Elon Musk has been hyping an idea for a “fifth form of transportation” that would travel between downtown San Francisco and Los Angeles in 30 minutes and cost only $6 billion to construct – a fraction of the $68 billion high-speed rail project slated to connect those cities. So far he’s been coy about the details, but he’s set to reveal the plans on August 12th. Tech aficionados are trying to predict the design before the announcement. Here are the hints they have to work with: It will travel between SF and LA in 30 minutes (roughly 700 mph, or twice the speed of a plane), an SF to LA version would cost about $6 billion to set up, and it is a "cross between a Concorde and a railgun and an air hockey table".
Most speculation centers on some kind of underground train propelled by fancy new technology, possibly acoustic waves. Personally, I think that these ideas miss something essential. The central problem for building new infrastructure is not the lack of futuristic concepts (see maglev trains), it’s that constructing something – anything – that stretches the 380 miles from LA to SF is extremely expensive. Simple rural highway construction costs around $15 million per mile (9.84 million/mile in 1997 dollars) and even building a sidewalk in an urban area can cost $2 million per mile.
Elon’s cost estimate of $6 billion divided by the 380 miles from SF to LA leaves about $15.8 million per mile for construction costs. Boring a tunnel through the ground would be far too expensive: subway construction costs range from $93 million per mile for Madrid’s Metrosur line, to $210 million per mile for Singapore’s new Circle Line, to the astounding $2.73 billion per mile for New York’s 2nd avenue subway. You could perhaps save money by putting the train in an aboveground tube, but pipeline construction is not cheap either. The average shale gas pipeline cost $200,000 per inch-mile to construct in 2011, meaning that a 2 meter diameter pipe would come in at roughly 15.8 million per mile. This pushes the limits of the hyperloop budget even without accounting for new technology development or the bureaucratic headaches that come with building an aboveground tube through populated areas along the route.
Construction projects are expensive, and massive construction projects are massively expensive. Regardless of whether the final product is a new highway or a fancy acoustic train, much of the work is trucks and excavators shoveling dirt around.
The point I'm making is that technology for a new kind of train is not enough, you also need new construction technology or a new way of managing large construction projects. Without this, the hyperloop will go the way of the maglev train -- a few expensive demonstration lines, but not widely used and not transformative for society.
One interesting idea along these lines is modular, factory-built construction. With modular construction, you build the structure offsite in a factory and then transport it in pieces on trucks to the final destination for assembly. This method has been getting attention lately because of modular skyscraper construction projects in New York and China. Musk has mentioned modular construction before, in some offhand comments about building adding a double-decker central lane to the 405 freeway, so he is familiar with the concept.
Factory construction allows for the use of big fixed machinery to increase productivity, and the factory floor is protected from bad weather that can delay on-site construction. The method has even more advantages for construction in populated areas. Because most of the construction is done in a remote factory, the construction time on location is much faster. In the video below, a new pedestrian bridge gets installed over a busy road. By assembling the bridge offsite, they are able to hoist the bridge into place all in one night, avoiding extended traffic snarls.
Even if modular construction isn't helpful for hyperloop construction (whatever the hyperloop turns out to be), the larger point stands. When designing practical technology, it’s not enough to think about just the device -- you need to consider the whole technological system from manufacturing or construction, to operation, to maintenance. Musk has proven himself to be a master system builder at Tesla and Spacex. At Spacex, for example, he did not set out to build one of the many futuristic launch concepts, like a space elevator. He stuck with the same multi-stage rockets that have been sending things into orbit since the 1950’s. All of the impressive cost reductions came from optimizing the technological system surrounding the manufacturing and launch of rockets. There was no one breakthrough technology, it involved everything from a new machine for friction stir welding together the fuel tank, to automated sensors that reduce the size of the launch control crew, to procedural changes like full-system testing of the rocket, when most aerospace companies will only test components and subsystems.