33. Starship Fleet Armada

When people remember the founding decades on Mars, they picture the pads flaring and the dust curling around silver hulls. 

I remember the drums behind the music: the fleet math. Each landing implied a braid of assets — the Mars-capable hull that touched down, the tankers that climbed and reclimbed to fill it in LEO, the reflights that kept the metal honest, and the quiet replacement rate that said: machines have lifetimes, plans need spares, and Mars punishes wishful thinking. 

This chapter is that braid, laid out from Settlement Stage S0 through S5 — about the first ten thousand Martians — before big Spacetrain Cyclers took over the migration long-haul and before Phobos felt like a port.

♦

The rule of thumb was plain enough to fit on the back of a glove. A canonical early superlift starship needed on the order of 8–12 tanker sorties to leave LEO full for a Hohmann-class Mars throw; we ran the average at roughly 10–12 per departure in the S0–S2 era, tightening to 9–10 by S3 and ~8–9 in S4–S5 as depot choreography and LOX-first top-offs improved. The tables below carry the real counts stage by stage, but carry that picture in your head: for every crew or cargo hull that pushed for Mars, there was a picket fence of tankers launching, returning, launching again, until the transfer burn lit clean.

The second rule was cultural rather than numeric: returns are part of the mission. We set explicit “% ships taking off” policies per stage and then held ourselves to them. In the thin early Hohmann Cycles we accepted that a few crew starships would remain for a season as static assets — power buses, spare shops, donor airframes — but the bias was always to lift crew rated hulls home and fly them again. A returned ship is a teacher. A stranded one is a confession.

♦

S0, the Crew Sorties, wore this logic like a learning bruise. Ten crew landings to deliver fifty-seven people, nine lift-offs (≈90%), and fifteen cargo hulls total for ~540 tonnes useful in stage (excludes the Stage -1 Prep Bootstrap kits). LEO Tanker flights for the campaign: ~263, with tankers reused on average ten times, implying a distinct tanker cohort of about twenty-seven metal birds to keep the buses fueled. Those are not poster numbers, but they tell you the truth of early cadence: ten crew ships is only the visible crest; beneath it, a small airline of tankers is beating the air to make rendezvous after rendezvous while guidance learns to be boring and plumbing learns not to surprise you at −40 °C. This of course doesn’t include the Mars surface refueling flights need to get people home.

S1, the Base Camp, sharpened the pencil. Twelve planned crew landings, adjusted, and ~10 returns (≈91%); roughly forty-three cargo hulls hauling ~1,710 tonnes useful; tanker sorties climbing to ~550 for the stage as the fleet learned to stage tankers and hold propellant at altitude, with about twenty reuses per tanker on average — meaning only twenty-eight distinct LEO tanker hulls were required to sustain S1’s tempo. The raw count matters less than the reuse: twenty flights per tanker in an era of crude ground ops is not a marketing line, it is a discipline line that says your turnarounds, inspections, and pad thermals are behaving. Again we haven’t included here the substantial number of Mars surface rocket refueling flights we had to do in this stage as ISRU had not scaled yet.

By S2, the Tribe, the fleet began to hum. Eighteen crew landings, sixteen returns (≈92%), and ~121 cargo ships for ~5,630 tonnes useful — a number that reads like ambition until you do the multiplication: each of those cargo hulls was preceded by a ladder of tanker climbs. LEO Tanker flights for the stage sat around ~1,330. Average reuses per tanker hit ~40. That allowed an effective tanker pool of thirty-four active airframes to support the entire stage. We used to write “S2: 34 tankers can feed a tribe” on the whiteboard as a joke and then point at the tally when someone proposed starting a second landing zone too early. While ISRU propellant was contributing it was still not much more than a trickle, so to get the complete picture we would also need to add the significant Mars refuelling flights to these LEO tanker numbers.

S3, the Village, is the first time the fleet felt large. Thirty-six crew landings, thirty-three returns (≈93%), cargo landings on the order of ~282 (adjusted) to bring ~14,620 tonnes useful, and tanker flights for the stage spiking near ~2,925. It sounds exhausting until you see the reuse line: ~60 flights per LEO tanker on average. That drove the distinct tanker requirement to under fifty hulls for the entire S3 campaign. We knew by then what we were: not a stunt show, we were an “airline.”

