VC11 In-space assembly and servicing
Depth of coverage
At its target depth of L3. What the depth levels mean.
What this layer covers
Autonomous and robotic assembly, docking, alignment of large apertures, in-space manufacture, on-orbit repair and replacement
Where it ends
Building and fixing it; running it is VC12
Deep area
This layer is studied in more depth than most. It belongs to the deep area In-orbit assembly and mass.
In-space assembly and servicing
No cited source reports in-orbit assembly of solar power satellite hardware at full scale. A 2010 NASA Langley paper proposes a much smaller 450 kW-class vehicle, built from eight 50 kW-class modules, as a first step. It is a concept, not a demonstrated assembly (CLM-0343). The only measured human assembly rate in the sources held comes from Skylab 3, reported in a 1975 Raytheon study: an astronaut joined sunshade-pole sections at 6.2 minutes per operation (CLM-0965). The 1978 US design assumed that 555 or 715 people, for its two solar-cell options, could build a 5 GW satellite in six months (CLM-0963). That six months was a rule set at the start of the programme, not an engineering result (CLM-0982). This site's own calculation applies a manual rate of 6 to 12 kg per person-hour, calibrated to Skylab, to the whole satellite. Built at that rate, it would need 1,561 to 7,764 people. The published crews therefore imply about 2 to 14 times as much output per person as that rate, and the 1978 report does not say how (CLM-0973).
Scope. A power satellite is too large to launch in one piece. This section covers how the parts would be joined, aligned and fixed in space, by robots, by astronauts, or by machines that make beams on site. It stops when the satellite is built. Running it is covered in Running it, wearing out, and retiring it.
What is settled. No source held here reports in-orbit assembly of solar power satellite hardware at full scale. A 2010 NASA paper proposes a much smaller 450 kW-class solar electric transfer vehicle, built from eight 50 kW-class modules, as a technology stepping-stone. It is a concept, not a demonstrated assembly (CLM-0343).
The only measured human assembly rate in the sources held here comes from Skylab, the US space station, in 1973. One astronaut assembled two 55 ft poles from 5 ft sections in 137 minutes, which is 6.2 minutes per operation, each fitting one section (CLM-0965). The poles were a sunshade, not structural members. A 1975 study projected 2.5 minutes per operation as skills improved, on a 90 per cent learning curve. That is a projection, not a measurement (CLM-0965). The same study assumed one to two years to assemble the transmitting antenna alone. It judged remote-controlled arms cheaper than hand work, which competed only if a two-year build and the cost of a 50-person space station were acceptable (CLM-0966).
The 1978 US reference design assumed six months per 5 GW satellite. Its initial crew estimates were 555 people for the silicon option and 715 for the gallium aluminium arsenide (GaAlAs) option, two kinds of solar cell. Two figures in that report disagree on the silicon crew in low orbit, 75 against 225 (CLM-0963). The 1981 programme position was about 400 people at the high-orbit base, with about 135 more at a low-orbit base, on 90-day rotations (CLM-0985). A preliminary 1980 assessment anticipated that five years in space, behind the reference shielding, might increase a worker's potential cancer risk by about 12 per cent (CLM-1083). Building two satellites a year needed about 300,000 kg a day delivered to the site (CLM-0964).
What is disputed. The first dispute is whether the crew figures can be believed. This site's own calculation takes a manual rate of 6 to 12 kg per person-hour. The 1975 study calibrated that rate to building-site steelwork on Earth and the Skylab pole, and the calculation assumes 6 to 10 working hours a day (CLM-0972). For the structure alone, which is about 10 to 17 per cent of satellite mass, a six-month build needs 226 to 896 people. The published crew falls inside that range (CLM-0972). Applied to the whole satellite, the same rate needs 1,561 to 7,764 people (CLM-0973). The published crew then implies output per person of about 2 to 14 times the Skylab-calibrated rate, or that 54 to 93 per cent of the labour was automated (CLM-0973). The report states neither. This multiplier comes from the same calculation, which combines two sources three years apart. It is not a finding in either source (CLM-0973).
