Oct 2, 2013

What does urban rail really cost to build?

The official estimate of the cost to build the Melbourne Metro rail tunnel is $9 billion but some advocates claim it's only $3 billion. What explains the huge difference in these two claims?

Alan Davies — Editor of The Urbanist

Alan Davies

Editor of The Urbanist

Estimated cost of constructing Melbourne Metro, by Beyond Zero Emissions Inc

Last week I discussed the worrying tendency of some involved in public discussion of urban issues to treat the facts as if they’re optional (Infrastructure: does getting the facts right matter anymore?). In particular, I pointed to a list of rail projects formulated by Trains Not Toll Roads where the claimed costs in some instances are, I believe, misleadingly low.

One of those projects is the proposed Melbourne Metro rail tunnel. Trains Not Toll Roads, acting on the basis of research by the transport team at Beyond Zero Emissions Inc (BZE), claims the cost of construction would be $3 billion. This is wildly at odds with the accepted estimate of at least $9 billion.

BZE haven’t made their calculations public, but the organisation’s Media Coordinator, Ben Courtice, sent me a spreadsheet showing the major cost categories used to arrive at the estimate (see exhibit). I’ve sought some expert opinion on BZE’s calculations in an effort to understand the enormous divergence between the two figures. What follows isn’t exhaustive, but it indicates some of the key cost areas omitted from the $3 billion estimate.

Most of the $3 billion was arrived at by two calculations. First, $1.6 billion comes from multiplying the length of the tunnel (9.6 km) by an assumed average cost of $0.17 billion per kilometre (taken from the Phase 2 HSR study).

Second, a further $1 billion comes from building five stations at an assumed average cost of $0.2 billion each (no source provided). These two figures come to $2.6 billion – the lion’s share of the remaining $0.4 billion is attributed to Planning and Project Management.

The first thing to note is the $3 billion estimate doesn’t include any allowance for Risk and Contingencies. However the spreadsheet provided to me by BZE allows $0.9 billion for this component, bringing BZE’s estimate of the total cost to $3.9 billion, as per the exhibit. Yet strangely, this much higher estimate isn’t shown on the Trains Not Toll Roads web site.

The 30% allowance for Risk and Contingencies assumed by BZE might be acceptable when a lot of detailed work to prove-up a project has been put into the front end (the Government has allocated about $50 million to Melbourne Metro). However when the costings are based on very simple formulas without regard to the special circumstances of a project, a much higher allowance must be made.

There are also some areas of serious cost under-estimation in BZE’s analysis. Nothing has been allowed for Land Acquisition; just $1.6 million for Earthworks and Substructure; $30.6 million for Superstructure; and a mere $9 million for Signalling and Communications.

This might seem natural to a layperson as the Metro’s a tunnel, but it will have to be retro-fitted into a dense and busy urban area. It will require acquisition of expensive inner city and CBD land for stations and for staging construction. Adjacent land, extensive earthworks and large portal structures will be required where the line comes to the surface at Kensington and South Yarra.

Junctions will be needed to connect to existing lines and they will have to be built without disrupting services. Introducing new signalling technology is likely to be complex and the existing rail lines will require upgrades to the more advanced train control system.

BZE assume the five stations will cost $0.2 billion each, but it doesn’t appear that they’re aware Metro line trains will be 220 metres long (1.5 times current length). The Phase 2 HSR report (which BZE relies on for guidance on tunnelling costs) allows $0.46 billion for a single CBD station in Melbourne (1).

The estimates also make no provision for train maintenance and stabling facilities, for design and contractors overheads, or for provision of a train control centre. As increasing the number of services on suburban lines is a key benefit of the Metro (and no doubt contributes to the positive benefit-cost ratio), the cost to remove bottlenecks on these lines is also arguably a cost of the project.

I’m not in a position to say definitively whether or not the $9 billion estimate for Melbourne Metro is right (although Bent Flyvbjerg’s research indicates transport projects tend to exceed budget, so preliminary costings are likely to be even more ropey), but I think it’s pretty clear the claimed cost of $3 billion is misleading. The real reason for the difference between the published figures is political, not technical (2).

As I’ve noted before, the key priority for improving cities shouldn’t be to deny reality, however unpalatable, but to understand why urban infrastructure costs so much and what can be done about it (e.g. see Why is infrastructure so bloody expensive?Why do subways cost so much more here than elsewhere?; and If WA can build rail lines cheaply, why can’t the other States?).


  1. Stations can be very expensive. Although it’s not directly comparable because it’s a renovation and is above ground, the indicative cost to redevelop Flinders St Station in line with the winning design in the recent architectural competition is in the order of $1 – 1.5 billion.
  2. Those who saw the story in The Age last Saturday (Time to unearth London secrets) claiming London is building 42 km of tunnel for $25 billion should note that the it’s actually 21 km (it’s dual tunnel; the reporter counted both). That comes to $1.2 billion/km, similar to the (real) estimated cost of Melbourne Metro.
(Visited 87 times, 11 visits today)


Leave a comment

Your email address will not be published. Required fields are marked *

11 thoughts on “What does urban rail really cost to build?

  1. Smith John

    What is the latest official publicly available estimate of the cost of the metro rail tunnel? When was it published? Where does the figure of $9 billion come from?

  2. Samuel

    The cost of land is often excluded when the land has already been reserved or acquired, for example the EastLink business case (if you can find a copy). I presume the analysts presume it is a sunk cost. This is a mistake, as putting the land to use has an opportunity cost, even if no further cash outlays are required. The land could have been released for other uses.

