Propellant Depots

The gear-ratio argument

Why Depots Change the Equation: The Gear Ratio

A kilogram of propellant at EML1 is worth several kilograms launched from Earth. That multiplier is the depot’s business case.

The Tsiolkovsky rocket equation is exponential. Every kilogram of propellant a vehicle carries from the surface must itself be lifted, which requires more propellant, which requires more lift. For a high-Δv mission — lunar landing, Mars transit, deep-space departure — the fraction of launch mass consumed by propellant climbs steeply.

A depot breaks the exponent. A vehicle launched nearly empty can carry structure, cargo, or crew where propellant mass would otherwise sit, then fill its tanks in orbit from a separately delivered supply. The total propellant consumed is the same; the total mass launched from Earth per unit of delivered payload drops. That ratio — kilograms launched per kilogram delivered — is the gear ratio, and the depot’s contribution is measured by how much it shrinks.

For Starship HLS, the arithmetic is direct. The vehicle needs a full propellant load in LEO to reach the lunar surface and return to NRHO. SpaceX’s current architecture calls for a dedicated Depot Starship in LEO, filled by roughly 5–16 tanker flights — the range depending on Starship block, tanker capacity, and boil-off assumptions — before the HLS vehicle arrives to receive its load. Without the depot, the mission does not happen. The depot is not an optimization; it is the architecture.

Storable vs cryogenic

Two Different Depot Problems: Storable and Cryogenic

One problem is already being solved. The other is harder by an order of magnitude.

Storable propellants — hydrazine, mono-methyl hydrazine, nitrogen tetroxide — remain liquid at room temperature and orbital thermal conditions. Transferring them between vehicles is primarily a plumbing and rendezvous problem: dock, pressurize, open a valve, pump. Storable refueling extends the life of GEO communications satellites, military assets, and science missions whose station-keeping budgets are finite. This problem is being solved now.

Cryogenic propellants — liquid oxygen (LOX, boiling point −183 °C), liquid hydrogen (LH2, −253 °C), liquid methane (LCH4, −161 °C) — are the high-performance fuels that power every vehicle leaving LEO. They boil. In microgravity, the liquid does not settle to the bottom of the tank; it clings to walls, floats as bubbles, and mixes with ullage gas. Transferring cryogenics between vehicles in orbit requires solving settling, gauging, insulation, venting, and pump design in an environment where none of the familiar gravity-assisted shortcuts work.

No one has transferred cryogenic propellant between two vehicles in orbit. The first demonstrations are underway in 2026.

The cryogenic problem

Boil-Off, Settling, and Gauging: Why Cryogenic Depots Are Hard

Gravity does three things for a propellant tank on Earth. In orbit, all three are gone.

Boil-off is the headline problem. Cryogenic liquids absorb heat from the spacecraft structure, from solar flux, and from Earth’s infrared emission. The absorbed heat vaporizes propellant. On Earth, the vapor rises to the top of the tank and is vented; in orbit, it can form anywhere. Multi-layer insulation, sun shields, and zero-boil-off active cooling (cryocoolers) can reduce losses, but none eliminates them entirely without significant mass and power. SpaceX’s 2026 test plan includes an extended-loiter demonstration — a single Starship sitting in orbit to measure insulation and pressure performance over days to weeks.

Settling is the operational problem. To pump liquid rather than vapor, the liquid must be pushed to the bottom of the tank. On Earth, gravity does this. In orbit, a small settling burn — firing thrusters to create a fraction of a g — pushes the liquid toward the drain. Every settling burn costs propellant and interrupts the depot’s quiescent state.

Gauging is the measurement problem. On Earth, you weigh the tank or read a sight glass. In microgravity, the liquid surface has no fixed location. Pressure-volume-temperature methods, capacitive probes, and mass-spring oscillation can estimate remaining propellant, but each has error bands that widen as the tank empties. LOXSAT will test several gauging technologies in its nine-month LEO demonstration.

Who is building what

Depot and Refueling Programs: 2026 Status

Three distinct depot architectures are converging on flight in the same year.

SpaceX — Starship Propellant Depot (LOX/LCH4, LEO)

SpaceX is building a dedicated Depot Starship — a modified upper stage optimized for propellant storage rather than payload delivery. The depot launches to LEO, deploys insulation and sun shielding, then receives propellant from a series of Tanker Starship flights. Once full, it transfers its load to an HLS or cargo Starship bound for the Moon or Mars. The 2026 milestones are an extended-loiter test and a ship-to-ship cryogenic propellant transfer demonstration between two Starship vehicles in orbit. If both succeed, the path to Artemis IV’s HLS fueling sequence opens. Block 3 tankers, stretched to maximize propellant volume, are in production at Starbase.

Eta Space — LOXSAT (LOX, LEO)

NASA’s LOXSAT cryogenic fluid management demonstration — built by Eta Space, launched on a Rocket Lab Electron — has a launch window opening no earlier than 17 July 2026. The nine-month mission will test 11 cryogenic fluid management technologies: boil-off reduction, propellant transfer, tank pressurization, liquid gauging, and an automated cryogenic coupling built by Altius Space Machines (Voyager Space). LOXSAT’s data feeds directly into Eta Space’s Cryo-Dock, a planned full-scale commercial cryogenic depot targeting LEO deployment by 2030.

