The Overland, which has linked Adelaide and Melbourne since 1887, is the oldest direct intercapital rail service in Australia. Barring an eleventh-hour change of political will, it looks like that 132-year run will come to an end in a little less than two months time; Great Southern rail is no longer accepting bookings for the service beyond December 31 this year. This will mark the end of regularly-scheduled regional passenger rail in South Australia, a dour milestone indeed for the state.
To be sure, the writing’s been on the wall a long time. Most of the other regional services were lost over 30 years ago, and although The Ghan and Indian Pacific remain iconic rail tours, they no longer serve as the essential transport links that they once were (and in any case, they no longer make stops in regional SA). The Overland has remained as the sole twice-weekly exception, yet even so, passenger numbers have been dropping off a cliff, dropping from 100,000 to 20,000 per year in the last decade. With the South Australian government withdrawing their (paltry) $330,000 subsidy last year, the Victorian government reluctantly picking up the slack, and Great Southern Rail itself wearing $130,000 in annual losses, this decision was only a matter of time.
There’s little sense in throwing yet more taxpayer money at a little-used and obviously dying service just to keep it on life support for a few more years, unless you’ve got a plan to make that service competitive again. And as we shall see, it certainly could be competitive. What if The Overland were significantly upgraded to become Australia’s first intercapital fast rail service?
South Australia is not often considered high on the list for potential fast rail projects, on the assumption that its population is simply too low to support such investment. However, a closer look reveals a different story. Adelaide and Melbourne are Australia’s two closest mainland capitals, at a rhumb-line distance of 654km. There’s lots of potential demand; while not in the same league as Sydney-Melbourne, Adelaide-Melbourne is still a respectably busy air route, with about 2.5 million passengers annually. Most importantly, the terrain between the two cities is mostly flat, and the existing rail corridor is highly suitable for faster speeds (at least between Murray Bridge and Ballarat). Fix the final hundred kilometres or so to both terminal stations, and you’ve pretty-much built your high-speed railway.
Victoria is already starting to do this. A Melbourne-Geelong fast rail link is in the early planning stages, with an expected price tag of $2 billion. Meanwhile, a group of five regional Victorian councils is also pushing to upgrade the entire V-Line network to medium-speed or better. The most direct route between Adelaide and Melbourne would be via Ballarat, however the standard gauge route diverges at Ararat, going via Geelong instead (a bit over 50km longer). Either option would be perfectly suitable for a fast Overland service.
As for South Australia, I’ve previously argued on this blog, and elsewhere with economist Dick Blandy and former state rail commissioner Luigi Rossi, that a fast rail link between Adelaide, Mount Barker and Murray Bridge is an essential part of the state’s near-future transport system, and could likely be built for a cost of around $1.5 billion dollars, depending on the preferred route. By bypassing the slow existing alignment through the Adelaide Hills, the present timetabled journey of 125 minutes could be reduced to about 30 minutes express, about half the time it would take to drive.
In addition to the standalone suburban rail projects mentioned above, the remaining 464km of track between Murray Bridge and Ararat would certainly require some work, though probably not as much as you might first assume. The terrain is mostly flat, and the existing railway alignment well maintained and eminently suitable for higher speeds than currently prevail. Almost all curves are well over 1000m in radius, and are presently rated for speeds of “130km/h+”. Given the existing curve geometry, there is no reason why speeds well in excess of 200km/h cannot be safely achieved on almost the entirety of the existing alignment with only minimal modifications (you can check out the curve and gradient diagrams for yourself at the ARTC website, here for the South Australian side, and here for Victoria).
What would those “minimal modifications” entail? The existing track formation is continuously welded rail, mostly 60kg/m, on concrete sleepers and stone ballast in good condition, so the track itself is already of suitable quality; at worst the rails might need to be resurfaced and the ballast refurbished and tamped; an upper-estimate cost for this operation might be $100,000 per kilometre, so let’s call it $50 million for direct trackwork between Murray Bridge and Ararat. 200km/h speeds would also necessitate upgrading the present level crossings (mostly signs only) to gated crossings with bells and lights. At a cost of $1,000,000 each, and (say) one crossing every 5km, that makes $140 million for new crossings. Existing crossings of the main highway are already grade-separated. Upgraded fencing would probably be required in the towns through which the railway passes, totalling perhaps another $10 million. We can also assume that several additional passing loops would be required; at a cost of $5 million each and once every 50km, let’s call it another $50 million. That’s a pretty conservative estimate of $250 million between Murray Bridge and Ararat, very reasonable for an infrastructure project of this scale and performance (for a sanity check, see the ARTC’s current upgrade of the North Eastern Line in Victoria, a comparable cost per kilometre for a similar package of works).
