Mars Base & Habitation

The base arrives

The Outpost Is Landed, Not Built

Roughly 230 t of hardware in two or three deliveries, and then years of work turning it into somewhere to live.

A minimum viable Martian outpost is about 230 t and a cargo vehicle lands roughly 100 t, so the whole site arrives in two or three landings. Nothing is manufactured from raw material to make the base exist. What the prepositioning phase actually does is unload, arrange, excavate, shield, pressurize and connect — and every one of those is an earthmoving or a labor problem rather than a fabrication one.

The mass ledger is the surprise. Energy chain and mobility together take roughly 44% of the outpost’s mass; habitat structure takes about 11%. The uncrewed outpost is a power plant with an excavator attached, and the habitat line item buys shielding, life support and outfitting rather than volume, because volume comes free with the hull that delivered everything else.

Cargo arrives divisible early and indivisible late, for the reason set out on Mission Architectures: at a working assumption of 50% landing attrition, solar panels and humanoids and habitat shells can be sent in the first window and their losses absorbed, while reactors, the full ISRU train and the ferry wait until the loss rate has been measured.

Offload

Getting the Cargo Out Is a Geometry Problem, Not a Mechanism Problem

The governing variable is drop height, and it is set before launch by where the bay sits.

The payload bay sits at the top of current vehicles for real reasons — load path, ascent stability, entry trim, and keeping the doors out of the landing plume. Every one of those is a recurring cost for a reusable vehicle and a one-time cost for one that flies once and stays. A bottom bay is correct for the expendable cargo lander and wrong for the reusable tanker, which splits the vehicle family by geometry rather than by fit-out.

Sizing follows the heaviest passive item rather than the total manifest. Of roughly 230 t, about 53 t is self-mobile and drives itself off, 42 t is palletized, and 76 t is passive and large. The actual requirement is one reactor of around 6 t lowered 25 m — a hoist, not an elevator. The design rule that falls out is to put a chassis under things that do not obviously need one: a reactor on wheels deletes the sizing driver entirely and buys the ability to move power away from the landing ejecta apron afterward.

The trench

Toppling Is the Configuration, Not the Accident

A vertical hull cannot be buried at any cost. A horizontal one takes about 2,740 m³ of excavation.

Tipping a 9 m by 50 m hull on Mars releases about 9.13 MJ as its center of mass falls from 25 m to 4.5 m, arriving at roughly 21 m/s. A prepared regolith bed 3 m deep brings that down to about 7.8 g — at or below the loads the same cargo already survived on ascent, so restraint is passive and the qualification already exists.

The reason to do it is shielding. Burying a horizontal hull under 3 m of cover takes 2,890–4,200 m³ of material moved; burying a vertical one is impossible at any price. Toppling is what a hull must do to become a habitat. And trenching rather than mounding cuts the work sharply: a free-standing mound at a 33° angle of repose needs about 7,900 m³, while trenching to the hull’s axis needs about 2,740 m³ net.

That fixes an order of operations which no published architecture states:

Two consequences of that last row are load-bearing and almost never stated. Overburden at 3 m gives only 11–19 kPa against 100 kPa of internal pressure, so a pressurized buried shell sits comfortably in tension — but a buried hull that is later depressurized becomes a buckling case. Depressurization is therefore a prohibited operation for the life of the structure, which makes the habitat serviceable from the inside or not at all. And the topside list is a design freeze: every penetration, feed and mount has to be right before the regolith goes on, and nobody has drawn that list up.

An inflatable bag as a toppling cushion works on the arithmetic — absorbing the energy over a 3 m stroke is about 1.33 MN spread over 240 m², or 5.5 kPa, needing roughly 54 kg of gas in a bag of about a tonne, vented rather than sealed. Its best use is not as a landing bed but inside the trench, where a conforming bag shapes the cradle to the cylinder and removes the hardest part of the excavation. A hull toppled onto a pillow at grade costs about 5,200 m³ of extra digging if it is ever buried.

Conversion

The Payload Bay Is the First Habitat, and the Tanks Come Later

2,800 m³ of enclosed volume, and 1,800 m³ of it is full of hardware that has to be cut out by hand.

