A habitat is not the mission. It is what keeps four people alive long enough to
build the mission. The landing pad. The road. The berm. The next habitat.
Everything here is designed backwards from that job.
Descend
4crewon a 180 day rotation, with a 4 day ride home
372 m³pressurisedacross five modules on a hub and spoke plan
85%closedof daily consumable mass, recovered or made on site
2 mplannedof regolith over every vessel, a later increment
48 kWpeaksolar, backed by 210 kWh and a fission unit
$1Mper kgthe number the whole architecture is arguing with
Habitat or outpost
NASA's Artemis Base Camp is a fixed habitat for up to four crew on stays of about a month,
growing toward two, and the reference surface habitat at delivery holds two crew for 30 days with
four as a seven day safe haven. The longest anyone has lived on the Moon is three days. ACT takes
the station's increment length, about 180 days, to the surface, and that is the step nobody has
made. It rests on station experience for isolation, life support and crew time, and it adds three
unknowns the analogs cannot retire: one sixth gravity for six months, surface dust in the cabin
for six months, and an open surface dose of about 1.37 mSv a day as Chang'e 4 measured it, which the radiation document carries as a design value of about 380 mSv a year, with the hull as the only shielding until bulk regolith is placed.
01Priority one
Survive
A compacted regolith skirt anchors each hull and protects its foot. The two metres of
regolith the design calls for over every vessel, a complete micrometeoroid shield as well,
is a later increment, so for now the hulls share the strike risk with the radiators, array
and cupola glazing, which is why the exposed hardware is paired. A water jacketed storm
shelter under the core for solar particle events, which can put protons on the surface
within about thirty minutes of a flare while forecasts from Earth aim to give up to a day
of notice. Until the cover is placed, the shelter, the 180 day rotation and limits on time
outside manage the crew dose. Dual string life support, because rescue is four days away.
02Priority two
Sustain
Close the loops that can be closed. Water comes back at 98%. Oxygen comes back at about
half through Sabatier, the station's figure today, and the shortfall is covered from regolith
rather than from Earth. Food is only partly closed, and that is deliberate.
03The actual job
Build
The reason the outpost exists. Laser sintered landing pads, roads, blast berms and
shielding, printed from the ground the base is standing on. Every finished structure
makes the next delivery cheaper.
The site
Drawn to scale from the same metre coordinates the virtual tour walks
through. The distances are not arbitrary. The pad sits 65 metres out on a north east bearing
because ejecta leaving a landing site in vacuum travels ballistically, with no air to slow it,
and the berm has to stand between that and the hull. The only measured case is Apollo 12, whose
descent plume pitted Surveyor 3 at 155 metres, so 65 metres is defensible only because the apron
is sintered, with no loose fines to entrain, and the berm catches the low angle fan. That is a
design assumption to be tested, not a measured result.
Site plan as the tour shows it, with the increment three roads and plant in place and the bulk shielding over the modules still to come. The habitat is hub and spoke so any one module can
be sealed off at its connector without stranding the crew. There is no wind on the Moon, so dust
moves by geometry, not by a prevailing flow: lander plume ejecta, rover wheel spray and what
rides in on suits. The excavation face and the rover routes are kept off the airlock approach,
with the sintered road between them, and every route between the four work areas is graded and
sintered to keep fines out of the bearings.
Why ISRU carries the architecture
This is the whole argument in one chart. A tonne of plant landed once, against
a resupply bill that never stops. Both lines use the mass balance on the Systems page, so the
crossover is not a marketing figure.
Cumulative mass landed from Earth. Without the plant the base needs 4.79 kg a day
of oxygen, water, food and nitrogen makeup, which is 1,748 kg a year for ever. With it, the
residual falls to 3.14 kg a day. The plant costs a tonne to land and pays that back in
about twenty months. Everything after that is the reason the outpost is a construction base
rather than a camp.
A day on the base
Four people, twenty four hours, and about 15 crew hours a day on the
work the outpost actually exists to do: suited construction and running the plant. The rest is
the overhead of staying alive well enough to keep doing it, on the station's day of 8.5 hours of
sleep, 2.5 hours of exercise and 6.5 hours of scheduled work.