The second dispute is whether six months was ever an engineering result. It was a programme ground rule, a constraint set at the start (CLM-0982). Boeing wrote that to fit the six months it used skeleton crews and minimal equipment, and slowed its beam machines. The same page says that running the machines a little faster would cut construction time significantly (CLM-0982). Two satellites a year was also a ground rule (CLM-0962). The programme's 1980 transport workshop called that rate "an overriding system driver" and the resulting timelines "probably unrealistic" (CLM-1034).
The machine rates were also planning assumptions. Beam machines form beams in orbit, and their rates are given in metres of beam per minute (m/min). NASA's 1981 summary derived four hours per bay, one repeating section, of antenna frame from a beam rate of 5 m/min. It took the wider range of equipment rates as "assumed to be reasonable" (CLM-0984). The 5 m/min was itself a ground rule (CLM-0986). Three rates, 0.5, 1.7 and 5 m/min, appear for different beams and purposes, and are not averaged here (CLM-0976, CLM-0986). Rockwell's 1981 analysis assumed 1 m/min. It sized 904 people on board for a dual satellite built in 90 days, 226 per shift on three 8-hour shifts a day, with four shift crews on a 6-on, 4-off cycle. That is a different satellite from the 1978 reference, and the source calls its crew table "a top level evaluation" (CLM-0983). Crew figures from 1978 to 1981 come from different designs and contractors. They are not a trend (CLM-0985).
Other designs. Boeing's 1978 cost trade favoured building in low orbit. The 1981 programme chose high orbit on structural loads and environment (CLM-0987). A 2025 Chinese design would assemble everything at high orbit, where there is less debris, with "a minimum of several dozen" robots (CLM-0687).
What is missing. No source held here gives a measured rate for robots joining structural members. The only modern build time held, an assumed two years in a 2022 European study, comes with no rate. The study models "sufficient" assembly capacity without saying what it is (CLM-0967).
Read next. Building a satellite in orbit: do the numbers close? Launch price per kilogram: what the models assume. Launch and the climb to orbit. Running it, wearing out, and retiring it.
Records
38 of the 49 records tagged to this layer are published. The rest are not yet published. A record is published only after a second reader has checked it and it has passed every publication check. What the gate checks.
| Record | In plain language | Tier | Confidence |
|---|---|---|---|
| CLM-0019 | The report sets out Space Solar's plan: a small in-orbit test, then 31 megawatts in 2030, 150 megawatts in 2033, and 650 megawatts in geostationary orbit in 2036. | T4 | MED |
| CLM-0290 | On 29 May 2025 UKAEA, the UK's national fusion energy research organisation, published a press release saying that it and the company Space Solar had finished a demonstration in which robots assembled a small test section of the kind of structure a solar power satellite would need, part-paid for by a UK research council grant whose amount was not published. We have the press release and nothing else; no independent party has confirmed the demonstration or what it showed. | T6 | MED |
| CLM-0339 | A 2019 paper said solar power satellites are often assumed to need visits in orbit for repairs and refuelling, but that no system yet existed to service satellites at geostationary orbit, though one was being developed. | T2 | MED |
| CLM-0343 | A 2010 NASA paper argues that building a solar power satellite in space is not just a robot problem: the module design, the assembly steps, and the supporting equipment all have to be developed together. It suggests starting with a much smaller test system first. | T5 | MED |
| CLM-0517 | Every date in the plan -- the ground demo (2026), the demo mission (2027-28), 2030, 2033 and 2036 -- is a hope, not a booking, and the report that states them says so itself. | T4 | HIGH |
| CLM-0539 | A UK consultancy, working for the European Space Agency and using the same design as its reference point, actually published a full weight breakdown -- something the original 2019 paper never did. It comes out lower than the headline claim, at about 0.70 kilowatts per kilogram rather than 1, even though it leaves out several categories of weight altogether. It also rates one of the satellite's power-conversion stages as barely more than an idea on paper, and its input notes say the lightest case for the radio equipment is an expert's judgement, not something anyone has measured or built. | T4 | MED |
| CLM-0670 | Caltech has two different space-solar-power efforts. One is a paper design that assumes a normal rocket-fuel-based trip to a high orbit. The other is a small test that actually flew, but only in low orbit near Earth, testing hardware pieces, not a trip to a higher orbit. | T5 | MED |
