19. Settlement Structures
We discovered early that Mars did not need a thousand kinds of building. It needed two that never ran out of uses.
One was a street you could live in — a pressurized link wide enough for people, carts, pipes, and light industry to pass in shirtsleeves. The other was a room big enough to matter — a hall you could re-fit from housing to workshops to green space without touching the shell. Call them what the frontline engineers called them: barrels and barns.
Two mega-SKUs, endlessly recombined, did what catalogs couldn’t: they let the city grow without new theology. AI simulations — thousands of them, cruel and patient — kept spitting out the same answer: minimize the kit, maximize the copies. Everything else is garnish.
The names stuck because they were accurate. A barrel is a straight-run pressure “street.” A link. A barn is the big-span “room.” A hall. With those, plus a tiny family of helpers — T and X junctions, bulkheads, shim lengths, clean–dirty locks — you could build almost anything that needed people inside. Planetary engineering lego sets.
And more importantly, you could keep building it tomorrow.
♦
The first barrels were not designed so much as extracted.
We flew uncrewed donor starships — purpose-built to die well — packed tight with a flat-pack of structure, ports, liner reels, and a single, stubborn fastener family. After landing, the hull was tipped-over, trimmed, and towed. Thirty-meter shell pieces slid from the belly like staves from a cask. We lined them on sleepers, clamped standard knees between staves, and snapped the liner flanges to a pressure ring we could service in gloves.
Yes, the liner was pre-installed on Earth in the very first runs. Yes, it offended the purists. It also pressurized in one afternoon, which shut the argument down.
We placed those early links in shallow cuts or sometimes no cuts at all, because our earthmoving throughput was a joke the first seasons. Hold-downs and ballast cassettes did the ugly work that better diggers later replaced. We borrowed shielding where we could — water in casks, sacks of regolith, panels laid like scales — and we promised to come back for a proper berm when the fleet caught up. Of course we laid them down tile-side up, as now those lovely steel and tile shells were born again as ready-made roofs. Did that mean we didn’t need regolith on top? Unfortunately not. The porous silica ceramic and stainless steel only contribute <5% of the needed Galactic Cosmic Ray shielding, but it did make a fine roof for 1-2m of regolith.
It wasn’t elegant. It was fast. The old cores of Port Harmonia still carry those first barrels. They creak if you listen, like all good ships should.
♦
Barrels evolved from cut-and-shut to purpose-made as our machines learned the ground.
The cross-section that won in later bloc designs was a haunched horseshoe barrel — about 12 m clear width, ~9 m to the crown in the mature types — double-layered: a light imported steel rib cage inside, and outside it regolith panels (geopolymer/sulfur-concrete with basalt fiber) doing the heavy compression and the radiation work. By mass, they were ≈86% local in later decades; the shipped bits were rings, ties, knees, ports, and the smart gaskets. Centuries on, they lean even further local by weight — Mars teaching itself to make Mars.
We stopped insisting on perfect plan-curves. Straight chords joined at small miters carried hoop forces cleanly through welded knees, made interiors simple to fit out, and — most important — let us swap a 30 m piece for a better one without touching the rest. The pressure liner ran continuous across the joints like a tendon.
Why did this beat Earth’s ships and skyscrapers on building kg per usable m²?
Because our burial technique makes the soil your structure; overburden counters uplift, and pressure loads become behavior you design for, not fear. Because low and wide wastes nothing on wind, thermal load, façades, or elevator cores to nowhere. Because pressure-efficient geometry spends steel on rings and ties, not vanity. Because the quiet medium outside — dust and dirt — favors compression and a cautious membrane, not a heroic hull.
On Mars, as on Earth, pressure is your friend if you treat it like prestress. A bicycle tube holds its shape from internal pressure. A carbonated bottle becomes rigid until you open it and hear the hiss. We learned to lean on hoop: keep membrane stresses smooth, let overburden and horseshoe geometry carry the argument, and design ties only where arithmetic insists.
0.38 g was a modest discount. It helped. Just not as much as pressure did.
♦
Barns — known to some as halls, or sometimes unkindly warehouses — were where scale happened.
They arrived in the starship barrels: IKEA for Martians. Frames, decks, end bulkheads, liners, the same ports and clamps so the spares stayed sane. We stood tied arches or portal frames at a regular pitch, set an intermediate deck as the tie, pulled the liner tight, and started living in rectangles civilian planners could understand.