Two crew go outside on a work day and never more than two at once, so there is always a
pair inside able to run a rescue. Each does one sortie, after an hour of suit checks and
prebreathe, because from a sea level cabin the prebreathe alone runs to hours; that is why the
base keeps NASA's exploration atmosphere of 8.2 psia and 34 percent oxygen. Sleep, exercise and
scheduled work follow the station's day of 8.5, 2.5 and 6.5 hours. Exercise is not optional,
because at 0.166 g bone and muscle loss is a mission risk rather than a fitness matter. Meals
are taken together on purpose: isolation studies keep finding that shared meals are one of the
strongest supports for crew cohesion.
How it gets built
Six years, four increments, eleven workstreams. The ordering is forced: nothing
else can land safely until the pad exists, and the pad is made from the ground it sits on.
The pad comes first because a lander touching down on bare regolith sandblasts
everything within hundreds of metres. Shielding follows the habitat rather than preceding it,
since there is nothing to shield until the modules are down, and the site is shown before that
step. The printed pressure shell trial
in increment four is the step that decides whether habitat volume keeps being launched from
Earth or starts being made on site.
Where the programme stands, August 2026
NASA reshaped Artemis in early 2026. Artemis III is now a crewed Earth orbit flight to test
rendezvous and docking between Orion and the Blue Origin and SpaceX landers, and Artemis IV makes
the first crewed landing at the south pole, planned for 2028. The Moon Base plan runs in three
phases: up to 25 Commercial Lunar Payload Services (CLPS) missions to 2029, semi permanent infrastructure and early habitation from
2029, and continuous presence from 2032. In May 2026 NASA awarded Blue Origin 188 million dollars
for cargo lander services and Astrolab and Lunar Outpost 219 and 220 million for crewed rovers,
with the first Moon Base lander due at the Shackleton connecting ridge in late 2026. NASA's August 2025 directive on fission surface power calls for a 100 kWe class reactor with a closed Brayton cycle, ready to launch by the first quarter of 2030. No surface habitat
contract exists yet, and the reference surface habitat is a two crew, 30 day hybrid inflatable.
Increment one here lines up with phase two.
ACT increment
Years
NASA Moon Base phase it sits inside
What has to exist first
Increment 1pad, habitat landing
0 to 1.5
Phase 2, from 2029: semi permanent infrastructure, early habitation and logistics
A crewed landing (Artemis IV), a cargo lander in the Blue Moon or Starship class, a power grid start
Increment 2shielding, berms, array, ISRU
1.5 to 3
Phase 2 into phase 3
Excavation and construction machines proven on CLPS scale missions
Increment 3fission unit, roads, full greenhouse
3 to 4.5
Phase 3, from 2032: continuous presence
A flight fission surface power unit, which NASA's August 2025 directive targets for launch by 2030
Increment 4printed pressure shell trial
4.5 to 6
Phase 3
A printed structure qualified as a pressure vessel, which nobody has done
The technical blueprint
Plan, section and elevation, dimensioned, with the volume, mass and power
budgets the rest of the concept has to live inside.
Twelve engineering documents, one per subsystem, each at the depth of a real
design review rather than a summary. Every figure on this site traces into one of them, and the pages and the documents were reconciled with each other in August 2026.
Mass delivered to the lunar surface costs on the order of a million dollars per kilogram.
Astrobotic's CLPS price was quoted at 1.2 million dollars a kilogram for a 265 kg lander in 2018,
and the modules themselves ride on Starship or Blue Moon class landers whose per kilogram
economics are not yet published, so the balance uses the CLPS price as a conservative benchmark.
That one number is why ISRU carries the architecture here instead of sitting off to the side as
research. A one tonne molten regolith electrolysis plant drawing 14 kW is rated at about a tonne of oxygen a year in the parametric model the ISRU document cites, and budgeted at 1.8 kg a day, two thirds of that, because most of a small cell's power leaves through its walls rather than into the melt.
How the loops interlock
No module stands on its own. The CO2 the crew breathe out feeds the greenhouse, the vapour it gives off feeds the water system, and the waste metal from the ISRU plant feeds construction. Take one node away and the rest degrade.
98%water recovered
50%oxygen loop closed by Sabatier
48 kWpeak generation
PowerOxygenCarbon dioxideWaterMaterialBiomass
Mass and power balance for a four person rotation. The oxygen loop never closes fully.
Sabatier recovers only about half of it, the station's figure today, because the reaction needs
four moles of hydrogen for every two the electrolyser makes. NASA's exploration target of 75
percent needs methane pyrolysis on top. On the ISS that gap is filled by resupply from Earth.