| CLM-0687 | The Chinese design says how it gets to orbit. Rockets carry the parts to a low orbit. A reusable space tug then moves each part up to the high orbit, where they are put together. The paper says that assembling it high up lowers the risk of hitting space debris, because there is much less debris there. The tug comes back down to be refuelled and to have its solar panels swapped. The design does not say whether the tug burns fuel or uses electric thrusters. | T2 | HIGH |
| CLM-0714 | The 2021 UK study by Frazer-Nash Consultancy proposed that the satellite fly itself to its final orbit, using its own solar panels to run its own electric thrusters. Chemical tugs would take the parts only partway, to about 5,900 km, just above the inner radiation belt, where the satellite would be built. The study said the plan needed further work and gave no figures for the self-powered trip. | T4 | HIGH |
| CLM-0886 | One design builds its satellite from six hundred pieces put together in orbit. The other says it needs no assembly in orbit at all. Nothing in between is recorded. | T2 | MED |
| CLM-0924 | In four parts of NASA's programme the 2001 committee found an analysis missing or a goal without a stated basis: for solar arrays, for how much of the assembly work robots should do, for structures, and for thermal control. Reading these as one pattern, targets set before the study that would tell you what to aim for, is an inference; the committee makes each finding separately. | T3 | MED |
| CLM-0962 | The 1978 US government reference design assumed two satellites would be finished every year for thirty years. That was a starting assumption for the cost work, not something the study worked out. | T3 | HIGH |
| CLM-0963 | The 1978 reference design expected each satellite to take six months to build in orbit, using a crew of roughly 550 to 700 people. Two figures in the same report disagree about how many of them were in low orbit. | T3 | MED |
| CLM-0964 | To build two of these satellites a year, about 300 tonnes of material had to arrive at the orbital construction site every day. Getting it there was over a third of the whole cost. | T3 | HIGH |
| CLM-0965 | In 1973 an astronaut on Skylab put together two long poles from short sections in just over two hours, about six minutes per join. A 1975 antenna study used that as its only real-world data point for how fast people can build things in space. | T3 | MED |
| CLM-0966 | In 1975 the antenna alone was expected to take one to two years to assemble in orbit with about fifty people, and robot arms were judged cheaper than astronauts doing it by hand. | T3 | MED |
| CLM-0967 | The 2022 European study assumed it takes two years to build a satellite in orbit, and that getting parts launched could be the thing that makes it take longer. | T4 | MED |
| CLM-0972 | If you take a 1975 estimate of how fast people could build structure in space, checked against one Skylab task, and apply it to the 1978 design's framework, you get a crew of a few hundred to build it in six months, which is roughly what the design said. For the framework, the 1978 numbers add up. | T0 | MED |
| CLM-0973 | Apply that same measured pace to the whole satellite rather than just its framework, and the 1978 design would have needed several thousand people, not a few hundred, to finish in six months. The design assumed something made people two to fourteen times more productive and never said what. | T0 | MED |
| CLM-0976 | In the 1979 Boeing design the machines that extrude the satellite's beams ran at 5 metres a minute, with about half an hour of fixed handling per beam segment, and the assembly line that laid beams and unrolled solar blankets together advanced at half a metre a minute. | T3 | MED |
| CLM-0982 | The famous six months to build a satellite in orbit was decided first and the designs were fitted to it. One contractor says outright that it slowed its machines and cut its crew to make the number come out. | T3 | HIGH |
| CLM-0983 | Rockwell's crew is around nine hundred because it runs three shifts round the clock with a fourth crew resting, and at any moment fewer than ninety of the two hundred-odd on shift are actually building the frame. The machine speed was assumed, not measured, and forty per cent was added for joints and downtime. | T3 | HIGH |
| CLM-0984 | The programme's own arithmetic for the antenna frame: each beam takes its length divided by five metres a minute plus twenty minutes to handle, which comes to about three hours per bay, padded to four for downtime. | T3 | HIGH |
| CLM-0985 | By 1981 the programme put about four hundred people at the orbital construction base, with roughly two-thirds of them on the actual building work and the rest running the base, each staying about ninety days. The prime contractor's 1979 plan for the same base had ten-hour shifts six days a week. | T3 | HIGH |
