Rockets that come home

Why the first attempt at reuse failed, and what changed the arithmetic

A The economics of spaceflight were fixed for fifty years by a single arrangement: the vehicle was destroyed on every flight. A rocket is mostly tank, and the tanks, engines and structure that reach orbit are abandoned there or allowed to burn up on the way down. The comparison usually offered is with aviation, where an aircraft that could carry passengers once and was then discarded would make a ticket unaffordable no matter how cheap the fuel. Fuel is, in fact, a trivial fraction of the cost of a launch — a few per cent at most. Everything else is the vehicle and the people who build it. The proportions are worth stating plainly because they determine what any attempt at economy must attack: a programme that halved its fuel bill would save almost nothing, and a programme that flew the same hardware ten times would change the industry.

B Reuse was therefore an obvious ambition, and the first serious attempt at it is the reason the ambition acquired a bad reputation. The American shuttle recovered its orbiter and its boosters and was promoted as the vehicle that would make access to orbit routine. It did not. The thermal protection required inspection and replacement on a scale nobody had anticipated, the main engines were removed and overhauled between flights, and the refurbishment consumed the savings and then some; each flight cost more than the expendable rockets the system had replaced. The lesson widely drawn was that reuse does not pay, when the accurate lesson was that reuse whose refurbishment is expensive does not pay.

C The problem is fundamentally one of energy. A first stage separates travelling several times the speed of sound and must be slowed, turned, guided back through the atmosphere and landed upright, and every one of those operations requires propellant that could otherwise have lifted payload. The penalty is real and is paid on every flight: recovering a stage reduces the mass it can put into orbit by something like a third. The proposition only works if the saving from reusing the hardware exceeds the value of the payload given up, which is an arithmetic question rather than a matter of engineering ambition. It also explains why the heaviest payloads are still launched on vehicles that are not recovered: when a customer needs the full capability of the rocket, the propellant reserved for the return journey is not available, and the stage is expended without apology.

D Making it work required solving a control problem that had defeated earlier attempts. A returning stage is an unstable, nearly empty tube descending at high speed with its centre of mass at one end, and steering it is closer to balancing a broom on a fingertip than to flying an aircraft. The solution combined grid fins that deploy in the airstream, engines that can be relit and throttled deeply, and software fast enough to correct continuously. The landings that resulted, first on land and then on floating platforms, were achieved after a long sequence of public failures, which turned out to be an advantage: each destroyed vehicle produced data, and the programme's willingness to fail in view of the cameras compressed the development schedule considerably.

E The economics that followed have been transformed in a way that is easy to state and hard to overstate. Individual first stages have now flown many times each; the interval between flights has fallen from months to weeks; and the cost of putting a kilogram into orbit has dropped by something approaching an order of magnitude compared with the vehicles of the previous generation. Launch has changed from a scarce and rationed capability into something closer to a scheduled service, and the consequences for what people propose to do in space — constellations of thousands of satellites, telescopes too large to have been contemplated before — follow directly from that. Cheap launch has also changed how spacecraft are designed, since a satellite that can be replaced easily need not be engineered to survive fifteen years without maintenance, and the discipline of building for the worst case has relaxed accordingly.

F The remaining question is how far the principle extends. Recovering the second stage is much harder, because it returns from orbital velocity rather than from a fraction of it and therefore requires the full thermal protection that made the shuttle expensive. Vehicles now in development attempt full reuse of both stages, using steel rather than aluminium and accepting a heavier structure in exchange for tolerance of heat. Whether that succeeds is not yet known, and the honest position is that the first stage has been solved and the second has not.

G It is worth noticing what has not changed. The engineering that permits a stage to land is not new physics, and most of it was proposed decades ago; what changed was the availability of cheap computation, the willingness of an organisation to destroy prototypes in public, and an economic model in which the same company both built the rockets and needed a great many launches for its own purposes. The last of these may be the most important and is the least discussed. A manufacturer with an internal customer can fly often enough to make reuse pay, and flying often is the condition on which the entire argument for reuse depends.