An emptied hull encloses about 2,800 m³ — roughly three International Space Stations by volume, qualified to about 6 bar against a 1 bar habitat requirement, so doors and penetrations can be cut without compromising the pressure vessel. That is the claim usually made for cargo hulls, and it is true with a large asterisk.

The asterisk is that roughly 1,800 m³ of that volume is tankage, and tankage is occupied: downcomers running through the methane tank, anti-slosh baffles, header tanks, feed and pressurization lines, thrust structure. Clearing it means 200–600 m of cut length at perhaps 0.2 m/min in a suit, multiplied by five to ten for rigging, extraction and staging — on the order of 500 to 1,000 crew-hours per hull for clearing alone, before any fit-out. That is half a crew tour to a full one, per hull, and a site needs at least two.

The escape is to start with the part that needs no clearing. The 1,000 m³ payload bay is already empty, and it is about 1.1 ISS-equivalents on its own — enough for the first crew by a wide margin. Making it the first habitat dissolves what would otherwise be a bootstrap problem, where the first hull has to be outfitted by people who have nowhere to live while they do it. The tanks become a later expansion, converted at leisure by a crew that already has a home.

One hazard belongs here and appears nowhere in the published literature: the methane tank carries hydrocarbon residue on its walls, and the habitat atmosphere under discussion runs elevated oxygen. Decontamination before pressurization is a flammability requirement, not a housekeeping one.

The same arithmetic makes an old recommendation concrete. Designing cargo vehicles for disassembly — tank internals that cut out cleanly, pre-planned penetrations, separable seams, residue that does not complicate habitation — costs no launch mass and has to be decided before the vehicle is built. The crew-hour figure is what gives that recommendation teeth.

Shielding

Regolith Over the Hull, Water Over the Shelter

The discriminator is area, and the two materials win at different scales.

Mars has no global magnetic field and 6 mbar of atmosphere, so chronic galactic cosmic ray exposure is defended with mass, and the mass is locally free. Three to five tonnes per square meter of regolith is the working areal density, which is what the 3 m of cover in the burial sequence delivers.

Water is better per unit mass against solar particle events specifically, because it is hydrogen-rich — roughly 2 t/m² does the work of 3–5 t/m² of regolith. But the scales separate sharply. Over a hull’s 450 m² footprint, water shielding would take 900 t, which no production rate supports. Over a 20 m² storm shelter it is 40 t — a credible reserve that also serves as the life-support buffer and the ISRU interstage buffer. Same tonnage, three ledgers.

Solar particle events give hours of warning and are survivable in a small, heavily shielded volume; galactic rays are continuous and need the whole habitat covered. So the answer is both materials, assigned by area: regolith over everything, water concentrated where the crew retreats.

Martian lava tubes are real, large, and frequently proposed as ready-made shielded volume — skylights in the Tharsis and Elysium provinces are well imaged. The concept is sound and the engineering is unexamined: nobody has characterized floor conditions, roof stability, access geometry or how a pressure boundary would be built inside one. Treat it as a genuine option for a later, larger settlement, not as a first-outpost plan.

The boundary

Atmosphere Choice and the Dust Line

Two boundaries define the base: the pressure boundary and the dust boundary, and they are not the same line.

A reduced-pressure, oxygen-enriched atmosphere — roughly 57 kPa at 34% oxygen rather than Earth-normal — cuts prebreathe time before every excursion and reduces the buffer gas inventory from about 76% of the mix to 63%. It saves gas and it saves crew-hours, which is the argument that matters, and it pays for that in flammability risk, materials compatibility and chronic hypobaric exposure whose long-term effects are not established. Mars is also the one destination where buffer gas is free: nitrogen and argon come out of the same compressor already running for carbon dioxide.

The dust boundary is the harder one. Martian fines are 1.5–3 µm, electrostatically charged, and carry perchlorate at 0.4–0.6 wt% — a thyroid toxin and an oxidizer. Dust that crosses into the habitat does not simply make a mess; it is a chronic exposure. That forces a hard boundary rather than a housekeeping regime, and it is why suitports dominate credible surface architectures over conventional airlocks: the suit stays outside, the crew never handles a contaminated garment indoors.

