Reusable launch systems

Reuse is not new. The Space Shuttle was reusable and made spaceflight more expensive. What changed in the 2010s was not the idea but the economics, and the difference is worth understanding precisely.

The recovery sequence

A propulsively landed booster separates at a few kilometres per second and then has to remove that velocity, survive re-entry and land.

Boostback burn

Flown only for return-to-launch-site profiles. Reverses the horizontal velocity so the booster flies back. Expensive in propellant, which is why it is used only for lighter payloads.

Entry burn

Fired against the direction of travel just before the dense atmosphere, cutting peak heating and dynamic pressure to what the airframe can take without a dedicated heat shield.

Aerodynamic control

Grid fins steer during the hypersonic and supersonic descent, holding the booster on its trajectory against wind.

Landing burn

A short, high-thrust burn timed so velocity reaches zero at the moment altitude does. There is no hover margin on a booster whose minimum thrust exceeds its empty weight.

That last point deserves emphasis. A nearly empty booster with a high-thrust engine cannot throttle down enough to hover; its thrust-to-weight ratio exceeds one even at minimum throttle. The landing burn is therefore a suicide burn, ignited at exactly the right instant so that the deceleration curve reaches zero velocity at zero altitude. Ignite early and the booster stops in mid-air and falls; ignite late and it does not stop.

What recovery costs in payload

Propellant reserved for boostback, entry and landing is propellant not spent accelerating the payload, and the recovery hardware is permanent dry mass.

The droneship exists precisely because of this gap. Landing downrange means no boostback burn, so the penalty drops from about 40% to about 20%, at the cost of operating a barge and a longer recovery cycle. Missions are assigned to profiles by payload mass, and the heaviest are still flown expendable.

The break-even analysis

Let $V$ be the cost of building the recoverable stage, $R$ the cost of recovering and refurbishing it, $F$ the fixed cost of the recovery infrastructure amortised per flight, and $N$ the number of flights per booster.

Expendable cost per flight is $V$. Reusable cost per flight is

Reuse pays when $C_{\text{reuse}} < V$, which requires

Two conclusions follow, and they are the whole story of why reuse succeeded recently and failed earlier.

Refurbishment cost is what matters, not the number of flights. If $R$ is 10% of $V$, the scheme is ahead after two flights. If $R$ approaches $V$, no value of $N$ ever helps. Reuse is a refurbishment problem, not a durability problem.

Flight rate has to be high enough to amortise the fixed cost. Droneships, recovery vessels, landing pads and inspection facilities cost money whether they are used once a year or fifty times. At low cadence $F$ dominates and reuse loses on its own.

Why the Shuttle failed the test

The Shuttle was designed for roughly $10$ to $20$M per flight at 50 flights per year. It achieved something between $450$M and $1.5$B per flight, depending on whether programme fixed costs are amortised, at four to five flights per year.

Both terms went wrong at once.

Refurbishment was the larger failure. The thermal protection system used roughly 24,000 individually shaped ceramic tiles, each requiring inspection and many requiring replacement after every flight. The main engines were removed and substantially overhauled between flights. The solid boosters were recovered from seawater, which meant disassembly, cleaning and recasting rather than refuelling. The orbiter was reusable in the sense that the same airframe flew again, but the labour to return it to flight status approached the cost of building new expendable hardware.

Flight rate then compounded it. A system with enormous fixed costs needs high cadence, and the Shuttle never achieved a fifth of its design rate, so every flight carried a large share of a standing army’s salary.

Why Falcon 9 passed it

No dedicated heat shield

The entry burn reduces heating enough that the aluminium-lithium structure and a modest ablative coat on the base survive. Removing the TPS removed the Shuttle largest refurbishment cost.

Kerolox with clean-running engines

Merlin was designed for repeated firing, and the recovery profile keeps the engines well inside their thermal limits.

Dry landing

Propulsive landing on land or a deck avoids the seawater immersion that made SRB recovery so expensive.

High cadence

A launch rate of a hundred or more per year spreads the fixed recovery infrastructure cost thinly.

Result

Individual boosters have flown more than twenty times, with minimum turnarounds measured in weeks.

The design decisions that made this work were made for reuse rather than retrofitted, and each one attacks $R$ rather than trying to make the hardware last longer. That is the correct target, and it is the lesson the Shuttle provides in negative.

Thermal protection

Anything returning from orbital velocity, rather than staging velocity, faces a much harder problem: roughly 30 times the kinetic energy per unit mass.

The Shuttle’s tile system is the cautionary example, and Starship’s is the current experiment in whether a tiled system can be made to work with maintenance costs low enough to matter.

The unsolved part

Booster reuse is a solved problem. Upper stage reuse is not, and it is where the remaining cost sits.

An upper stage reaches full orbital velocity, so recovering it means a full re-entry with the thermal protection that implies, plus the mass of that protection carried all the way to orbit where it directly displaces payload. The mass penalty falls on the stage that can least afford it.

Falcon 9 does not attempt it. Fairing recovery was pursued instead, first with net-equipped ships and now by retrieving them from the water and refurbishing, which recovers a few million dollars per flight from a component that is comparatively easy to recover.

Full reuse of both stages is the central bet of Starship, and whether it closes determines whether launch costs fall by another order of magnitude or settle near where they are now. See Launch economics.

References

• Jenkins, D. R. Space Shuttle: The History of the National Space Transportation System, 3rd ed., Specialty Press, 2001.
• Pielke, R. and Byerly, R. “Shuttle programme lifetime cost”, Nature, 472:38, 2011.
• Blackmore, L. “Autonomous Precision Landing of Space Rockets”, The Bridge, 46(4):15, National Academy of Engineering, 2016. On the landing guidance problem.
• Açıkmeşe, B. and Ploen, S. R. “Convex Programming Approach to Powered Descent Guidance for Mars Landing”, Journal of Guidance, Control and Dynamics, 30:1353, 2007.
• SpaceX Falcon User’s Guide, for published payload performance.