9. Gen 2 – Depots & Convoys
The root why of Generation Two was not ambition. It was mercy. Mercy for schedules, for crews, for engines, and for every poor planner who had learned to fear the word “window.”
We had proved the route (G0) and taught it to keep time (G1). Gen Two’s objective was simple to say and hard to build: turn a planetary alignment into an operating season — and make every ship depart with a net of friends.
♦
We did it with two moves that look modest on paper. First, we made the low-Earth depot into a utility instead of an experiment — a place where propellant behaved like inventory and time could be bought in small, civilized purchases. Second, we declared a doctrine that sounds like superstition and calculates like survival: no singletons. Ships would fly in convoys, designed to rescue and to be rescued, their geometry choreographed like something you could audit rather than admire.
Those two moves — plus a thickening layer of dust manners at the pads — changed the flavor of the whole corridor.
♦
LEO Depots
Let’s talk depots without the romance. The early sump in LEO, our G1 buffer, grew muscles and a spine. STN-DEP (LEO) matured from a cryo puddle into a LOX-first utility that could publish its truth in units that bosses and pilots both understood: days of departures at planned rate. Not tonnes, not tweets — days. LOX production (from orbiting water cracked on a sober schedule) gave us the big oxidizer clock; tankers shifted to CH₄-only sorties by default. It felt wrong the first week and inevitable the third. Oxygen became a public service; methane became a courier.
You could tell it had matured when arguments changed tense. In G1 we asked, “Do we have enough prop?” In G2 we asked, “How many days of Green do we want to buy?” The Green Zone around the Hohmann date broadened from ±60 to ±90 days, and we paid a published toll for it: +200–300 m/s across TMI and predicable DSMs, never taken from EDL reserves. Depot gauges stabilized into ZBO uptime ≥ 97–98%, with boil-off under high single digits per month even when Florida remembered how to rain. Slip authority stayed local (±7 days), but a depot that reported ≥45 crew-days and ≥35 cargo-days on hand made slips feel like accounting, not drama.
We funded the boring thing that saves careers: tanker overbuild. Idle tankers, an offense to the aesthetic age, turned out to be the antivenom for scrub cascades. We flew them anyway, year-round, because cadence is a habit, not an event. The result showed up on the graphs that never trend on networks: SUF (Synod Utilization Factor) climbed into the 0.8s, then flickered higher, while miss-to-launch shrank to a number you could lose under a long lunch.
The depot changed payload behavior, too. MIUs staged early, barcoded and smug. Plug-to-productivity for critical kits fell to hours because everything that needed to decide had already decided. Refueling choreography — once a ship-to-ship romance — became a published queue with approach lanes, collar reservations, and hold points that felt like air traffic from a kinder century.
Depots did not become hotels. We kept that boundary like a law. They stayed utilities — tank farms with gentle claws — and the crews who insisted on living there were carefully introduced to chairs elsewhere.
♦
Starship Convoys
Now the other half: convoys. You can frame them as courage wrapped in geometry, but I prefer the plainer truth: redundancy that fits in a sky.
A single ship is a haiku about luck. In G2 we stopped writing poetry and started publishing bulletins.
A convoy was three hulls with distinct roles in their hearts and shared abilities in their bones. Early on we called them by function rather than paint: CREW/HAB, TANK, MEDIC/SPARES. All three could host humans in shirtsleeves for a while; all three could mate collars without inventing standards; all three published their state of health in the same, unforgiving units: seconds, meters, kilograms.
Convoy geometry was the quiet art. Too tight and a plume turns a family into a headline; too loose and rescue becomes elegy. We flew along-track separations in the hundreds to low thousands of kilometers, typically tightening to sub-thousand after DSM-1. We budgeted for any-to-any rendezvous inside eight hours with ≤50 m/s while preserving each ship’s EDL reserve as untouchable. LEAD — the convoy’s conductor — published a navigation bulletin like scripture: DSM-n delta-v, windows, accept/reject bands, and the exact conditions for invoking the reserve plan. Pride and habit did the rest.
Rescue became a procedure rather than a prayer. A helium bottle lazy on one hull? The TANK cross-pressurized across a tethered collar while everyone pretended to be bored. A pump controller with opinions? The MEDIC slid a spare MIU into the rails as if it were a drawer and logged the serial with the gentleness of a librarian. When people needed to move, they didn’t go EVA with a camera and a wish; they rode a small pressurized shuttle module that treated relative motion like an instrument. We will take that craft apart in a later chapter; here I’ll simply record the consequence: Time-to-Stabilize for inflight anomalies fell into dozens of minutes, not hours, across the fleet.
The result of convoys was not the end of emergencies; it was the end of lonely emergencies. A ship could fail gracefully — capture to a hold orbit, publish its state, and be helped by friends inside geometry designed months earlier. In the language of auditors: common-mode faults were fenced; single-mode faults were forgiven.