An upgraded Overland would therefore be, by far, the cheapest fast rail route of any yet proposed in Australia, including the much shorter Sydney-Canberra route. Including the major works on the Adelaide and Melbourne approaches, the total construction cost would be $3.75 billion. Furthermore, the vast majority of that cost (over 90%!) is represented in two standalone exurban fast rail projects (Adelaide-Murray Bridge and Melbourne-Geelong) each of which would be economically viable in its own right. The $250m add-on cost of the full Overland upgrade is very affordable.
If the service were to use a high-end diesel tilting train with a top speed of 220km/h (along the lines the Spanish Talgo XXI, or even a modern version of Queensland’s Hitachi-based Diesel Tilt Train), we might reasonably expect average running speed in the Murray Bridge – Ararat corridor to be perhaps 180km/h, including stops. This would reduce the section’s travel time to 2 hrs 35 minutes, from the Overland’s present time of 5 hrs 45 minutes.
Adding 30 minutes for the Adelaide – Murray Bridge sector, that makes it 3hrs 5 minutes from Adelaide to Ararat. The remaining time from Ararat to Melbourne depends whether we take the Ballarat or the Geelong route. Assuming the travel times given in the Victorian Councils proposal for the terminal sections, and a 180km/h average elsewhere, this would imply an additional 90 minutes for the Ballarat route, or 99 minutes for the North Geelong route. That pretty much answers the question of which route is the more suitable; a nine minute difference sure won’t be worth the hundred million or so dollars that dual-gauging 211km of track will cost.
Therefore the total travel time on the 805km Adelaide to Melbourne route would be 4 hours 43 minutes, well under half the 11:05 taken by the existing service, and at an average speed of 171km/h it comfortably fulfills the definition of high-speed rail (top speed greater than 200km/h, and average end-to-end running speed greater than 150km/h). That kind of performance would once again make the train a serious competitor to all existing modes of transport; it would be miles faster than driving and not too much slower than flying once check-in, baggage collection and airport transfer at both ends are taken into account.
There are four existing transport modes that an Adelaide-Melbourne fast rail service could attract ridership from – they are air, private car, coach, and of course the Overland itself. The characteristics of each mode, and the ridership likely to be captured by the new fast rail service, are as follows:
- Air – This will be the main competitor to the train, and the main source of diverted demand. While the scheduled gate-to-gate time of air travel is far faster than any practical train at around 1 hour, the actual CBD-to-CBD travel time will typically be more like 3 hours (assuming arrival at terminal 1hr prior to departure, 15 minutes baggage collection, and 20 minutes CBD transfer at each end) which is beginning to make a sub-5hr train ride look more appealing. A typical full-fare economy ticket is $145 with Virgin, but often as low as $105. According to a demand model by Danish economist Peter Jorritsma (pictured below), at a rail travel time of 5 hours we might expect to capture at least 20% of the 2.5 million passengers who presently fly each year, which would mean 500,000 passengers diverted to the new Overland.
- Car – Driving is a smaller share of the market than you might think; according to data from the South Australian DPTI, just 1700 non-commercial vehicles cross the border on the Dukes Highway daily. Assuming that maybe a third of these are intercapital journeys (total guess), and using a typical vehicle occupancy of 1.6, that makes about 330,000 people per annum. The train could expect to capture a modest percentage of this traffic, perhaps 15%, which would mean about 50,000 trips per year. Intermediate trips could be expected to add perhaps the same amount (again, total guess), for a total demand of 100,000 passengers per annum. The cost of driving is also substantially higher than generally assumed; adopting a conservative value for the total cost of vehicle use of 68c/km (the ATO guideline), the cost of a one-way drive between Adelaide and Melbourne is a surprisingly high $494.
- Bus – Firefly is now the only coach operator offering Adelaide-Melbourne trips; if we assume the two daily services in each direction are at capacity (40 passengers), that makes a shade under 60,000 passengers per annum. At a fare level of $60, it is likely to remain an attractive option for price-sensitive consumers despite the 12-hour duration. We will assume the train will attract 50% of the market.
- Rail – The existing ridership of the Overland can also be added to the total; it’s not a lot, typically about 20,000 per year (estimates range from 17,000 to 24,000), at a normal economy fare of $164. Given that the new service will replace the old, as long as the fare is comparable we would expect 100% capture of this ridership.
- Induced – In addition to the above sources of demand, we could expect a small amount of induced demand: additional trips that are not captured from any existing mode, but are facilitated by the existence of the new service itself. In this case it will probably not be a significant percentage of the total, as despite the improved travel time, most intermediate destinations will remain too far from the capitals for regular commuting. Let’s say 100,000 passengers per annum (an average of about 25 trips per day, per town).
