The Ceres Mass Driver

Why, given the elevator

The Launcher Outthrows the Elevator Sixfold and Arrives Decades Earlier

Both get cargo off Ceres without propellant. They are not competing for the same job.

The Ceres elevator beats the launcher on energy — about 0.015 kWh/kg against 0.037 — and it is gentle enough for people and instruments. On the two things a bulk exporter actually cares about, tonnage per day and when the thing starts working, the launcher wins outright.

Throughput. A cable moves what its tension budget allows at the speed a climber travels. A 10-tonne-capacity cable running climbers at 200 km/h over the 1,032 km to the release point delivers roughly 46 tonnes a day, and that cable costs around 120 tonnes of native wire. A single track firing one-tonne slugs every five minutes delivers 288 tonnes a day on 443 kilowatts of average power. Six times the tonnage, and the track is a structure rather than a hanging load. Matching the launcher with a cable means a 62-tonne-capacity elevator, which is something like 750 tonnes of drawn wire.

Sequence. The elevator’s cable is the foundry’s first large product, made on site from local nickel-iron. That means the foundry has to be running, and shipping, before the cable exists. A launcher is a graded cut, a set of coils and a power supply, and it works at whatever length has been built — a shorter track simply means higher acceleration. The launcher is what bootstraps the elevator, not the other way around.

Redundancy. A single cable is the one structure in this regime whose failure disconnects a settlement from its own body, and no redundancy design has been sized for it. A launcher on the surface is the second route off, independent in every component.

Elevator throughput is capacity multiplied by climber speed divided by cable length, which gives the same answer as counting climbers. Both figures scale with the money spent; the comparison holds at the entry scale where the decision is actually made.

So the split is the obvious one and it is the same split the Moon arrives at from the other direction: bulk water, metal and regolith go out through the launcher; people, instruments and anything fragile go up the cable. Modeled

Track length and g

610 Meters at 30 g, or 1.36 Kilometers at 10 g

Track length and acceleration trade directly, and both are small because the target speed is small.

An electromagnetic launcher is a straight run of coils that accelerates a carriage, releases the payload at the muzzle and recovers the carriage. Length follows from constant acceleration:

d = v² / 2a  ·  t = v / a

At Ceres, v is 516 m/s. That is the number that makes everything else small — a rifle round leaves a barrel faster, and naval railgun tests have reached five times it.

The 610-meter figure carried elsewhere on this site is the 30 g case sized to 599 m/s rather than to bare escape — escape plus about 83 m/s of departure margin, which any real trajectory needs. Bare escape at 30 g is 452 meters. Both are correct for what they describe.

Two things that complicate a launcher on Earth or the Moon do not arise here. There is no atmosphere, so no aerodynamic heating, no drag loss and no need for a muzzle shroud. And the surface is not a problem to build on at this length — a few hundred meters of graded track on a body whose largest relief is crater walls.

One constraint does bind. A single impulse from the surface of an airless body cannot produce a bound orbit: the trajectory comes back to the launch radius. A payload is therefore either thrown clear to escape, or thrown on a ballistic arc and circularized by a small kick motor near apoapsis. For an exporter that is fine, because everything the belt ships is leaving anyway.

What a shot costs

133 Kilojoules per Kilogram, and Rotation Pays a Third of It

Launching eastward at the equator collects 93 m/s for free.

Ideal launch energy is kinetic energy at the muzzle:

½ × 516² = 133,128 J/kg = 0.0370 kWh/kg

Ceres turns once every 9.074 hours, so a point on the equator is already moving at 2πR/T ≈ 93 m/s. A payload launched eastward starts with that speed, and the track only has to supply the remaining 423 m/s:

½ × 423² = 89,465 J/kg = 0.0249 kWh/kg — a 33% saving

Because energy goes as the square of velocity, a 18% head start in speed is a third of the energy. That is a large and free discount, and it is the same term the elevator collects intrinsically through the rotating-frame potential. The 0.037 figure carried elsewhere on this site does not credit it, so it is a conservative ceiling rather than a working number. The Moon, turning once in 27.3 days, gets 4.6 m/s from the same effect and may as well get nothing.

Real launcher efficiency is the other adjustment and it runs the other way. Coilgun and railgun efficiencies from muzzle energy against stored energy land in the 20 to 50% range on terrestrial hardware, so a delivered 0.025 kWh/kg becomes 0.05 to 0.12 drawn from the plant. That is still small: throwing a tonne of water off Ceres costs about as much electricity as boiling a few dozen kettles. Modeled

Pulsed power

The Hard Problem Is 152 Megawatts for Two Seconds

Average power is trivial. Pulse power is two orders of magnitude higher, and that is what the design is really about.