Astroscale + Orbit Fab — APS-R Refueler (Hydrazine, GEO)

Astroscale’s 300 kg APS-R Refueler is manifested for summer 2026, targeting GEO. The spacecraft will rendezvous with a U.S. Space Force Tetra-5 satellite, dock using Orbit Fab’s RAFTI interface, and transfer roughly 30 kg of hydrazine. It will then fly to an Orbit Fab fuel depot — also launching in 2026 — to refill its own tanks before servicing a second DOD satellite. If successful, this will be the first on-orbit refueling of a DOD asset and the first operational demonstration of the depot-to-servicer-to-client supply chain.

ULA — Simple Depot Concept (LH2/LOX)

ULA’s Simple Depot concept, first proposed in 2010, adapts flight-proven Centaur upper-stage cryogenic technology. The depot launches empty on a single Vulcan Centaur, deploys insulation in orbit, and receives propellant deliveries. ULA has targeted 95% fuel retention over 60 days and estimated a 10-tonne prototype could be built for roughly $1.2 billion. No firm launch date has been announced.

ESA — ARGON (EML2, 2030 target)

ESA approved its ARGON program in December 2022 with €220 million in funding, targeting a 20-tonne propellant depot at EML2 by 2030 to support lunar lander refueling. This would be the first depot positioned at a libration point rather than in LEO or GEO.

Program reference

Depot and Refueling Programs at a Glance

Three propellant classes, four orbit regimes, five programs.

The interface standard

RAFTI and the Gas-Pump-Nozzle Problem

A depot without a standard interface is a gas station that fits only one car.

Orbit Fab’s Rapidly Attachable Fluid Transfer Interface (RAFTI) is the closest thing the industry has to a universal refueling port. RAFTI has been flight-qualified and adopted by the U.S. Space Force — at least three Tetra-5 satellites carry the interface. It is designed for repeated mate/de-mate cycles and supports hydrazine transfer today, with xenon transfer under commercial agreement for Astroscale’s LEXI servicer fleet.

RAFTI solves the storable-propellant interface. The cryogenic interface is a harder problem — cryogenic fluid transfer disconnects must seal against vacuum at temperatures below −160 °C, resist thermal cycling, and handle two-phase flow. Altius Space Machines (Voyager Space) is testing a cryogenic coupling on the LOXSAT mission.

Without a standard interface, every depot-client pair requires a bespoke connector. That kills the network effect. The gas station works because every car has the same filler neck.

Where to put the depot

Depot Placement: Junctions for Throughput, Warehouses for Reserves

The warehouse-and-junction logic applies directly.

An active depot — high throughput, frequent docking, tanker arrivals, client departures — belongs at an unstable junction. EML1 is the natural site: it has Earth visibility for communications, near-side lunar access for incoming mass-driver deliveries, and departure manifolds for outbound vehicles. SpaceX’s LEO depot is a special case — it sits in the departure orbit of its only client (Starship HLS) rather than at a system-wide junction.

A strategic reserve — propellant stored for months or years against future need — belongs at a stable warehouse. DRO is the natural site: near-zero station-keeping, no urgent departure manifold, and a thermal environment that can be optimized for long-duration cryogenic storage.

ESA’s ARGON at EML2 sits at a junction rather than a warehouse — consistent with its role as a refueling node for departing lunar landers.

The noble-gas gap

The Noble-Gas Gap: Electric Propulsion’s Missing Supply Chain

The depot can store and transfer LOX, LH2, and LCH4. It cannot produce the propellant that electric tugs actually burn.

Hall-effect and gridded-ion thrusters — the workhorses of solar-electric propulsion — use xenon, krypton, or argon. These noble gases are extracted from Earth’s atmosphere by cryogenic air-separation plants. No off-Earth production source exists for any of them.

A cislunar depot can store and transfer noble-gas propellant launched from Earth. But it cannot manufacture a resupply. Every kilogram of xenon at EML1 was lifted from Earth’s surface at full gear-ratio cost. That makes electric-propulsion logistics in cislunar space a terrestrial supply chain with an orbital warehouse — not a closed loop.

The Moon can plausibly supply oxygen and hydrogen via electrolysis of polar water ice. Using LOX/LH2 chemically for the same tug round trip that an electric tug would fly on xenon drives the propellant fraction toward three-quarters of departure mass. The noble-gas gap is not merely an inconvenience. It is a structural constraint on which propulsion architectures can scale beyond terrestrial supply.

See The Tug with No Propellant in the corridors page for the full arithmetic.

Lunar water as propellant

Closing the Loop: Lunar Water Electrolysis

If the Moon has accessible water ice, it has LOX and LH2. That closes the propellant loop for chemical engines — and only for chemical engines.

Electrolysis of water produces oxygen and hydrogen in an 8:1 mass ratio. A depot at EML1 receiving lunar-sourced LOX and LH2 could refuel chemical vehicles — Starship-class landers, Orion service modules, or dedicated cislunar tugs — without importing propellant from Earth. The gear ratio for those vehicles drops toward one.

The loop closes only for chemical propulsion. Electric thrusters still need their noble gases. Nuclear-thermal engines use hydrogen as reaction mass (the loop partially closes). Solar-thermal systems could use water directly as propellant at lower specific impulse.

The open question is whether lunar polar water ice exists in the quantities, concentrations, and accessibility that an industrial-scale electrolysis operation requires. Orbital measurements from LCROSS, LRO, and Chandrayaan-1 confirm the presence of water ice in permanently shadowed craters. Ground-truth confirmation at mining-relevant concentrations remains the single most important near-term measurement for cislunar industrial architecture.