  3. Teddy Bear

    In built up urban areas TBM methods are often the most cost effective because it minimises surface disruption. You would be surprised to see the amount stuff under our streets. Water, sewage, storm water, gas, electricity, telecoms, other roads, other railways and the secret passage from St Pauls to Young and Jacksons. All of this must be relocated under cut and cover approach.

    Once a TBM machine is in the ground and running it is very efficient. The major costs are often associated with portals and stations where it is necessary to break into the surface and find enough real estate.

  4. Jacob HSR

    About Bent Flyvbjerg, Alon Levy says in France and Spain most railway projects are built within budget:

    That’s interesting Alan, only 30% of the Metro Sur was built using cut-and-cover?!

    The South Morang extension of the Epping line is cut-and-dont-cover aka trench. I am wondering if it could it be cheaper to build a railway beneath an existing railway (in parallel)? eg, the Epping North decoupling, would it be cheaper to just build another railway parallel to and beneath the existing railway rather than TBM to Flagstaff?

    Especially new lines like HSR to Sydney or Melbourne Airport. Does it really have to be TBM or would it be cheaper to build a cut-and-cover railway beneath (and parallel to) an existing railway?

    As for noise during construction, how about the silent piling machines being used to build Crossrail in London.

    Disruptions are taking place for the Regional Rail Link, but that is not a reason to built it using TBMs apparently. It is being built in trenches.

  5. Waffler

    Alan, you are no doubt correct that the BZE estimate is only part of the story. On the other hand rail tunnels are much smaller and cheaper to dig than road tunnels.

    I still find it amazing that Governments are happy to talk about multi-billion dollar projects that will supposedly solve all the worlds problems. Rail engineers have been seeking relatively small amounts (still in the tens or hundreds of millions though) for decades to untangle some of the operational constraints that bedevil our rail system – single tracks sections, flat junctions, limited platforms, etc. and have had no luck.

    Same with roads. Let’s spend $6-9B on a big tunnel, but we haven’t been able to afford the myriad of smaller projects like duplications in growth areas, rail grade separations or intersection upgrades. I read somewhere that VicRoads can’t even afford enough traffic engineers to make sure the signals are operating efficiently (which is apparent from driving around) – surely the first task in “sweating the asset”!

    All these multi-billion dollar projects carry huge risks, especially when the planning hasn’t been done. Smaller projects reduce the risks considerably. And even if one goes wrong, the funding impacts are considerably more modest.

  6. boscombe

    Not sure if this has been included somewhere, but if the money is borrowed and will be paid back over many years, you should add the cost of the interest on the loan.

  7. Andrew

    It is essential to budget substantial amounts for land acquisition whether it’s for urban rail or urban road infrastructure. Even deep tunnelling (for road or rail) will require acquisition of deep-stratum where private property is not depth-limited to 15 metres.

    Usually road requires a wider alignment (at greater acquisition cost) than rail for a given capacity, but in both road and rail cases the more costly construction options (such as deeper and/or longer tunnelling) can be outweighed by the savings in having to acquire less surface land and/or less land overall. The cost of acquiring urban land has driven much engineering innovation.

    The cost of temporary occupation of land during construction (for laydown and storage areas, for example) would be additional.

    BZE needs to rework its model to include a credible allowance for what is an unavoidable and substantial cost.

    Whether BZE seeks to fund such infrastructure by betterment levies or land-value capture is another question entirely.

  8. mook schanker

    Those costs are laughable. Might see this kind of crap in an election soundbite, jesus BZE, way to paint yourself as some loon lobby group.

    A few things:-

    Where does the cost of permanent way (and slab track for the tunnels) fit in? That’ll be closer to a billion+. Let alone the tunnel facilities which add a bit; fire, ventilation & comms.

    High Capacity Signalling will be required for the whole rail line end to end to work properly (plus the depots will be near these end locations). Be closer to a billion to do that.

    $0 for rolling stock. Every applicable piece of rolling stock will need to be modified to use HCS, that is the train control system, through vehicle wiring and comms – no small feat. They would rather be looking at a completely new fleet of rolling stock for this line than modify anyway – there goes another half billion for a smallish fleet.

    Digging tunnels is all about the upfront Capex no matter how short the tunnel section is.

    The station cost will be at the top of the scale for the CBD, going from cut/cover street access to true underground to link with the underground rail loop. Never mind the costs to Yarra Trams & Swannie traders for chopping up Swanston St for 5 years.

    Jacob – cut & cover and TBM costs are easy to find on the internet.

  9. Alan Davies

    Jacob HSR #1:

    I’d be interested to know what scope there is for cut-and-cover once allowance is made for going under the Yarra, dealing with existing shallow underground infrastructure, and the cost of disruption. Even Madrid’s ultra-cheap Metro Sur (<$0.1 billion per km) reportedly only had 30% of undergrounding done by cut-and-cover tunnels.

    But I suspect disruption would be the big problem.Here's an interesting quote:

    Because of the negative community sentiment that often accompanies "cut and cover" construction, almost all new subway construction is done using the "deep bore" method. One exception was Vancouver B.C.'s recently opened Canada Line, and proves to be an excellent example of problems caused by the disruptive nature of the "cut and cover" method. One merchant has already won a lawsuit for C$600,000 – since overturned on appeal – due to damages caused by construction disruption, and 41 additional plaintiffs filed suit last year to recover damages. Interestingly, the amount of money they wish to receive is equal to the savings realized by building the line using the "cut and cover" method instead of the "deep bore".

  10. Jacob HSR

    It would be great if somebody could reveal the cost of Cut-and-Cover tunnels vs TBM tunnels.

Share this article with a friend

Just fill out the fields below and we'll send your friend a link to this article along with a message from you.

Your details

Your friend's details