Rail Construction Costs

I wanted to have a look at prospects for rail to Chadstone and Monash Uni in this post, but Eddington has inconvenienced me somewhat. He came up with some outrageous cost estimates for rail construction – far higher than the Northern Central City Corridor Study figures I’ve been using for costing projects like a north-south tunnel and airport line.

Clearly, I can’t go on using the NCCCS figures over the Eddington ones just because I like the fact they are lower – ultimately they are both theoretical studies and it would be poor form to rank one above the other without good reason. So I’m going to take a more empirical approach, and look at the costs of two new Australian rail projects – the Mandurah rail line in Perth and the Epping-Chatswood line in Sydney. Mandurah opened a few months ago and Epping-Chatswood should open in a few months, so the cost structures facing these projects should be broadly similar to those in Melbourne. This should give a more accurate picture of what new rail projects will cost.
The NCCCS Estimates

The NCCCS, released in 2002, costs double track surface rail at$1.9m/km, overhead at $0.8m/km and signalling at $0.8m/km. That’s basically $3.5 million per kilometre of railway in 2001 dollars. Indexed to 2007 dollars, the cost is $4.12 million per kilometre for above ground double track rail, assuming no property acquisition. Stations are not included.

For deep level underground rail, the underground railway at $40 million/km and underground stations at $50 million. In 2006 dollars (the most recent full year figures the RBA has on their calculator), that’s $47.14 million for the rail and $58.93 million for the stations.

Note these figures are slightly higher than those used in my posts on a north-south tunnel and airport line. This is because they were costed in 2006 dollars (2007 dollars not available at the time), whereas the figures above are in 2007 dollars. The difference is small though.

The Eddington Estimates

Getting the actual costings in the Eddington report is actually fairly difficult. This is partly because the report is so long, but it also takes the approach of making it as difficult as possible to find detailed explanations of the costs. Getting marginal costs per km seems to be impossible, and there is a failure to even separate station and track costs. Furthermore, the costs provided are vague – the headline figure for the public transport option is between $7.5 and 8.5 billion, without really explaining why.

The best information I could get was in Appendix C of the Engineering Design and Costing Report (watch out it’s a big file). It gives us slightly lower costs – $5.9 billion for a 17km rail tunnel, $1.1 billion for a 26kmTarneit rail line and $200 million for DART; yielding a total of $7.2 billion, all in January 2008 dollars. Strangely, other parts of the report cost the tunnel at $7 billion and DART at $250-300 million – there’s no consistency.

The lowest cost figures give a per km cost of around $347 million for rail tunnel (presumably with stations included) and $42.3 million for above ground rail. The above ground rail figure is basically useless for estimating rail construction costs though, because the Tarneit line requires substantial land acquisition. Interestingly, if we assumed that the land acquisition costs were massive (and let’s remember that this line goes through farmland) and took up half of the total cost, the rail construction cost would still be 5 times what the NCCCS estimated for surface rail.

Mandurah rail cost

The recently completed Mandurah line in Perth is 70km long, and cost $960 million in 2006/07 dollars. The whole New Metro Rail project cost over $1.4 billion, but included numerous other works and the purchase of a fleet of trains. The Mandurah line has 11 stations, two of which are underground, which are included in the $960 million cost. Per km, Mandurah cost $13.7 million, including 11 stations and an underground section.

Epping-Chatswood rail cost

The soon to open Epping-Chatswood line in Sydney runs underground for 12.5 km. It cost $860 million to build. Given $860 million figure was not given a year, I’ll assume that it’s in 2002 dollars (the year the project started), giving us a cost of $984 million in 2007 dollars. The per km cost is $78.72 million. However, the Thiess portion of the project only covered tunnelling, rail and station shells. Total project cost was higher – $2.22 billion in 2007 dollars. This gives at total per km cost of $177.6 million. To find out how I calculated these figures, click here.

Conclusion

The table below sums up the $m per km construction costs as measured by the NCCCS, Eddington, Mandurah and Epping-Chatswood, measured in more or less current dollars.

* does not include stations
# includes land acquisition
^ includes only tunnel, rail and station shell cost

The upshot is that empirical evidence suggests that the NCCCS estimates are slightly too low (not that low – we should remember that these estimates are the only ones which allowed us to completely remove stations from the equation), and that the Eddington estimates are way way way way too high when compared with actual rail projects happening in Australia. From now on, I’ll rely more on the empirical data (alongside the NCCCS figures) when looking at potential new infrastructure.