A one-tonne slug taken to 516 m/s absorbs 133 megajoules. At 30 g the whole event lasts 1.8 seconds, so the track draws

133 MJ ÷ 1.8 s = 76 MW mean over the pulse
and because P = Fv rises linearly under constant force,
152 MW instantaneous at the muzzle

against a belt industrial plant sized in single-digit megawatts. No reactor delivers that, and none should have to. The energy per shot is small; it simply has to arrive very quickly. The launcher therefore runs on stored energy — capacitor banks, or flywheels, or superconducting magnetic storage — charged steadily between shots and dumped in two seconds. A mass driver is a pulsed-power problem wearing a launch costume, and the pulsed-power hardware, not the track, is what sets its mass and its failure modes.

Ideal energy, no efficiency applied, and no credit for the 93 m/s of rotation. Pulse power depends only on acceleration; charging draw and throughput depend only on cadence. The two are independent knobs, which is why a modest plant can run a launcher that momentarily outdraws it two-hundredfold.

Two secondary constraints follow from the same pulse. The carriage has to be stopped and returned after every shot, which is a regenerative braking problem that gives some of the energy back, and the track has to shed resistive heating with nothing but radiators — see the radiator arithmetic. At one shot every five minutes neither is severe. At one a minute both become design drivers.

Against the Moon

Twenty Times Cheaper Than the Lunar Launcher, and No Catcher Problem

The Moon needs 2.38 km/s. Energy goes as the square, so the gap is not a factor of five.

The mass driver concept belongs to the Moon, where Gerard O’Neill proposed it in the 1970s and where it has been studied ever since — the full treatment is on The Lunar Mass Driver. What matters here is the delta, and it is large.

Ideal kinetic energy at bare escape, which is the only basis on which the four are comparable. On that basis Ceres is 21 times cheaper than the Moon, and crediting rotation at both ends widens it to about 32. The source analysis records 17.6× like-for-like against a specific lunar delivery trajectory rather than bare escape; the earlier 24× figure compared the Moon at 2.52 km/s against Ceres at bare escape and is retired. Mars is listed to show why nobody proposes this there: it has an atmosphere as well as the velocity.

The second difference is the one that decides whether the thing is buildable. A lunar launcher throwing at 2.38 km/s needs a catcher that can absorb 2.38 km/s of relative velocity, and building that catcher is a harder engineering problem than building the gun — it is the unsolved half of every lunar mass-driver architecture. At Ceres the muzzle velocity is a fifth of that and the kinetic energy a twenty-first. A receiving station takes the payload at a few hundred meters per second, which is a docking problem rather than an impact problem, and a payload thrown slightly wide simply coasts for recovery instead of destroying something. Modeled

What it can throw

Anything That Does Not Mind 30 g, Which Is Most of What the Belt Exports

Water, metal and ore do not care. People, electronics and optics do.

Thirty gravities for two seconds is survivable for a great deal and fatal to some things. Frozen water in a shaped canister, cast nickel-iron billets, sintered regolith blocks, bulk chemical feedstock and packaged consumables all take it without special provision — artillery shells routinely endure thousands of g and contain working electronics. What does not take it is anything with a free liquid surface, a large unsupported optic, a deployed structure, or a person.

That maps almost exactly onto what the belt actually sells. The water argument concludes that the belt exports water as water rather than as propellant, and water as ice in a canister is the ideal slug: dense, inert, unpressurized, indifferent to acceleration and needing no thermal control at 160 K. Throwing a kilogram of that water off Ceres costs about 6% of the energy spent liberating it from the ground and under 1% of the energy it would take to split it.

The remaining question is packaging. Every slug needs a structural shell that survives the track, carries whatever guidance the trajectory needs, and is either recovered or written off. At 288 tonnes a day that is a lot of shells, and an operation that has to import them has no business running a launcher. The launcher and the foundry are the same decision: cast the canisters from local nickel-iron and the loop closes; import them and the economics collapse. See Nickel-Iron & Carbonyl Metallurgy. Modeled

State of the art

Electromagnetic Launch Is Flying Hardware, at Velocities Far Above What Ceres Needs

Nothing on this page waits on a physics result.

The US Navy’s railgun program reached muzzle velocities around 2,500 m/s in ground testing before it was shelved in 2021, roughly five times Ceres escape. The Navy’s Electromagnetic Aircraft Launch System has been catapulting aircraft off the USS Gerald R. Ford since 2017 — a linear induction motor that accelerates tens of tonnes to about 70 m/s, which is the mass regime a Ceres launcher needs at a seventh of the velocity. Maglev trains have run commercially on linear motors for decades.

On the space side, SpinLaunch has been testing a kinetic launcher in New Mexico since 2021 — centrifugal rather than linear, and aimed at the far harder problem of launching through an atmosphere from Earth’s gravity well. Its existence matters here mainly as evidence that investors will fund propellantless launch. Gerard O’Neill’s group built working mass-driver demonstrators at Princeton and MIT between 1976 and 1980, reaching accelerations above 500 g on small payloads.

What has never been done is operating any of this in vacuum, at cryogenic ambient, on a rubble surface, unattended, for years, with local repair. That is the belt’s standing problem rather than a launcher problem, and it is the reason the 30.8-month spares queue decides the design. Observed

Record

Provenance and Data Sources