That makes a total of 750,000 passengers per annum, or a little over 1,000 per day in each direction, which is a fairly respectable ridership for long-haul interstate rail. Such a demand could be met by five daily services in each direction, running on a schedule of one service every three hours (say, departing at 6am, 9am, 12pm, 3pm and 6pm). This would require four trainsets, each of four cars and 200-pax capacity, and would allow a 75 minute turnaroud – plenty of time for refuelling, cleaning and personnel changes. Alternatively, a twice-daily return service using two larger trainsets of eight to ten cars each would have a similar capacity. In both cases, the cost of rollingstock is likely to be in the order of $160 million (conservative estimate of $10 million per car)
Keep in mind also, that the above ridership is the expected usage if the system were operational tomorrow. In reality, population growth between now and the completion of the system, as well as changes in population distribution and commuter behaviour as a result of the new transport option, would result in additional demand over and above what we have estimated here, perhaps significantly so.
Operating costs and fares
If the proposed service had similar operating costs to existing regional rail services elsewhere in Australia and the world, we would expect total operating costs of the four-car trainset to be around $30 per train-km, or $0.15 per seat-km. Assuming a reasonable capacity factor of 75%, that implies a cost-recovery fare of $0.20 per passenger-km, or $161 for an Adelaide-Melbourne fare. That’s about the same as the existing fare, but for a far superior service.
Nevertheless, such a fare level would be high compared against competing modes (an economy airfare is typically around $150, while the coach fare is just $60), without a material level of public subsidy. This isn’t surprising; most regional rail services in Australia and elsewhere in the world require at least some level of subsidy. The typical amount in Australia ranges from about $10-$20 per passenger trip, or around $0.10 per passenger-km (see e.g. this very long and detailed thread at Railpage)
The Overland’s present subsidy is about $1.08 million per year (wholly funded by Victoria since SA cut its $330,000 subsidy in 2018), plus a further $130,000 in operating losses to Great Southern Rail, for a total annual shortfall of $1.21 million, or about $61 per passenger (taking the conservative estimate of 17,000 passengers per year). That’s about $0.09 per passenger-kilometre, pretty typical for Australian regional rail, if anything somewhat on the low side. If the subsidy were continued at half the present rate per passenger, the increased ridership of the upgraded service would entail a total annual subsidy of $27 million, and enable a fare level of $125, which would be competitive against all transport modes.
Such a level of public subsidy, as well as the up-front capital costs of track upgrades and new rollingstock, would be easily justified by the significant external benefits derived, as outlined below.
- Direct User Benefits – The fast rail service will provide faster service at lower cost than most existing transport modes are able to achieve, resulting in total direct user net benefits of $66 million per annum. These benefits take the form of fare differential and travel time savings; user time-preference values are taken from assumptions used in the 2013 High Speed Rail Study.
- The most significant benefits accrue to users who switch mode from car, who save around $0.50/pax-km in direct costs. Also, drivers will save around 18 seconds per kilometre in reduced travel time, yielding an additional financial benefit of $0.10/km (even at the low estimate for time preference value of $20/hr, below minimum wage). For 100,000 trips per annum, this is a very significant annual benefit of $48 million.
- Benefits to coach users comprise much improved travel time (7 hours per trip, having a value of $140), offset by a disbenefit of higher fare level ($65 per trip). This represents a net annual benefit of $2.6 million.
- Rail users also gain a significant benefit, with travel time savings of at least 6 hours per trip resulting in net benefits of $2.4 million. Fare level is assumed to be equivalent.
- Using the above assumptions, Air users would derive net-neutral benefits, with higher travel time of 1.5hrs offset by lower fare level of $30. We will however assume that the portion of air users who switch to rail do so because their time preference places a lower disbenefit on rail travel time than air travel time (for reasons of comfort, convenience, security, productivity or amenity). Assuming these users value the disbenefit of rail travel time at a low $20/hr ($95 per 4.75-hour trip), while valuing the disbenefit of air travel time at a higher level of $40/hr ($120 per 3-hour trip), the net benefit of mode shift for these passengers is $25/trip, or $12.5 million per annum.
- Affordable Housing – The railway would facilitate a modest increase in population of the towns along the route and thus an increased provision of affordable housing, due to the median house price differential between the state capitals and regional centres. Median house prices in the capitals are $655,000 in Melbourne and $438,000 in Adelaide, with relevant regional townships significantly lower (see e.g. Horsham at $255,000, or Bordertown at $162,000). Based on these examples, a sustained housing price differential of $300,000 per dwelling is a reasonable expectation. Very conservatively assuming a railway-induced population growth of 1% in the served townships, this would result in a population increase of 400 people per year, or approx. 150 households. While the total number of people is small, the magnitude of the affordable housing benefit would be significant, around $45 million per year.
- Road crashes – a 2006 study by BITRE estimated the total cost of road crashes to the Australian economy to be $0.083 per vehicle kilometre travelled, for car travel. An upgraded Overland would take about 100,000 cars off the road annually, reducing car VKTs by 54.5 million per year (assuming an average trip length of 75% of the full distance). Car-related road trauma costs would be reduced by $4.5 million per annum.
