Fast rail on the freeway: Another approach for the Canberra line

Hot Rails has previously looked at the Southern Highlands corridor, estimating the cost of a 200km/h corridor all the way from Goulburn into Central Station to be in the order of $3.2 billion. However, the most critical part of that railway is the section between Campbelltown and Mittagong, which bypasses a long and very slow section of existing track. If this section were built in isolation, and the existing railway used elsewhere, a significant fraction of the benefits could be obtained for relatively low cost.

There are very few options for transport corridors in this part of the world; wedged between the rugged Nepean and Nattai national parks, the Hume Highway already occupies by far the most practical route. It’s got gentle curves (a minimum of 1800m radius), a typical gradient well under the 4% practical limit for fast passenger rail, and a median strip more than wide enough to accommodate a dual-track fast railway. This is therefore the obvious play.

There are numerous examples of railways being built on freeway median strips; see for example the Kwinana Freeway in Perth, and Melbourne’s Eastern Freeway, whose wide median was originally designed to take the planned Doncaster Railway Line. Wikipedia has a long list of further examples. For Campbelltown to Mittagong, the advantage of using the Hume corridor cannot be overstated; the economic and political difficulties of acquiring private land or national park are avoided, the railway can use existing grade separations, and engineering design difficulty (and expense) is minimised in general.

The proposed new track would be 54km of double-track. The Hume Freeway median would be used for 46km between Macarthur and Mittagong, as well as a short section of the mostly-disused Picton Loop. We also propose electrification to Moss Vale, to enable the use of efficient and powerful electric or bi-mode rollingstock. In addition to the section on the Hume median, there are three major works that would be required: the transitions onto and off of the freeway corridor (at Aylmerton and Macarthur, respectively), and a viaduct at Alpine to keep the gradient within acceptable limits. We’ll look at each of these in turn.

Existing track (yellow) and proposed new route on Hume Highway median strip (blue)
Proposed fast rail alignment on Hume Freeway median strip.
Dimensions (in metres) of Hume Freeway and adjacent railway corridor near Yerrinbool. Base image: Nearmaps

Aylmerton Junction

From Mittagong, the new route crosses over to the Hume Freeway shortly after Mittagong Station. Using the former Picton Loop line until Braemar, a short embankment followed by a 700m viaduct (about 10-15m high) brings the alignment over the southbound lanes of the freeway and onto the median strip. Minimum radius is 1800m, meaning trains will be able to accelerate to their full speed of 200km/h + on departure from Mittagong Station.

  • Earthworks: 54,000m3 – $864,000
  • Fencing: 3.6km chainlink (both sides) – $360,000
  • Viaduct: 700m with 25m spans – $28 million
  • Dual track turnout – $1.5 million
  • Dual track control point – $2 million
    • TOTAL – $33 million

Macarthur Junction

The new route departs the freeway median strip just south of the University of Western Sydney campus at Macarthur, dropping into an approximately 550m-long tunnel, starting about 500m before the existing railway underbridge. Aside from cutting off a small area at the rear of several rural residential blocks, the alignment remains entirely within the existing road and rail corridors. Minimum radius is 800m, restricting speed to 160km/h (tilting trains) or 120km/h (conventional).

The alignment remains on a dedicated dual track on the southern side of the rail corridor until shortly after Macarthur Station, where it rejoins the existing East Hills Line tracks.

  • Tunnel – $40 million ($73 million/km)
  • Land acquisition – $1 million (max)
  • Fencing – $0.5 million
  • Dual track turnout – $1.5 million
  • Dual track control point – $2 million
    • TOTAL: $45 million

Alpine Viaduct

A 1.4km long viaduct will be required to keep the gradient to a maximum of 4%, on a small section of highway between the towns of Alpine and Yerrinbool where the highway gradient briefly exceeds 6%. The required viaduct would be very similar to the Hallerbach Viaduct on the Cologne-Frankfurt high-speed railway in Germany (this YouTube video of the viaduct is mesmerising)- it has 40m spans, and is a maximum of 36m in height above the valley floor. The Alpine Viaduct would have similar specifications.

Hallerbach Viaduct, Germany. Image: Wolkenkratzer, Wikimedia Commons


  • 56km of dual track (200km/h class) – $126 million
  • 46km of dual concrete barriers to segregate freeway median: $32 million
  • Bridges – $61 million, comprising:
    • Nepan 1 – 300m length with 150m mainspan – $25 million
    • Allens Creek – 250m with 50m mainspan – $13 million
    • Nepean 2 – 285m with 50m mainspan – $14 million
    • Nepean 3 – 185m with 50m mainspan – $9 million
  • Aylmerton Junction – $33 million
  • Macarthur Junction – $45 million
  • Alpine Viaduct – $70 million
  • Overhead Line Macarthur to Moss Vale (70km) – $105 million
    • SUBTOTAL: $472 million (direct construction costs)
  • Soft Costs (contingency, preliminaries & general, and client costs): $283 million
    • TOTAL: $755 million


The existing line from Campbelltown to Mittagong is 76km long, and the XPT service takes a timetabled 55 minutes, making an average speed of almost 83km/h. The new proposed route would be 19km shorter, fully grade-separated, and have a minimum radius of 1800m, meaning the allowable operating speed would exceed 200km/h. If we assume an average speed of 180km/h allowing for accelerating and deceleration at either end, the travel time in this section alone would be cut by 36 minutes.