♦
All of this would have degenerated into theatre without one more invention: Spacecraft Traffic Control. (STC) began as a sheepish spreadsheet and became a civics lesson taught in vectors.
Space is large; LEO is not. The refueling stacks we flew in G2, plus everyone else’s satellites, boosters, junk, and ambition, required something sterner than “look out.” We formalized phasing lanes and keep-outs around the depot shell. Approach corridors ran prograde at defined radii; hold boxes were numbered and boring; slot times respected tanker trains and the range. Autonomous collision-avoidance stopped being a novelty after the first three times it whispered “no” to a proud pilot and saved nothing more dramatic than the schedule.
Departure choreography tightened too. Pad-to-ascent windows overlapped less and aligned more. LEO rendezvous sequencing moved from art to math: no two ships flew a collar approach with relative motion that could cause an argument later, and the STC ledger refused to be charmed. Even roll flips for collars got their own lanes. Inflight, convoy lanes were notionally “painted” onto the porkchop, invisible but enforced by the bulletins and by the social penalty for ignoring geometry.
Mars side, the sky stayed generous, but the ground stopped being shy. Landing zone farms thickened; each pad had approach corridors, plume priority, and berms that did less to look brave than to catch rocks. Pad density rose until you could watch guidance solutions negotiating with neighbors like polite drivers at a four-way stop. We built plume conflict maps and obeyed them. STC-Mars learned to name winds like sailors and to treat pad opening/closing as weather, not shame.
A note to romantics: none of this killed wonder. It replaced improvisation with choreography so the wonder could survive on a workday.
♦
Metrics, because we kept score to stay honest. With depots healthy and convoys doctrinal:
Green Zone adherence stopped wobbling; departures inside ±90 days accounted for the overwhelming majority of TMI, with DSM budgets meeting plan >95% of the time.
SUF plateaued above 0.85 for crewed waves and approached 0.9 for cargo, because the depot and tanker train broke the causal chain between one scrub and a synodic tragedy.
Refuel collar utilization in LEO ran at 70–80% of plan with approach aborts rare enough to investigate each one with coffee and shame.
Pad throughput at the busy Martian fields doubled without raising anyone’s heart rate; tile wear per landing trended down; filter change intervals became a calendar note, not a story.
Rescue index — spare seats available × intercept Δv available × TANK prop after DSM-2 — became a published number. If it fell below the band, the convoy didn’t carry crew. Programs ended careers over that, and good.
On cost, the truth is flatter than legends. Hardware efficiency crept; $/kg to TMI fell mostly because we stopped wasting propellant, people, and windows. Convoys reduced mission loss variance, which is how accountants sleep. Depots converted weather into inventory. Together they turned a spiky campaign into a predictable service.
♦
Choreography lived on the ground as much as in the sky. The pads adopted check-in/checkout rituals that would bore a poet and feed a civilization. Dirty vestibules stayed filthy by design. MIU rails made install-days sing. EDL kits (shoes, skirts, backshells you could use twice if you treated them nicely) formed the beginning of an EDL Library that let cargo landings stop destroying their own future. We pre-positioned backup kits at the big fields; when a hull rolled a shoulder after hypersonic pride, the pad chief shrugged back and swapped the skirt.
I remember one arrival wave that reads like parable. The convoy came in hot-and-correct, DSMs spot-on. LEAD called the reserve burn off with a voice like an accountant closing books. Capture burned gently into LMO, where nothing waited because G3 hadn’t been born yet — no tugs in this era, just honest ships doing honest delta-v work. On the surface, Pad North declared yellow after a wind-shift and closed its corridor. STC-Mars pivoted traffic without a speech. Pad East took the CREW/HAB. The MEDIC held and then dropped to Pad Far fifteen minutes later. The TANK stayed upstairs, gave two friends a polite sip, and went to bed. No hero shot made it to the public reels. The only line anyone framed was T-to-Stable: 22 minutes for a balky pump downstairs, closed by a spare MIU we used so often it had a nickname.
That’s G2 in a glove: the sky wide, the ground tight, the system forgiving.
♦
If Generation One made cadence believable, Generation Two made it durable. We took the smallest real depot we could stand and taught it to act like a utility. We took three ships and taught them to behave like one organism. We accepted the civil service work of STC so pilots could keep their poetry for the windows that mattered. And we thickened our dust discipline until pads became polite factories rather than expensive anecdotes.
There is a line I wrote in a margin during those years and found again a century later in a museum drawer, scratched in graphite hard enough to bruise the paper: “Spend Δv to buy days. Spend ships to buy certainty. Spend pride to buy procedures.” That was the exchange rate of Gen Two. We paid it gladly.
The interplanetary corridor didn’t become beautiful in G2. It became kind — to schedules, to lungs, to luck. From that kindness, the next generation could afford to specialize: capture split from landing, trucks took over the down-leg, and Mars orbit learned to be busy. But that is another chapter. For now, remember the feel of this one: depots humming, convoys talking in numbers, and pads that look bored in the best possible way.