Here it is filled by the ISRU plant, and that is what makes the base independent rather
than just efficient.
0%
of all water recovered, the figure the ISS reached in June 2023
0%
of the oxygen loop closed by Sabatier, the station figure, before ISRU tops it up
0t
of oxygen a year from a one tonne electrolysis plant at the duty cycle the power budget allows
0d
longest closed bioregenerative run on record, at Yuegong 1
The four loops, one at a time
Each loop closes to a different degree, and the gaps are what determine how much has to be flown from Earth. These are the numbers the whole architecture turns on.
Water
98%
The most closed loop on the base, and the only figure here with real flight hours behind it. Urine, sweat and cabin condensate are distilled back to potable water. A brine processor recovers what the main still leaves behind, which is what took the International Space Station from 93 to 98 percent in June 2023.
Crew demand
3.6 kg per person per day
Makeup required
0.29 kg/day for four
Recovered from
urine, sweat, condensate
Oxygen
50%
A four bed molecular sieve strips carbon dioxide from cabin air and a Sabatier reactor recovers about half of the oxygen bound up in it, which is where the station's system stands today. The loop cannot close on Sabatier alone, because the reaction needs four moles of hydrogen for every two the electrolyser produces, and the methane carries the rest away. NASA's exploration target is 75 percent, with methane pyrolysis. On the station the shortfall arrives by cargo vehicle. Here it comes out of the ground.
Crew demand
0.84 kg per person per day
Shortfall
1.65 kg/day for four
Covered by
ISRU, 1.8 kg/day capacity
Food
1%
Deliberately the least closed loop. Growing staple calories on the Moon costs more in power and volume than shipping them, so the bay grows what shipping does worst: fresh leaf crops with vitamin content that degrades in storage. Lettuce is about 95 percent water, so 4.1 kg a week fresh is about 0.03 kg of dry matter a day against a 2.72 kg dry food demand, about 1 percent by mass and a much larger share of the crew's fresh vitamins. The rest of the diet is flown up.
Grow area
22 m² over six tiers
Yield
4.1 kg per week, fresh
Harvest cycle
28 days, staggered
Power
100%
Fully closed in the sense that nothing is consumed, but the constraint is the dark. Near the pole the sun stays low and the array is mounted vertically to suit it. This site is not in the equatorial 14.7 day night: modelling on Lunar Orbiter Laser Altimeter (LOLA) topography gives the best Shackleton rim sites 240 days of sun without a break and a longest dark spell of about 1.5 days, but a battery cannot carry even that at this load, which is why a fission surface power unit is in the architecture from the second increment and on line in the third.
Array peak
48 kW
Storage
210 kWh, 7.6 h at 27.6 kW
Longest dark spell
about 1.5 days at the best rim site
Daily mass balance, four crew
What the base consumes against what it recovers or makes on site. The residual column is the only part that has to arrive from Earth, and shrinking it is the entire point of the ISRU plant.
Consumable
Demand per day
Recovered on loop
Made on site
Residual from Earth
OxygenSabatier 1.68, greenhouse 0.03
3.36 kg
1.71 kg
1.65 kg
0.00 kg
Water
14.40 kg
14.11 kg
0.00 kg
0.29 kg
Food, dry4.1 kg a week fresh is 0.03 kg dry a day
2.72 kg
0.03 kg
0.00 kg
2.69 kg
Nitrogen makeup
0.16 kg
0.00 kg
0.00 kg
0.16 kg
Total
20.64 kg
15.85 kg
1.65 kg
3.14 kg
85%of daily mass closed on site
3.14 kgresidual per day from Earth
1,146 kgresupply per 365 days
$1Mrough cost per kilogram landed
Whether 1,146 kg a year is credible, and what flies it, against NASA's own forecast and the landers that exist or are in development.
Figure
Value
Source and note
NASA forecast, recurring logisticssustained south pole presence
2,500 to 10,000 kg a year
NASA lunar surface cargo white paper, 2024. Food, water, air, spares and utilisation, delivered about once a year.
This concept, consumables residualthe table above
1,146 kg a year
Excludes spares, filters, lithium hydroxide, clothing and packaging, which NASA's figure includes, so the real bill is higher.
Blue GhostFirefly
up to 240 kg
Flown, March 2025, with LuGRE aboard.
GriffinAstrobotic
625 kg
First Moon Base flights from 2026. One Griffin class delivery a year covers the consumables residual with margin.