Existing RouteNew Route
Distance76 km57 km
Speed82.9 km/h180 km/h
Time55 min19 min

Subsequently, the existing route would be used without modification. If a tilting bi-mode train such as the Talgo 250 Dual were used, we could expect 25% faster speeds throughout. This would result in a further 13 minute saving between Mittagong and Goulburn, and 19 minutes saved between Goulburn and Canberra (similar savings were seen during the 1995 trial runs of the Swedish X2000 tilting train). Assuming no further time saving in the exit from Metropolitan Sydney (due to existing traffic), this would result in a travel time between Central and Canberra of 3:02hrs, a saving of 68 minutes over the existing timetable. This could easily be brought down well below 3hrs by offering a limited-stops or express service; the current timetable has 9 intermediate stops, each of which adds at least 2 minutes to the schedule.

This may not sound spectacular, but 3 hours is a significant performance milestone. For starters, it makes the train time-competitive with driving (3 hours flat, if you’re lucky), and way better than the bus (3:30hrs). It’s even fairly good against the plane, which takes just 50 minutes gate-to-gate, but adding check-in (45 min if you’re pushing it?) and CBD transfer at either end (optimistically 15 min?), the total is easily over two hours. Furthermore, the core Southern Highlands towns of Mittagong, Bowral and Moss Vale would all be brought within one hour of Central, and Goulburn to well under 2hrs. The train would become – by far – the fastest commute method, and a significant player in the transport mix.

Passenger Demand

Modifying the estimate from a previous post (Passenger demand for a Sydney-Canberra fast train), we can expect ridership of 3.65 million passengers per annum (mppa), comprising passengers captured from the following modes:

  • Air – At a 3 hour travel time, we would expect to capture around 60% of the air travel market of approximately 1 million passengers per annum (mppa); this capture would equal 0.6 mppa.
  • Car – I previously estimated a contestable travel market of 46,505 car passengers per day on the Sydney-Canberra corridor (17.0mppa). AECOM estimated a mode capture of 20% for a very high speed service (1hr SYD-CBR); the capture for a 3hr service would be significantly lower (say, 5%), which would come to 0.85mppa.
  • Coach – the 2013 High Speed Rail study estimated 3,515 coach passengers per day between Sydney-Southern Highlands and Sydney-Canberra. If the faster train were to capture 50% of coach ridership, this would be 0.64mppa.
  • Rail – Pre-COVID, the Southern Highlands line was averaging around 1.1mppa (see Train Patronage by card type), while the intercity services (Sydney-Canberra and Sydney-Melbourne) would add another 0.5mppa (see Train Statistics 2014, pp57). We can assume that all these services will switch to the new track, therefore rail mode capture will equal 1.6 mppa.
Sydney-Moss Vale02102711223013251624625
Passengers per day for various transport modes, and the estimated number that would switch to the faster rail service.

This demand could be met by the following timetables:

  • Sydney to Moss Vale: 18 trains per day in each direction, hourly between 4am and 10pm (similar to the current timetable), using 4-car, 200-seat trainsets
  • Sydney to Goulburn: Four daily trains in each direction, including two en-route to Melbourne, each of 8 cars / 400 seats
  • Sydney to Canberra: Nine trains daily in each direction, every two hours between 4am and 8pm, each of 8 cars / 400 seats

Focusing on the Canberra service, with a three hour travel time, this timetable would require four trainsets, allowing each service almost a full hour for turnaround. Compared to the existing daily service of three 3-car trains in each direction, this would be a tripling of service frequency, as well as 2.5x the capacity per train. The expanded service hours would enable journeys that either arrive before business hours, or depart later in the evening, making the service far more attractive to a wider variety of travellers.


Assuming a typical fare structure of $0.20 per kilometre, one-way fares would be $23 to Bowral, $39 to Goulburn, or $57 to Canberra (comparable to existing rail or coach fares). At the ridership levels estimated above, that would entail annual farebox revenue of $160 million.

Comparable intercity rail operators in the UK have operating cost structures of around $2.75 per carriage-km (including rollingstock lease costs). Applied to the train sizes and frequencies above, annual operation and maintenance of the rail service would be $69 million, giving an annual gross margin of $91 million. This would be around triple the finance costs on the $755 million capital expenditure required to achieve such profit; such a cost structure should therefore make a project such as this highly attractive to the private sector, or indeed to NSW Trains.


  1. A perfect plan as we see superb developments in Western Australia immediately attracting passengers across the whole day. This is very important for people whose work site changes and they can remain in their existing home. See the cross over line fro. Beeliarbwest to Fremantle creating a circle.

  2. Edwin, I hope you find support for this. Good thinking.

  3. Access to Canberra (ACT) is trumped by QANTAS and the Airport Owner channeling the traffic selling the so called “convenience” of a terrifying and cramped air ride to Sydney. Murrays provide a stately bus trip. The airlines have the political market in their hands. A fast (not really fast) train provides a smooth regular ride equalling the air trip via airports from city centre to City Centre. That option is constantly quashed indicating some strange pressure effect.

  4. I have thought for a long time this is the section where the most gains could be made. Good job well done. I hope they (DoT) will listen and get on with fast rail in NSW.
    I suspect the risk profile of combining Freeway & High speed rail maybe an issue, but with the right barricades can be minimised. There is no such thing as a risk free journey.

  5. Doing it on the cheap is entirely the wrong approach. Doing it expensively means we get the right project – the one that kills aviation domestically and sets Australia up for habitation right along the eastern coast.

Comments are closed