S4, the Town, is where the numbers begin to look like the floor of a stock exchange. Eighty crew landings, seventy-five returns (≈94%), and ~1,799 cargo landings to deliver ~117,500 tonnes useful in the stage. LEO Tanker sorties: ~16,346. Average reuses per tanker: ~80. That allowed us to run the entire town-scale supply with a tanker fleet on the order of two hundred five hulls. You can hear the policy beat under that: reuse makes fleet size tractable; tractable fleet size makes schedules civil; civil schedules make humans behave. We put those three lines in the pad ops handbook and fought any fashion that tried to break them.

S5, the first designated (small) City, pushed that discipline to its limit without breaking it. Two hundred thirty crew landings, two hundred eighteen returns (≈95%); ~2,672 cargo landings to deliver ~220,000 tonnes useful mass; tanker sorties of roughly ~26,760; and reuses per tanker around ~100. The distinct tanker cohort needed was ~268 hulls — a startlingly small number for a campaign that massive, and a testament to a maintenance culture that had learned the art of “boring heroism”: engines that relight like lighters, tiles that seat without drama, valves that close when told. The total starship fleet on the books by S5, counting Mars-capable crew/cargo hulls and the tanker cohort, sat just north of three thousand metal birds; the adjusted, truly active fleet deployed at any one moment ran closer to three thousand even, as retirements, overhauls, and refits cycled through.

♦

There is a temptation to think of “fleet size” as a headcount. It is a cadence count. Earth take-offs per stage tell the story from the ground up: ~288 for S0, ~605 for S1, ~1,470 for S2, ~3,250 for S3, ~18,269 for S4, and ~30,105 for S5. In each of those piles, crew and cargo Mars-bound departures were the minority. The majority were the tankers — quiet, relentless, and sublime in their ordinariness. Every time an ops team tried to hide a shaky assumption under optimism, the Earth take-off ledger walked it back into the light.

Returns mattered just as much. A cargo ship return rate of ~6% in S0 (one bird back by policy), ~8% in S1, ~10% in S2, ~20% in S3, ~30% in S4, and ~40% in S5 is not a measure of failure; it is a measure of confidence. When returns went up, reflights followed. By S5 we were seeing more than a thousand cargo ship reflights in the stage, and crew ship reflights creeping into the hundreds. The “new ships needed” line fell in response — a program that doesn’t need to buy its way out of attrition is a program that can afford to build smarter.

What did this feel like on a pad at Port Harmonia in S4? I remember walking a night turn with my thermal drone and counting the rhythm by ear: one dust plume fading at Pad C, another footprint appearing on the horizon, a third hull from Phobos skimming overhead to drop into terminal. The spreadsheet says 1,843 cargo landings in S4, 75 crew lift-offs, 16,346 tanker sorties. My memory says it sounded like rain on a metal roof.

♦

Two tables, then, to keep the bones straight. The first is the fleet’s visible motion — crew and cargo starships and their returns. The second is the invisible heartbeat — the tanker ladder and the fleet size that made it survivable.

Fleet movement and returns (S0 → S5)

(Excluding S0-S3 Mars return refueling flights)

*In S2–S3 the “>100%” notation on crew take-offs vs landings reflects reflights within the stage window (lift, return, lift again) rather than time-shifted accounting. We learned to keep that footnote after an unhelpful audit tried to call arithmetic a scandal.

LEO Tankers, reuse, and fleet size (S0 → S5)

(Excluding S0-S3 Mars return refueling flights)

The “Starship fleet (all hulls)” column combined crew/cargo Mars-capable ships and the tanker cohort carried on the books; the “adjusted active fleet” is what an ops chief could actually schedule in any given season once refits, overhauls, and long-cycle inspections were removed from the orchestra pit.

♦

Two closing observations that saved us from ourselves.First, LEO tankers are not a side quest. They are the mission’s multiplier. Every point you add to average tanker reuse shrinks the distinct tanker cohort and frees shipyard capacity for Mars-capable hulls. The difference between fifty and a hundred reuses per tanker is the difference between a village served by a heroic stunt and a town served by a timetable.

Second, fleet size is only meaningful when paired with boring. A 80-90% crew return policy in S5 is not “bold”; it is a confession that the engineering is mature and the culture has learned to say no on the days when Mars says no. We taught young ops chiefs to put their ear to the pad deck and listen for that kind of silence. It sounds like a schedule that will still be running in ten years.

The next chapters will take this fleet and bend it into convoys, depots, and eventually cyclers. Before we do, hold this simpler picture: the first ten thousand Martians were carried by a few hundred Mars-capable starships, a few hundred more tankers, and a maintenance religion that turned reuse from aspiration into arithmetic. The posters show the plumes. Civilization was the drum.