Blue Moon Mark 1Blue Origin
3,000 kg
In development.
Blue Moon Mark 2Blue Origin
20,000 kg reusable, 30,000 expendable
In development. The class that lands the 9.4 t core.
Starship HLSSpaceX, human landing system
about 100,000 kg
In development, uncrewed demonstration planned for 2027. One flight per increment covers every module in it.
Where it fails
A closed loop is a set of single points of failure wearing a disguise. Each of these is why a subsystem is dual string, or why a consumable buffer exists at all.
Failure
Time to crew impact
Buffer held
Response
Sieve bed saturatescarbon dioxide removal stops
6 hours
lithium hydroxide, 96 h
Swap to the second string, regenerate the failed bed under vacuum.
Distiller losswater recovery stops
3 days
340 L reservoir
Greenhouse condensate becomes the primary source while the still is rebuilt.
Array dust accumulationgeneration falls off
weeks
210 kWh
Electrostatic clearing on the panel faces, scheduled during the lunar day.
Solar particle eventdose rate spikes
about 30 minutes, forecasts up to a day
12 g/cm² shelter
Crew into the water jacketed shelter under the core, cupola shutter closed in 90 seconds.
Meteoroid or secondary ejecta strikean exposed element is holed
hours to days
second radiator wing, second array string
Isolate the loop and patch. Until the bulk regolith is placed the hulls carry the strike risk along with the exposed hardware; the 0.993 five year no penetration goal NASA sets for surface habitats is what the later cover is sized to meet.
Pressure breacha module vents
minutes
hub and spoke isolation
Seal the affected spoke at the connector. The other four modules stay habitable.
Where the mass actually goes
The same daily balance as the table above, drawn so the proportions are visible.
The width of every ribbon is kilograms per day. Almost all of it turns round inside the base.
Water dominates the flow and is also the most closed loop, which is why the brine
processor matters out of all proportion to its size. Food is the opposite: a small flow that
is almost entirely open, and the single largest reason a resupply ship still has to come.
Nitrogen is tiny and completely open, because there is no nitrogen worth extracting in
lunar regolith and the cabin still leaks.
Surviving the dark
A lunar day is 29.5 Earth days, and at the equator the night half of it lasts 14.7.
This site is not at the equator. LOLA modelling gives the best Shackleton rim sites 240 days of
continuous sun each year and a longest dark spell of about 1.5 days, and average illumination of
77 to 88 percent for a 2 m mast at the rim and ridge sites, best on the ridge toward de Gerlache.
Storage has to cover the longest gap, not the average, and no battery does that at this load.
The array is mounted vertically because near the pole the sun stays low and a
horizontal panel would see it edge on. 210 kWh at the 27.6 kW habitat load is 7.6 hours, enough
for reactor transients and terminator crossings, not for a dark spell. Until the fission unit is
on line the early increments accept a reduced night mode. Once it is, the 40 kWe unit carries the
habitat through any spell on its own; the ISRU plant runs on the daytime surplus and is not a
night load. Eclipses and multi day winter gaps still happen, so the design case is the longest
gap at the chosen spot, which has not yet been computed for this exact site, not the average
illumination.
The thermal design case is polar
The 300 degree swing usually quoted for the Moon is the equatorial case, about
120 C at noon to minus 170 C at night, and it does not apply at 89.9 degrees south. Here the sun
never climbs more than a few degrees above the horizon, so one flank of every module stays lit and
the other stays cold, all year. Diviner puts the illuminated ridge between Shackleton and de
Gerlache near 200 K, about minus 70 C, in summer, with interspersed light and shadow in winter,
and the permanently shadowed floors nearby fall below 20 K at the coldest sites. Radiators
therefore face the cold sky and the shaded ground, never a hot noon surface, and the cold soak
case is a dark spell of a day or two, not fourteen. Diviner's summer maximum map shows the ridge itself reaching 220 to 300 K where slopes face the low sun, so the design range for exposed hardware runs from that down to the 50 to 70 K winter minimum.
Module index
Hub and spoke, so any one module can be sealed off without stranding the crew. The diameters follow analog stations people have actually lived in, from Utah to Ladakh.
5modules
372 m³pressurised volume
4crew, 180 day rotation
Specification, module by module
Pressurised volume is the currency of a surface habitat. Every cubic metre has to be launched, landed and then shielded, so the sizing follows what analog crews have actually tolerated rather than what would be comfortable.