| CLM-0986 | The beam-machine speeds in these studies were chosen off a curve, with the operating figure set where going faster stopped helping, and the five metres a minute used elsewhere was a planning assumption that the machine-builders thought could be doubled. | T3 | HIGH |
| CLM-0987 | They built in high orbit not because it was cheaper - the 1978 numbers said low orbit was - but because a half-built satellite in low orbit gets pulled about by gravity gradients and drag, cools every time it passes into shadow, and cannot be flown up to high orbit in one piece without breaking. | T3 | HIGH |
| CLM-1027 | Getting a crew from low orbit up to the construction site took half a day and a stage the size of a small airliner, refuelled at the top with fuel the slow electric ferry had carried up. Counted fully, moving one worker's year of labour into orbit cost about a million 1977 dollars. | T3 | HIGH |
| CLM-1034 | The programme's transportation workshop said the schedule was probably unrealistic, and the programme's cost reviewer said the launcher fleet was too small for it: about one launch a day on a five-and-a-half-day turnaround with no reserve vehicles, and a fleet of slow electric ferries with no slack for accidents. | T3 | HIGH |
| CLM-1057 | The safety study's own picture of the system: about $13 billion of hardware per satellite in 1978 money, 640 people in orbit for six months to build each one, forty heavy launches per gigawatt, a launch site loud at six kilometres and windows rattling at 125, and one rocket failure that could kill a thousand. | T3 | HIGH |
| CLM-1071 | Running the programme at full stretch was reckoned to employ about half a million people on the ground, nearly all of them in factories, with roughly a thousand in orbit at any moment. | T3 | HIGH |
| CLM-1083 | Building and keeping sixty satellites was reckoned at fifty thousand working years in orbit, mostly at geosynchronous altitude where the radiation is worst; a preliminary estimate expected that a five-year career there might raise a worker's cancer risk by about an eighth, and nobody knew what heavy cosmic-ray ions would do to eyes. | T3 | MED |
| CLM-1371 | A 1977 NASA design study planned a 10 km electromagnetic track on the Moon to fling small blocks of lunar rock into space at 2.4 km/s, about a million tonnes a year, as raw material for a space colony and the solar power satellites it would build. | T3 | MED |
| CLM-1372 | The 1977 NASA design was a wheel-shaped habitat about 1.8 km across, with the living space inside a 130 m wide ring, spinning once a minute so people feel normal weight. A long cylinder of the same radius was also sized, and would hold far more people. | T3 | HIGH |
| CLM-1373 | Two years after SP-413, a later NASA study's optimised Moon launcher was far shorter, about half a kilometre, because it accelerated at 1000 g instead of 30 g. It would throw about 42 kg a second at lunar escape speed. | T3 | MED |
| CLM-1374 | A 1979 NASA paper let small, hand-picked crews live in faster-spinning habitats (3 turns a minute), but held large settlements of ordinary people to 1 turn a minute or less, to avoid motion sickness. | T3 | HIGH |
| CLM-1443 | The 2021 study for the UK government says the leading designs need no new materials, but that building them at a sensible cost needs two things that are not yet mature: robots that assemble large structures in orbit, and cheap reusable rockets. It says both are developing fast, that the whole system is at an early stage with considerable engineering risk, and that a programme of design and technology demonstration could address that risk. | T4 | MED |
| CLM-1459 | A 2013 design study for a large constellation of solar power satellites proposes that each one climb to its final orbit on electric thrusters powered by a separate, smaller set of solar panels rather than its main panels, which stay folded during the long trip to limit exposure to space weather and debris. The study expects the separate panels to be damaged on the way and reuses them afterwards to help keep the satellite pointed. | T5 | LOW |
| CLM-1468 | The basic idea, a solar power satellite climbing to its final orbit on electric engines run by solar panels it carries, with no separate fuel tanker, is real: it has flown on communications satellites, though not at the scale of a power station. A 2013 design does it with a separate, smaller set of panels. The stronger version, where the satellite's own main panels run the engines, has been proposed, for example by a 1977 US study, which found the panels used for the trip would be badly damaged on the way, and by a 2021 UK study for the last part of the climb only, after chemical stages carry the parts past the inner radiation belt. None of these designs has flown. | T0 | MED |