Elon’s Tunnel

To my considerable surprise, Elon Musk has suddenly become part of the high-speed rail debate in Australia. In a Twitter exchange with independent NSW member of parliament Jeremy Buckingham last week, he gave a rough estimate of 750 million US dollars for a 50km, two-way tunnel through the Blue Mountains.

The Australian media promptly went beserk, reporting that Musk had “quoted” a billion Australian dollars for one of the longest tunnels in the world, a claim which raised eyebrows in a country long desensetised to infrastructure projects with nine zeros or more on the end of the price tag. Many didn’t even try to hide their disdain; Jamie McKinnell of the ABC called it a “thought bubble” right in the headline. The most widely-quoted expert rubbishing the idea was retired engineer and hydrogeologist Dr Philip Pells, who told the ABC “He’s out by at least a factor of 10 and up to a factor of 50”, even going so far as to disparage it as “Alice in Wonderland stuff” (but then adding, “I’m not trying to knock the guy” – gee, thanks). When one engineer says this sort of thing about another, you can be sure that one of them has said something very silly. Which one is it?

Musk’s Twitter estimate comes to $10 million per bore-kilometre, while Pells’ statement implies a range of $100 million to $500 million per bore-kilometre. I’ll assume that Pells’ numbers refer to a total project cost, rather than just the tunneling component, but there’s still way too much daylight between these two estimates. Let’s compare their numbers to a few real-world examples.

The most recent major tunneling project completed in NSW was the North-West Rail Link in 2018, 15 kilometres of dual-bore rail tunnel. Out of the $8 billion project, the tunneling contract was $1.15 billion, or approximately $42 million per bore-kilometre. This includes both tunnels and underground stations, so the true number would be somewhat lower than $40 million. Along the same lines, the 2013 High Speed Rail Study commissioned by the Rudd Government estimated the cost of tunneling alone to be about $25 million/bore-km (including digging the tunnel, lining with concrete, waterproofing, drainage, pavement, cross-passeges and service ducts). So far, Musk’s estimate is looking better than Pells’.

But a Blue Mountains tunnel would be one of the longest transit tunnels in the world if completed, and additionally would be very deep. The world’s currently-longest transit tunnel, the 57km Gotthard Base Tunnel in Switzerland, would therefore appear to be a close proxy for the proposed Blue Mountains tunnel. The total project cost at its completion in 2016 was 9.56 billion Swiss Francs – about $13 billion AUD at 2016 exchange rates, or $114 million per bore-kilometre. That’s close to Pells’ low estimate, so is that a win for Pells? Well, not really. We’re looking at total project cost, not just the tunneling itself. The North-West Rail Link tunnels were barely a seventh of total project cost. If we reasonably assumed that the Gotthard tunnels themselves were a third of the total project cost (a typical proportion), that’s $38 million per kilometre, which I call as another win for Musk.

Even projects renowned for their cost overruns and complexity don’t support Pells’ assertion. The Channel Tunnel, widely cited as one of the most expensive tunnels ever built, isn’t remotely that expensive. The total project cost was 9 billion pounds at completion in 1994. Assuming a 1994 exchange rate of 2.1 AUD/GBP, and using the RBA’s inflation calculator tool, that’s $34.6 billion AUD, or under $350 million per bore-km. Going even further, Boston’s “Big Dig” is the case-study in civil infrastructure cost blowouts. The incredibly complex project included a mixture of new tunnels (about 20 lane-km), upgraded existing tunnels, bridges and surface alignments, as well as environmental rehabilitation and the difficulty of undertaking a megaproject in the middle of an active CBD. The total construction cost ended up being $14.6 billion in 2006 USD ($26 billion 2018 AUD). Tunnels would have to be over 40% of total project cost in order to come to $500 million per lane-km. This is highly unlikely for such a complex project, and in any case, using the world’s worst-ever example of a cost overrun to argue that someone has severely underbudgeted their sums isn’t exactly a strong argument.

My verdict? If Musk’s sums are “out”, it looks like they’re out by between a factor of between 2 and 4 compared to conventional wisdom cost estimates. There’s no basis to accuse Musk of a 10x underestimate, and certainly not 50x. I’ve no reason to believe that Philip Pells is anything other than a perfectly competent engineer, but he got this one wrong.

For the NSW government’s part, they’ve responded to the idea positively – Transport minister Andrew Constance said that their own internal estimates of the cost of a Blue Mountains tunnel come to around $3bn (LINK) – an estimate which is far closer to Musk’s than Pells’. And given that Musk started The Boring Company in an attempt to bring down tunneling costs by a significant factor, it’s not surprising that he’s estimating something far lower than what conventional wisdom would tell you .

If there is anything to criticise in Musk and Buckingham’s plan, it’s the proposed mode of transport itself – an unproven system of lowering cars into the tunnel on platforms via an elevator, and then whooshing the platform along the tunnel at high speeds (up to 150mph, says Elon). With all due respect to Elon, this isn’t the best use of a tunnel. I’d guess that the elevator contraption might be able to handle 6 cars per minute at best. That’s 360 cars per hour, per direction – probably no more than 500 people at typical occupancy. Even if Musk achieved his highly ambitious goal of 4000 (!) vehicles per hour, that’s still well short of the passenger capacity of high-speed rail.

Let’s be serious. If we’re going to build a tunnel under the Blue Mountains, for $1 billion or $3 billion or $whatever, it should be for conventional fast rail, which can easily do 12,000 people per hour, per direction, and is proven technology for speeds well exceeding 150mph. Not only that, but it can get you right into the CBD on the existing tracks. Come to think of it, if we’re building 50km tunnels for a billion dollars, just extend the tunnel the whole way in to Central Station. Go to north to Newcastle and south to Wollongong while we’re at it. If Musk’s new tunnel boring technology is as good as he says it is, it really will change the game.

If I were Andrew Constance, I’d hit up Musk with a reply tweet: Make that a firm quote, and you’re on.


  1. Richard Macfarlane

    Good assessment, but you’re missing an important point – tube diameter. Musk’s tube is only 4.2 m diameter, compared to 10m+ for road or rail. Allow for the considerable reduction in material volume and concrete walls, Elon’s estimates could be right.

    Of course, he’s proposing a completely different form of transport. Individual autonomous electric cars (or car-carrying buggy) in close convoy, would give good capacity and convenience.

  2. I think that’s more or less what I was getting at – if Elon is correct in saying he has a way to significantly reduce tunneling cost, he could well be right (it would only be a 50-60% reduction on current costs, not a 90-98% reduction as claimed by his critic).

    The completely different mode of transport is a big red herring as far as I’m concerned – practicalities are entirely untested, and not even addressed in his article (ie, how on earth he plans to increase capacity to 4000+ vehicles per hour, more than 1 per second). The complexity of this would be mind-boggling; stations sure wouldn’t cost just $50mil. Really think it would be better to concentrate on conventional fast rail if his tunneling tech is as good as he says it is.

Comments are closed