is not really a whole lot when it comes to faster rail. It’s not enough to do any major realignment of slow sections of track, or even widespread upgrading of existing track. It’s just about enough for a couple of eight-car trainsets. There’s not a lot of point upgrading tracks if you’re not also going to buy fast trains to run on those tracks, but there is still benefit to be had in using rollingstock that can take maximum advantage of your existing tracks. So if $100 million is your budget, then new trains are the only solution that’s within your budget.

We’re therefore going to explore how much faster The Overland could be by running faster trains on the existing track. To do this, we’ll consider the performance that we could get by using tilting trains. Such trains – which work by leaning into the curve in a similar manner to a motorbike rider – are in regular operation in every country with a modern rail system; even Australia has them – Queensland has been operating them for decades with great success. Depending on the technology and the cant (bank angle) of the existing track, their speed can be between 25% and 50% higher through curves than for normal trains. And because the whole point of a tilting train is to go faster, their top speeds are typically pretty high as well – between 160 and 250km/h.

Most major rollingstock manufacturers have some type of tilting train in their lineup, however one in particular stands out for use on The Overland – the Spanish-made Talgo XXI. It has a top speed of 220km/h, and a passive tilt mechanism that offers mid-range speed improvements through the curves – perhaps 30%. But the feature that really sets the Talgo apart is its dual gauge capability, meaning it can operate seamlessly between the broad-gauge suburban networks of Melbourne and Adelaide, and the standard-gauge interstate railway. Not only that, but Talgo have been actively courting the Australian market in recent years, even offering to bring a demonstration trainset out for a trial run on the Sydney Canberra line, at no cost to the government. It’s likely that the company would be very enthusiastic if invited to participate in a bid to revive The Overland.

We’ll assume no trackwork will be required at all. The Overland currently uses the standard gauge ARTC-owned interstate railway for its entire length; as it’s the main freight route between Melbourne and Adelaide, it’s well maintained with concrete sleepers and heavy, continuously welded rail – already well suited for higher speed operation. The route is 828km in length; at a travel time of 10:35 hours, that makes an average speed of about 78km/h. Driving, on the other hand, takes about 8 hours. To make the train faster than driving (a minimum performance baseline) would therefore require a 38% overall speed improvement.

That sounds fairly achievable – let’s see how it could be done. The railway lends itself to analysis in three sections:

Adelaide to Murray Bridge (96km) is characterised by steep grades and tight curves. Speeds here will be mostly limited by curvature; we’ll assume a 30% speed increase through this section, which would translate to a time saving of 29 minutes.

Murray Bridge to Ararat (464km) is mostly flat and straight, and particularly well suited to higher speeds. Even the curves are very gentle, almost all well above 1000m in radius and rated to “130km/h+” even for non-tilting trains. 200km/h or above should be easily feasible on the majority of this section with minimal trackwork, however regulatory speed limits may be imposed due to the presence of ungated and unlighted level crossings on much of the route, particularly on the South Australian side. We will assume speed limits of 160km/h without level crossing upgrades, or 200km/h if all level crossings are upgraded with gates and lights, which would correspond to time savings of 129 and 160 minutes, respectively.

Ararat to Melbourne (268km) has moderate curvature and grades, but also increasing traffic limitations as Melbourne is approached. We’ll assume a 30% improvement on currently prevailing speeds in this section. Additionally, by using the Talgo dual-gauge system, we can make a significant shortcut by switching to the broad-gauge V/Line network at Ararat, taking the Ballarat route instead of the Geelong route. This will cut 56km off the journey. With faster speeds and a shorter route, 87 minutes can be saved in this section.

Altogether, simply changing The Overland to a 160km/h dual-gauge tilt-train could cut slightly over four hours off the journey, reducing the total trip time to 6:30hrs, with an average speed of 119km/h, easily beating our minimum goal to be faster than driving. With a modest further expenditure to upgrade or remove all ungated level crossings would allow the train to reach its top speed of 220km/h, and bring the total duration below 6 hours.

In either case, this would be a gamechanging step up in performance, far outpacing private automobile or coach, and offering a genuine and unique alernative transport mode. While not able to match the CBD-to-CBD travel time of flying, The Overland would serve a multitude of intermediate regional destinations that do not have regular air service. The faster travel time would allow one daily return service per trainset, which in turn would facilitate more frequent and convenient departure times, further expanding The Overland’s market.

But in the end, would this be enough to “save” The Overland? That’s a two-part question. As the SA government has made plainly clear that it has no interest in providing even the most modest level of public subsidy towards keeping The Overland going, you’ve not only got to attract more passengers, you’ve got to do so at a profit.