Module
Diameter
Height
Pressurised volume, budget
Dry mass
Power draw
Crew hours per day
Airlocksuit port and dust lock
4.0 m
4.5 m
42 m³
3.1 t
1.4 kW
2.5
Habitation corequarters, galley, medical
8.0 m
5.5 m
148 m³
9.4 t
6.8 kW
11.0
Laboratorycuration and bioscience
5.0 m
5.0 m
54 m³
4.6 t
3.9 kW
4.0
Greenhouse bayhydroponic, six tiers
5.0 m
5.0 m
54 m³
5.2 t
14.0 kW
1.5
Cupolaobservation and comms
4.5 m
4.5 m
36 m³
3.8 t
0.9 kW
1.0
Connectorsfour spokes
2.2 m
horizontal
38 m³
2.6 t
0.6 kW
none
Total
varies
varies
372 m³
28.7 t
27.6 kW
20.0
93 m³pressurised volume per crew member
28.7 tdry mass to be landed
27.6 kWcontinuous habitat load
2 mregolith over every vessel, planned for a later increment
Open item, the volumes do not yet reconcile
Diameter and height are outer hull dimensions. The volume column is the pressurised volume
carried in the blueprint budget, the mass estimate and the tour, and it is smaller than the outer
geometry implies: an 8.0 m by 5.5 m cylinder encloses about 276 m³, not 148. Treat the
volumes as budget figures until the structural definition closes that gap. For scale, NASA's
human research programme sets a minimum acceptable net habitable volume of 25 m³ per person
for long missions; the blueprint's net figure here is 263 m³, about 66 m³ per crew
member. The flown benchmark for mass is the ISS Destiny laboratory, 106 m³ pressurised in a
4.3 m by 8.5 m cylinder at 14.5 t, so a 148 m³ core at 9.4 t assumes a far lighter shell
than a rigid aluminium module, an inflatable or a composite one.
Why five, and why this shape
The layout is not an aesthetic choice. Three constraints produce it almost entirely.
Any module can be lost without losing the base
Hub and spoke means a breach is contained to one spoke. Seal the connector and the other four modules stay habitable. A single large volume is cheaper to build and impossible to isolate, which is the trade every submarine and spacecraft has already made in favour of compartments.
Dust must never reach the living volume
Regolith is jagged, electrostatically charged and abrasive to seals and lungs. NASA's lunar dust toxicity advisory group set a permissible exposure limit of 0.3 mg per cubic metre of respirable dust for a six month mission, and the particles that matter are the ones under about 3 micrometres, so the airlock's capture figure has to be demonstrated in that band, not just under 20. Putting the airlock at the end of its own spoke means every EVA return passes through two boundaries before reaching the core, and the suits themselves never come inside at all. Apollo crews reported burning eyes and throats from cabin dust after every moonwalk.
The greenhouse has to be separable
The bay runs deliberately humid, warm and carbon dioxide enriched, none of which suits crew quarters. Isolating it on a spoke lets each volume hold its own atmosphere, and lets the bay be shut down entirely without touching life support anywhere else.
The eight metre core follows the evidence
MDRS in Utah has run crews of six and seven in an eight metre cylinder for more than two decades, and HOPE in Ladakh, built by Protoplanet with ISRO, used the same eight metre span for its crew module in 2025. Two independent programmes converged on it. That is a stronger argument than any volume model.
Volume, mass and power, side by side
Three budgets that pull against each other. Volume is what the crew live in,
mass is what has to be landed, and power is what has to be generated through the night.
The core takes 40 percent of the pressurised volume and a third of the dry mass,
which is what you would expect of the module people actually live in. The greenhouse is the
interesting one. It is tied with the laboratory for the smallest habitable volume and draws
more power than every other module combined, because photosynthesis is the one process on the
base that has to manufacture its own sunlight.
What the crew build
A lander touching down on bare regolith sandblasts everything within hundreds of metres. So the launch and landing pad is the first structure built, not the last. Everything after it gets easier.
57.2million dollar award
2 mshielding planned, a later increment
0material shipped in
Increment 1
Landing and launch pad
A high powered laser fuses the top layer of regolith straight into a slab that behaves
like ceramic. This is the technique NASA's MMPACT programme (Moon to Mars Planetary Autonomous Construction Technology) and ICON's Project Olympus are developing under a 57.2 million dollar award. No cement, no water, nothing shipped in. The
pad is made from the site itself.