- Congestion – The 2015 BITRE report Traffic and congestion cost trends for Australian capital cities estimated the marginal avoidable unit cost of congestion in Australian capital cities to be around $0.13 per passenger car-km (curiously, Adelaide’s value is almost exactly the same as for Melbourne). We will assume that each rail trip diverted from car avoids 15km of driving through the Adelaide suburbs, and 25km of driving through the Melbourne suburbs, for a total benefit of $5.20 per trip. Trips diverted from air will save roughly the same amount, due to avoided CBD transfers to Adelaide Airport or Tullamarine. The total avoided congestion cost is therefore $3.1 million per annum.
- Option Value – The provision of alternative transport modes in addition to car and air brings significant benefits to all residents, not just those who regularly use the service. It is typically measured by asking residents, via survey, the amount they would personally be willing to pay annually (by way of increased council rates, for example) in order to ensure the service were introduced, even if they did not themselves plan to regularly use it. While such benefits are difficult to quantify, they can be significant. The British Department for Transport uses a simple method that applies an annual “option value” to all households within the passenger catchment zone of each newly opened station – the value used is 241 GBP (2010 prices). If we applied this to The Overland, it would at minimum include the entire population of the regional townships served (41,000, excluding the satellite cities of Geelong, Murray Bridge and Mount Barker), and it would be reasonable to assume a significant option value for some proportion of the Adelaide and Melbourne population as well; perhaps the same population again would be a conservative estimate. At 2.2 persons to the household, and converted to 2019 AUD, this would represent a benefit of $19.4 million (though it could be much higher depending on the actual option value preferences of the large population base of the two capitals – it would be interesting to perform a survey to quantify the option value more accurately).
- Carbon abatement – The social benefit of carbon abatement measures is notoriously dependent on a wide variety of assumptions, and typically ranges from lows of $US10-50 per ton in less developed countries, to well into the hundreds in highly developed, densely populated nations such as Japan. The United States EPA in 2015 estimated an average, low-estimate of the social cost of carbon to be $US36/ton ($AU55/ton), which we will use here. With air travel typically emitting around 250g/pax-km more than rail travel, diversion of 500,000 passengers per annum from air to rail from the 650km Adelaide-Melbourne route would reduce CO2 emmisions by 81,250 tonnes per annum, with a commensurate social benefit of $4.5 million.
Assuming financing costs of 5%, up-front capital costs of $250 million for the railway (excluding the expensive terminal sections, which stand on their own financial merits) and $160 million for rollingstock, the total annualised public cost of the upgraded Overland would be $20.5 million in capital costs, plus the $27 million operating subsidy. With external benefits of $143 million annually, the Benefit-Cost Ratio (BCR) would exceed 3.0 in the first year. This calculation assumes a very modest proportion of induced demand and population growth in the regional townships served, and additionally there is a strong possibility that the above estimate for option value is highly underestimated. Should these conservative projections be exceeded, the BCR could very easily reach 4 or higher within a few years of the service’s introduction.
A BCR of 3 or above is indicative of a hugely beneficial investment (for comparison, the 2013 East Coast HSR study predicted a BCR of 2.3, and that was using an unusually low discount rate of 4%, for a time horizon of 50 years or more). However, the nature of the benefits are such that it would be difficult for a private railway operator to internalise them – an attempt to do so via increased fares, for example, would likely reduce patronage to such an extent that the benefits would be greatly reduced. There may be some potential for the railway operator to pursue value capture opportunities in regional townships, however this would be highly dependent upon appropriate development policies at state and local governments.
It is therefore likely that the most appropriate model would be either a fully government owned and operated service, or a semi-privatised model whereby a private operator is contracted by the government to provide a set level of service for a given timeframe.
A revitalised and much faster, much more frequent Overland rail service between Adelaide and Melbourne would revolutionise regional transport and regional development in the South East and Western Districts regions of SA and Victoria, delivering wide-ranging benefits far in excess of the quite sensible public costs. This is a project that should be adopted by both State governments as a top-level infrastructure priority. Given the scale of construction needed and the good condition of the existing track, there is no reason this service couldn’t be up and running before 2025.
In the meantime, let’s keep The Overland’s unbroken record intact. The present level of subsidy required from the SA government – just $330,000 per year! – is incredibly modest, both on a per-passenger basis, and in absolute terms. Heck, it’s in the same ballpark as a television commercial during a major sporting event. The South Australian government was happy to spend millions airing its controversial “Old Mate” tourism ad at this year’s AFL and NRL grand finals, which suggests a curiously inverted hierarchy of of priorities. Even in its current form, The Overland remains a critical piece of state infrastructure, relied on by thousands of regional citizens. South Australia deserves better from its government – it deserves to keep and improve its regional rail services.