They began in the early decades as single-storey for speed. They grew quickly: 80 m span was a comfortable mid-evolution number; by the second century, the celebrated great halls ran ≈250 m long, 80 m wide, ~8 m to crown — about two old-Earth football fields of shirtsleeve space in one breath. The outside stayed quiet, buried and boring. The inside changed configuration every decade or so.
We kept a hard line between shell and fit-out. The shell was Tier-A life — fifty-year stuff: buried ribs, panels, primary liner. The fit-outs were Tier-C: seven- to twelve-year lives (in reality twenty plus), meant to be unbolted by tired humans and enthusiastic apprentices. Pressure bulkheads turned barns into independent cells; liner doors made rooms where we needed them today and let tomorrow unmake them without grief.
Was it glamorous? Not usually. Did it deliver m² fast at scale with low leak and kind maintenance? Yes. Ultimately, was it super mass-efficient? Absolutely. That was the point.
♦
None of this was guesswork. We let an army of polite, merciless AIs build cities in simulation until the evaluation scores stopped moving up. They penalized EVA minutes, gasket families, and clever curves. They rewarded shirtsleeves, repeatability, and kg→m² conversion. Ask models to optimize for pressurized area per imported tonne under realistic power and robot fleets, and they rediscover barrels and barns the way water rediscovers downhill.
We worked the five-step rubric like a ritual:
Make the requirements less dumb. No tall showpieces. Pressure + shielding + maintenance first.
Delete parts. One port. One corridor section. One knee set. If the drawing needed a new fastener, the drawing was wrong.
Simplify/optimize. Straight chords over curves; panels over monoliths; liners that swap alive.
Accelerate cycle time. Thirty-meter pieces, 6 m frame pitch, one crew one cell per shift; let robots learn our rhythm and then take it.
Automate. Only after the human takt stabilized did weld-crawlers, print-bots, and seal-bots inherit the night shift.
In those first decades we made peace with shallow or no-cut when regolith handling was weak — more steel, more anchors — and then deepened and bermed a bay at a time as fleets matured. Interfaces never changed. Versions did, constantly.
If you walk the old quarters now, the genealogy is visible: donor-shell barrels dark with age; early barns with fat ribs and stingy cover; later horseshoes cool and quiet under meters of dirt. Systems drifted toward mass thrift, ISRU share, speed of install, and no-drama maintenance. The form factors stayed recognizable enough that a child could draw them. That was deliberate. Children become mechanics.
♦
The pressure-liner ecosystem was the quiet hero. Shells lasted decades; liners did the daily work of civilization.
We standardized on reels, not rooms: laminated membranes (gas barrier + abrasion fabric) shipped as 30m rolls for barrels and 40m spans for barn bays. Clamp-rings captured them to continuous flanges; penetrations were all via gland cassettes you could swap alive. Every 120 m pressure cell — barrel or barn — used the same bulkhead door: hinge geometry, gasket profile, latch count. A tired tech could service street or hall from muscle memory.
Barn subdivision never touched the shell. We hung liner curtains and door modules from the interior frame, creating pressure sub-bays for clinics, class blocks, or fab lines. When use changed, we re-clipped the curtain rails, moved riser panels one bay over, and updated blower setpoints. The shell shrugged; the city turned over new leaves without losing sap.
Leak management was discipline, not drama: acoustic sniff on install, slow-bleed telemetry in the crown plenum, UT/IR patrol bots at night. Mean time to isolate a faulted bay stayed under five minutes — and if you’ve never watched a bulkhead swing in that time, may you never need to.
♦
Barrels grew up the way good tools do: by versioning without renaming.
First generation: donor-shell staves, four × 30 m lengths, knee-joined into a 120 m pressure cell. Liner pre-installed, hold-downs to keep uplift honest, ballast cassettes to fake the berm we couldn’t yet move. We called those “cut & shut.” They offended sensibilities. They also opened on schedule.
By block 14 we had horseshoe cross-sections — ~12 m clear width, ~9 m crown — straight chords at 6 m frame pitch, knees at small angles so hoop forces passed cleanly without inventing curved ribs. Outside: regolith panels laid into ribs like tiles. Inside: the same liner family, now with better edge beads. By mass, the street was ~86% local. We kept the 120 m cell discipline and the one-door hardware rule, so maintenance and isolation felt identical across generations. Swapping a 30 m chord for a better one became a Tuesol task.
Evolution hid inside the familiar. The counters rose. Morale did the same.