Increment 2
Blast berms and shielding
Stacked sintered pavers raised around the edge of the pad to catch whatever the slab does
not. Then the excavator places two metres of regolith over the pressure vessels, the step
the site is shown before: for now each hull stands on a compacted regolith skirt to about a
fifth of its height. Two metres of regolith is the cheapest shielding in the solar system,
for the simple reason that it is already there.
Increment 3
Roads and graded routes
Regolith is fine enough to behave like talc, and it ruins bearings and seals. A sintered
surface between the pad, the ISRU plant and the excavation face cuts the dust thrown up on
each trip and buys the rovers years of service life.
Increment 4
The next habitat
The last step. Once the printer can lay a pressurised shell, habitat volume stops being
launched from Earth and starts being made on site. That is the point where a base turns
into a settlement.
How the pad is actually made
No cement, no water, no binder shipped from Earth. A laser melts the soil that is already there and it cools into something that behaves like ceramic. This is the process NASA's MMPACT programme and ICON are developing under a 57.2 million dollar award.
Grade and compact
The rover drags a blade across the site to flatten it and drive out the loosest surface fines. Sintering an uncompacted surface produces a crust with voids underneath, which cracks the first time a lander puts weight on it.
Melt with a directed beam
A high powered laser on the gantry raises the top layer past 1200 degrees. Sintering of polar simulant starts near 1180 degrees and a full melt needs about 1360, and lunar regolith is largely silicates and metal oxides, so it fuses rather than burns. The gantry tracks back and forth laying a course at a time, which is why the head in the scene above never stops moving.
Let it cool into a slab
In vacuum there is no convection, so the melt loses heat only by radiation and conduction into the soil beneath. Cooling is slow and even, which is what gives the finished slab its strength. Rushing it introduces thermal stress and the slab fails at the first landing.
Raise the berm
Stacked sintered pavers around the perimeter catch whatever the apron does not. Without them, ejecta leaving a landing site in vacuum travels ballistically and sandblasts anything on the horizon, including the habitat sixty five metres away. Apollo 12 pitted Surveyor 3 at 155 metres.
What regolith actually yields
Lunar soil is roughly 43 percent oxygen by mass, 40 to 45 percent in returned samples, locked into metal oxides. Breaking those bonds gives both the oxygen and the metals, which is why the ISRU plant and the construction gantry are the same supply chain rather than two separate projects.
Output
Process
Yield
Energy cost
Where it goes
Oxygen
Molten regolith electrolysis
1.8 kg/day
14 kW
Life support makeup, closing the half of the oxygen loop that Sabatier leaves open.
Iron and silicon alloy
Electrolysis byproduct
2.4 kg/day
shared
Feedstock for the printer. Brackets, frames and repair stock that would otherwise be launched.
Sintered slab
Laser vitrification
11 m²/day
9 kW
Landing apron, roads between the pad, plant and excavation face.
Pavers
Formwork sintering
40 per day
7 kW
Blast berms, retaining walls, and the module shielding once it is placed.
Bulk shielding
Excavate and place
6 t/day
4 kW
Two metres over every module, planned for a later increment. The cheapest radiation protection available anywhere, and the same machine builds the compacted skirt at the foot of each hull first.
43%of regolith is oxygen by mass
1 toxygen per year from a one tonne plant at the budgeted duty cycle
0 kgbinder or cement shipped from Earth
1180°Cwhere regolith simulant starts to sinter
The sintering cycle
Six steps, repeated one course at a time, until there is a landing pad where
there was loose soil. Nothing in the chain needs a material that came from Earth.
Cooling is the step that cannot be rushed and the one with no power cost. In
vacuum there is no convection, so the melt sheds heat only by radiation and by conduction into
the soil underneath. That makes it slow and even, which is exactly what gives the slab its
strength. Force the pace and the thermal gradient cracks it at the first landing, which is the
failure mode the ultrasonic inspection step exists to catch before another course goes on top.
Grounded in real programmes
Nothing here is invented. Every number traces back to a facility somebody has lived in or a programme somebody is funding today, across four continents and low Earth orbit.
12programmes cited
370day closed loop record
378day printed habitat run
Facility
Operator
Location
Scale against ACT
What ACT takes from it
MDRSMars Desert Research Station
The Mars Society
Hanksville, Utah, USA
A two storey cylindrical hab with a tunnel linked GreenHab and science pod, crewed in two week rotations, with an 80 day crew of seven in 2016. The direct precedent for the hub and spoke plan and the 8 m core.