The choice is also a crew-size decision rather than an EVA-systems detail, which is the finding that connects this section to the next. Ingress and servicing overhead multiplies every hour of useful outside work by something between 1.5 and 4.0 depending on whether the site runs mature suitports or conventional airlocks — and at a lunar-derived cadence of nearly 300 ingresses per tour, that multiplier is worth several crew.

The garage

The Garage Is a Bus Load, Not a Shed

Unpressurized, thermally controlled, dust-excluded — a third zone between the habitat and the open surface.

Every credible architecture names a vehicle and equipment shelter and none of them size it. It is a third environment: not pressurized-clean, not unpressurized-raw, but unpressurized-controlled — lit, warmed, dust-excluded, umbilical-served, with its own boundary and its own currency.

Sizing depends entirely on one unasked question: does the fleet shelter? Servicing bays alone need 50–100 m², because roughly 160 machines needing about 5.5 hours of service each is 880 servicing-hours against 4,368 clock-hours in a tour — one to three bays. Sheltering the whole fleet needs 400–500 m², driven by a humanoid at about 2 m² and an excavator at about 20. The two answers differ by an order of magnitude, and the discriminator is the site’s hazard schedule rather than any property of the machines. Plume ejecta from landings is the interesting hazard: scheduled, repeated, and avoidable by distance rather than by structure — and distance costs traverse time, which costs crew-hours.

What makes it a power question is thermal. A roughly 700 m² envelope held near 290 K against the outside sink draws about 3 kW standing, plus roughly 2.5 kWh per 100 kg of machine warmed from ambient — 3 to 5 kW continuous, which against 13 kW of spare dispatch margin on a 150 kWe bus is a quarter to a third of everything the site has left over.

The neat resolution is that the garage is already on site. A toppled hull floored two meters up gives a chord of about 7.5 m over 50 m — roughly 375 m², against the 400–500 m² the shelter case independently asks for. And a garage hull is far less work than a habitat hull: no leak-tightness requirement, no habitat-grade decontamination, and a large door can simply be cut.

Crew

The Crewed Phase Is Scheduled by Crew-Hours, Not Sized by Mass

Every other resource can be stockpiled or shipped ahead. Crew-hours arrive at a fixed rate and unspent ones are gone.

Station experience gives about 6.5 tasked hours a day over 6.5 days a week, or 42.2 hours per person per week once exercise and personal time are excluded. Over an 18-month surface stay that is 3,292 hours per person, and over a 30-month round trip 5,492. That is the supply, it is fixed, and it cannot be banked.

Demand is where the model breaks down honestly. Scaling the lunar EVA rate to an 18-month Mars stay gives a crew of 7.5; halving that rate gives 3.4. The answer is unresolved by a factor of three, and the reason is not an unmeasured quantity — it is that what the crewed Mars phase is for has never been stated. A science expedition, a construction campaign and a settlement seed consume wildly different hours, and the architectures do not say which one they are.

Two findings follow that change how the robotic fleet should be justified. First, the humanoid’s value against a crew-hour yardstick is the crew-hours it does not consume, not the work it performs — one excavator does the mechanical work of 330 humanoids, so as a labor argument the fleet deflates, but as an hour-saving argument it inverts and wins. Second, EVA hours are simultaneously labor supply and dose expenditure, so more work means more exposure means shorter careers means a larger standing crew. Reducing EVA hours is the only lever that breaks that loop from outside.

Medical care sits under the same shadow and scales with evacuation time rather than crew size. In low Earth orbit that is hours; on the Moon, days; on Mars, never. Autonomous or telementored surgery is the only option past the Moon, and relative to the consequence it is the most under-invested capability in the entire architecture.

Before departure

Five Things That Must Work, Uncrewed, Before Anyone Leaves Earth

Each failure costs a window, and the windows do not overlap.

The scheduling consequence is the part that gets missed. A failed demonstration does not cost weeks; it costs a window. Three demonstrations that must run in sequence rather than in parallel is a 6.5-year serial chain before the first crew departs — which is why prepositioning campaigns are long even when nothing goes wrong, and why the value of running them in parallel is measured in years rather than in dollars.

Record

Provenance and Data Sources