♦
Barns tuned three dials as ISRU rose: span, pitch, cover.
Span: 40 m early, 80 m once ribs and panels matured, with 250 m run lengths common by the second century. Pressure, not gravity, set most sections; 0.38 g helped, but only at the margins.
Pitch: frames at a regular 6 m rhythm, the mid-deck as a continuous tie. Two honest floorplates; tied-arch thrusts kept where anchors liked them. Portal frames swapped in where programs wanted straighter walls; the counters barely blinked.
Cover: start 1–1.5 m regolith in geocells — quick to place, good thermal inertia — then thicken to 2–3 m as earthmoving throughput hit its stride. Overburden did triple duty: radiation, thermal mass, uplift counter. We added water jackets above sensitive rooms and called that both shielding and storage.
Every increase in cover shaved leak-rate variance in storms and steadied thermal cycles. People noticed in small ways: fewer drafts, fewer blankets in the mess. Your best metrics sometimes drink tea with you.
♦
How were these spaces used?
Barrels carried the everyday. Ground decks were dual-lane: one lane for walking and hand-carts, one for electric tuggers and pallet bots. Overhead galleries held utilities and a catwalk for inspectors. In the later decades, a narrow mini-tram ran down selected barrels — the Tube, named with impeccable irony — gliding between stops so quietly you could hear your own breath through a mask filter. Frontages opened to small stops, clinics, repair benches, tool libraries. Shift whistles marked the crew change; market hours colonized the calm between dust events.
Barns held the reconfigurable heavy life. Living decks stacked in modular bays — privacy pods, shared wash cores, dayrooms that could become classrooms in a night. Machine shops lined with demountable fume hoods and crane rails, agri bays with stacked grow frames and warm water loops, life-support plant tucked against the service edge where the noise was a comfort. A single hall might be housing this decade, precision fab the next, a field hospital after that. The shell never cared.
Even power visited indoors: not the big prime movers, which liked the cold and the distance, but their switchgear and brains, inside where gloves came off and diagnostics were fast.
A city is what it does in its corridors. Ours worked there.
♦
And yes, we tried domes and polygons. People love a skyline (especially expensive architects and click-bait space-influencers). On first landings and quick-deploy science camps they were honest — fast to raise, photogenic, and easy to find in a storm.
At scale they lost. Curved panels multiplied spares and slowed swaps; seam sealing in dust was a sport we did not have time to master; interiors curved away from the rectilinear logic of racks, beds, benches, and shelves. Polygons promised compromise and delivered hungover geometry — too many unique corners, too few repeatable parts. We built a few civic domes anyway — ceremonies need somewhere to echo — and two centuries later the materials caught up enough to make them somewhat painless.
But they didn’t — couldn’t — do the heavy lifting of the first century of settlement. On Earth, towers and bubbles exist; they do not house the majority. Mars proved mundanely similar.
♦
We cheated where honesty allowed: start shallow, promise deep. Early yards could not move dirt fast; so we traded dig depth for more steel and anchors, then came back bay-by-bay to thicken cover when fleets and power allowed. The geometry tolerated the upgrade path by design: add geocells, pour geopolymer skins, drop water cassettes, re-check thrusts. No heroics. No full shutdowns.
People grumbled about the “low roofs” in those first halls. They forgave us when the second deck arrived — and the winter felt shorter.
♦
If this all sounds inevitable, it wasn’t. We nearly lost the plot to three temptations:
Curves everywhere. Beautiful on posters, expensive in spares, hard to swap. Straight chords and knee sets won; our welders stopped swearing.
Systems-in-walls. Gorgeous in mockups; murderous in maintenance. We dragged services back into walkable galleries and left access honest.
Geology as destiny. Weeks lost waiting for perfect caves. We built where the fleet could stage, and we said yes to gifts only when they arrived.
A civilization is a list of things it said no to. These were just some of ours in those first decades on Mars.
♦
So there they are: barrels and barns — two shapes, many lives. Pressure shells that age like infrastructure; interiors that spin like fashion; a tiny kit that taught itself to make a city by repetition rather than revelation.
We have described the what — form-factors, materials, the SKUs that mattered — and the why — kg→m², shirtsleeves, maintenance without theatre. That was half the argument.
The other half lit up mess tables and design boards for decades: given these ‘lego’ bricks, how do you arrange them? People drew lines. Robots drew others. The math, patient as dust, took sides.
The answer won’t fit in this chapter. It did, however, fit on Mars.