FMARSFlashline Mars Arctic Station
The Mars Society
Devon Island, Nunavut, Canada
Running a crew in a real polar desert beside an impact crater since 2000, including a four month crew of seven in 2007. The source of the EVA field tactics and crew selection protocols.
MARS OzArkaroola analog site
Mars Society Australia
Flinders Ranges, SA, Australia
A design study for two 20 tonne road transportable horizontal lander modules housing up to eight. The argument for shipping habitat as separate movable pieces instead of one large structure.
HOPEHimalayan Outpost for Planetary Exploration
Protoplanet, Bengaluru, with ISRO Human Space Flight Centre; The Mars Society as senior partner, Mars Society Australia assisting
Tso Kar Valley, Ladakh, India
The two module interlinked pattern, an 8 m crew module joined to a 5 m utility module, proven over a ten day crewed run that began on 1 August 2025. Not the same programme as LHAM, ISRO's separate 21 day analog near Leh with AAKA Space Studio, IIT Bombay and the University of Ladakh.
LUNALunar analogue facility
ESA and DLR
Cologne, Germany
Opened September 2024. Regolith simulant handling, 3 m drilling depth, a sun simulator that reproduces polar lighting, and FLEXhab, a four berth habitat for stays of up to 30 days. The basis for the dust mitigation and lighting assumptions.
CHAPEAMars Dune Alpha
NASA, ICON and BIG
Johnson Space Center, USA
A 3D printed habitat lived in for 378 days by four crew, June 2023 to July 2024. Proof that printed structure is habitable and that the internal zoning works.
Yuegong 1Lunar Palace 1
Beihang University
Beijing, China
The longest closed bioregenerative run on record: 370 days from May 2017 to May 2018, two teams of four, the longer single stay 200 days. Every assumption about the greenhouse carrying real life support load rests on this dataset.
MMPACTProject Olympus
NASA MSFC and ICON
Huntsville, Alabama, USA
Laser Vitreous Multi material Transformation, the mechanism behind the pads, berms, roads and shielding in section 04.
ISS ECLSSLife support, flown
NASA
Low Earth orbit
The only closed loop numbers with real flight hours behind them. Both headline figures on this page come from the ISS, not from a model.
ArtemisFoundation Surface Habitat
NASA
Lunar south pole
NASA's reference surface habitat is a 4.4 m metallic lower deck under a 6.5 m inflatable, sized for two crew for 30 days and four in contingency. The rigid core plus inflatable pattern, and the rover range that sets how far the construction face can sit from the hab.
Lunar HabitationRegolith shell, printed in place
Foster + Partners / ESA, 2013
Study, lunar south pole
The argument that the shielding is the architecture. An inflatable is unfolded from a tubular module, then a robot prints a regolith shell over it as a hollow closed cell lattice, the way a bird bone is built: most of the stopping power, a fraction of the launched mass. The berms in section 04 are the same idea at a lower technology readiness.
GaganyaanIndian human spaceflight
ISRO
India
The programme timeline this concept is pitched against: the first crewed flight now targeted for 2027 behind an uncrewed test flight due in late 2026, a station by 2035 and Indians on the Moon by 2040. ACT is sized for that decade, not this one.
Where ACT sits against what has been done
Every facility in the table above, plotted by the longest continuous crewed run
against crew size. The point of the chart is that this concept is not asking for a first.
CHAPEA put four people in a printed habitat for 378 days. Yuegong 1 ran a closed
bioregenerative loop for 370 days with two teams of four, one of them staying 200 days. ACT
proposes 180 days with four crew, which is comfortably inside both. LUNA is in the table but not
on the chart, because its FLEXhab has no published long run yet. What has never been done is the
combination: that duration, that crew, in one sixth gravity, in vacuum, while operating a plant
that makes its own oxygen. The analogs retire the human factors risk. They do not retire the
engineering.
Building the surface
Surface approach
Exterior, north west rim
DragOne finger to look aroundScrollPinch to zoomClickTap a ring to walk thereClickTap equipment to read it
DrivingWASD or the arrow keys0.0 m/s
Building the tour
A real time walkthrough of everything above, running in a browser with nothing to install. You steer with the mouse, indoors and out, and the camera travels between stops rather than cutting.