For passenger numbers, we can use the same assumptions from my previous Overland post, but assuming a 6.5 hour travel time instead of 5 hours. At that level of service, we might expect to capture 5% of the air travel market. With 2.5 million air passengers annually, that’s a potential market capture of 125,000 passengers per annum. Add in potential market capture from car (maybe 50,000), coach (30,000) and the existing ridership of The Overland itself (say 15,000) and we’re at 220,000 passengers per annum, or about 600 passengers per day.

Such a demand could be met by a single daily return service, operated by an eight-car, 400-seat trainset. Alternatively, if the market demanded a more regular service, that same ridership could be served by two smaller trainsets (say, 4 cars and 200 seats) which would offer two daily return trips in each direction. This would have the advantage of allowing the same departure times from both Melbourne and Adelaide, rather than having just a Melbourne-bound morning service and an Adelaide-bound evening service. While an overnight service would be possible, the relatively short duration would necessitate a departure time inconveniently late in the evening, or an arrival far too early in the morning.

It certainly appears that an improved Overland would attract sufficient passenger demand. But could it do so profitably?

Assuming an average fare level of $150 (accounting for a mix of first-class and economy fares, as well as shorter intermediate trips), annual revenue would be $33 million. A reasonable return-on-revenue of 30% ($9.9 million) would give the project an attractive payback-period of about 5 years. Achieving this commercial return would require that total expenses be kept below $23.1 million per year.

Capital costs would be limited to the purchase of the new tilting rollingstock – whether we go for a single 8-car trainset, or a pair of 4-car trainsets, in both cases the cost is likely to be in the order of $50 million. Assuming financing costs of 5%, that represents an annual cost of $2.5 million.

We’ll assume that track access is charged according to the ARTC’s price schedule, available here. For a 270-ton, eight-car trainset, the effective charge would be $3.12 per train-km; that works out to about $1.8 million per year.

This leaves $18.8 million for direct operation and maintenance costs, which would include fuel, maintenance, depreciation, salaries, cleaning and ticketing. On a unit basis, that’s $33.43 per train-km, or $0.11 per passenger-km.

Is this a reasonable expectation? Data from the Victorian V/Line system gives average operation and maintenance costs around $40 per train-km, so the new Overland service would have to be about 15% more efficient than V/Line. This should be eminently achievable – there would not be any track maintenance or network operation costs, and the long route distance would mean that variable costs would tend to dominate, diluting the fixed and turnaround costs. More generally, US transport expert and blogger Alon Levy estimates that a lean, low-cost intercity rail operator could, in theory, keep total O&M costs to as low as $US0.033 per passenger-km ($AU0.05), less than half the $0.11 we have assumed here.

So based on these numbers, The Overland can indeed be saved, and profitably at that. It would cost just half of the ABC’s suggested $100 million budget to purchase the new trainsets, and require the co-operation of South Australian and Victorian rail network operators to access their metropolitan networks and terminal stations. This could quite likely be achieved within 12 months, not 10 years. All of this of course would have to be subject to a far more rigorous cost-benefit analysis, but it strongly passes the back-of-the-napkin test, which is a very good start indeed.

Looking perhaps another 5 years ahead, if the upgraded Overland were as successful as predicted, there would be demand for further speed improvements. Upgrading all ungated level crossings remaining on the line would likely cost in the order of $100-$150 million. As the benefits would accrue to all road and rail users as well as the wider community, this cost would be borne primarily by government, not the railway operator. This would allow the trains to achieve their maximum speed of 220km/h, reducing travel time to below 6 hours. At that speed, we could expect patronage to almost double, requiring a doubling of the fleet size to meet the increased demand.

Further ahead, perhaps by 2030, exurban fast rail projects in Adelaide and Melbourne could bypass the existing slow approaches to each terminal station, bringing travel time down nicely below 5 hours and attracting perhaps 20-30% of the current air travel market. Total ridership would be around 750,000 passengers per annum, and due to the much faster travel time, this additional demand could be met by the same fleet of trainsets, together running a total of five return services per day. The cost of these projects would certainly be in the billions of dollars, but they would each serve a significant commuter market close to each city, and be profitable in their own right. The additional cost to The Overland would therefore be small.

A future of fast, comfortable and energy efficient travel will be far more likely if The Overland is retained. A well-patronised transport service drives demand for its improvement; a poorly patronised service drives calls for its abandonment. This is why a simple continuation of the present subsidy will do little more than delay the inevitable – without significant improvements in performance, passenger numbers will remain low, and there will be no impetus for growth. The Overland is a remarkable business opportunity that has been hiding in plain sight, and if neither state government is willing or able to grasp it, then they should at least make every effort to encourage a private sector project, should it be seriously proposed.