9tour stops
8build phases
60fps target
00
Foundation
Static site, no build step, Three.js vendored locally so nothing is fetched at runtime. Waypoint graph, hotspot copy and telemetry authored. Paged navigation and the app shell.
Complete
01
Lunar environment
Hapke reflectance shader for the regolith, including the shadow hiding opposition effect that standard lighting models miss. One hard sun, true black shadow, ACES tone mapping, starfield, Earth low on the horizon.
Complete
02
Habitat and pad
Five modules with hull seams, hatches and handrails, pressurised connectors, compacted regolith skirts, vertical solar array, ISRU plant, comms dish, rover, sintered landing pad, and two suited crew with visors down.
Complete
03
Walkable tour
Nine stops, four exterior and five interior. Drag to look, scroll to zoom, click a ground ring and the camera travels there rather than cutting. Ambient audio synthesised in the browser, and a different bed in every room. Every piece of equipment is a hotspot that names itself, the rover is drivable with a chase camera, a generated perimeter road rings the site, and the sun can be dragged from the horizon through noon and back.
Complete
04
Page heroes and themes
A live scene on every page, each showing that page's own subject, with a palette per page solved numerically so every combination clears the contrast standard. Seven scenes share one WebGL context whose canvas moves between them.
Complete
05
Subsystem papers
Twelve engineering documents at the depth of the Mars 5G paper. Communications, power, thermal, life support, ISRU, navigation, radiation, autonomy, dust, construction, food and medical. All twelve written, each with its own figures and reading list.
Complete
06
Earth to Moon sequence
Click the rocket. Crew board, it lifts off, coasts, and lands on the pad. One continuous shot that hands off into the habitat exterior, scrubbable at any point along it.
Complete
07
Realism pass
Interior fit out, dust and wheel tracks, crew working the site, then the performance budget. Draw call batching, instanced foliage, ejecta and route markers, a crater field under the whole site, and shadows the regolith actually receives.
Complete
Open item, asset licensing
Habitat geometry is written parametrically and dressed with CC0 materials, which keeps the
pitch commercially clean. Most lunar base models on the asset marketplaces carry a non
commercial licence and cannot ship in client work. The GLB pipeline is being built anyway, so
licensed or commissioned models can drop in later without touching the scene graph.
Delivery schedule
Every workstream, in the order it was built, all of it now complete. Weeks
rather than dates, because the sequence matters more than the calendar.
The tour and the content depth pass overlap deliberately. The tour is the
centrepiece and takes the longest to get right, so it runs in the background while the written
and drawn material lands around it. The twelve subsystem documents were the largest single block
of work and the one the client's Mars 5G paper set the standard for.
Sources
Every figure in this concept traces to one of these. Agency documents and peer reviewed work first, reporting only where it is the primary record.
Analog stations and habitats
Places people have actually lived, in deserts, on ice, at altitude and under a sun simulator. Every habitat dimension on this site is checked against one of them.
The measurements behind the polar case: how long the sun stays up on the Shackleton rim, when Earth is in view, how cold the ground is, what the open surface dose is, and what lands on a habitat from above.
Why the cabin is not at sea level pressure, how much room a crew needs for six months, what a station day looks like, and the actual strength of the shared meals evidence.
DragOne fingerLook around. The view turns with your hand and settles when you let go.
ScrollPinchZoom in and out.
Click a ringTap a ringWalk there. The camera travels rather than cuts, outside and in.
Click equipmentTap equipmentRead what it is, what it weighs and what it draws.
StopsNine stops, four outside and five inside, in the bar at the bottom.
SunDrag the slider to move the sun through a lunar day. Near the pole it never rises far.
SystemsDraws the power, air and water loops onto the base as you move.
SoundAmbient audio: a hum inside, silence outside. Off until you switch it on.
RoverOn the exterior leg, drive with WASD or the arrow keys.
Compassion8Innovation
The study book, in four parts
A study book in space systems engineering, written for readers of fifteen and over. Every checkable figure in it was put against the published record. Choose a part to read.
In the headset browser, open the address above, or scan the code.
Open Virtual tour and press Enter VR. Aim at a ring and pull the trigger to walk there.
Any headset with a WebXR browser will do: a Meta Quest, a Pico, a Vive Focus, or a PC headset through Chrome or Edge on Windows. Two controllers with a trigger and a thumbstick are what the tour expects.
