Powering the lunar base, version 2

Part of my series on misconceptions in space journalism, which by this point we may as well admit is really just a paper thin cover for me to write a seemingly endless series of blogs on esoteric space topics.

I’ve previously written a bit about powering a potential lunar base. The vision is that we want a decent sized, permanently occupied lunar base, probably at the lunar south pole. Generally speaking, the Moon “enjoys” 14 day (~340 hr) long days followed by 14 day long nights. This makes consistent levels of power a challenge over night. In particular, a solar+battery cycle would need so many batteries that the solar panels would be barely an afterthought. At the conclusion of that post, having eliminated 3 km tall power towers from consideration I suggested that the best approach might be to exploit thermal contrasts between permanently shaded regions near the pole and to use a bunch of water as thermal mass, heated during the day. At the poles, some places have nights that are only five days long, so provided the lunar base wanted to have enormous caverns full of water, this could be doable. Needless to say, constructing caverns full of lunar water is not cheap.

Original source. Map showing the relative illumination and slope of the Lunar south pole region. Nearby Malapert mountain is tall enough that during Lunar winter the region endures relatively long periods of shading.

There is, however, another way. This proposal is slightly tongue in cheek, but its advantage is that it can be costed because the supply chain already exists, and it is almost certainly cheaper per Watt than any other approach, because it circumvents the requirement for almost any fixed infrastructure on the Moon.

Proponents of space-based solar power have long advocated using gigantic solar arrays in Earth orbit, beaming power to the surface using microwaves. The business case misses by several orders of magnitude but this much is true – electromagnetic power can be transmitted using microwaves.

Okay, spraying microwave power at the Lunar south pole from a series of transmitting stations on the Earth seems excessive, but on the other hand, electricity on Earth is about $0.10/kWh, which is easily a million times cheaper than building a power station on the Moon. Maybe this actually closes?

How would this actually work?

Physicists who are playing along at home can repeat these calculations, but I’ll describe how a particular point design might work. Let’s assume we want a power density on the Moon of 100 W/m^2, which is about 7% of peak solar flux in terms of energy density. In other words, no-one will be getting directly fried from microwave flux.

For this design, I used microwaves at 83 GHz corresponding to one of the highest frequency atmospheric opacity windows, shorter wavelengths (~3 mm), and relatively low utilization in communication networks as of yet.

Also, microwave receivers on the Moon with a certain power demand are sized accordingly. My home has a 100 A service, meaning I could theoretically draw 22 kW before tripping the breaker. Capturing this much power on the Moon would require a receiver of 220 m^2, in the form of a fixed panel perhaps 44 m long and 5 m high, oriented to face toward where Earth is in the Lunar sky. This panel is significantly simpler than a solar panel, instead consisting of a network of small (O(cms)) antenna wires sized to the given microwave frequency, in turn connected to the main power bus via a simple rectifier.

The next architecture question comes down to the size of the “hot region” on the Moon. A smaller region requires less power, but a larger transmission station to create a narrower diffraction-limited spot. There are practical limits both to how much power can be obtained on the Earth in one place, and also to the achievable power density of a transmitting station. These two considerations cut against each other.

For this point design, I specified a beam that is 10 km wide and 500 m high, which projects over quite a wide area on the Lunar south pole, since Earth is very low in the Lunar sky there. For comparison, Shackleton crater on the Lunar south pole is 21 km in diameter. Any given system could be flexibly upgraded with the addition of ground stations on Earth to support other areas of operation, increased power flux, or modifications of the shape of the beam. A 10 km x 500 m beam would carry about 500 MW of power, ignoring side lobes.

On Earth, each ground station would need to be 1200 m in the north-south direction and 60 m in the east-west direction, to achieve the needed phase coherence at the diffraction limit. If we relax the shape constraint to account for the fact that the Earth rotates relative to the Moon and that stations may not be on the equator, the actual transmission antenna array would need to be about 1600 m (1 mile) NS by 150 m EW, and we would need at least three of them, since the Earth’s rotation will point them away from the Moon sometimes.

On this foot print, the transmission station would consist of thousands of discrete microwave transmitter stations, each with a passive dish element that tracks the Lunar south pole. Phase coherence can be performed either open loop, by careful computation of the target location, or closed loop by locking onto a carrier beam transmitted from the Moon. The latter option requires (very basic) machinery on the Moon but automatically compensates for line-of-sight refractive index variations, such as ionosphere electron density.

At this design point, the transmitter station achieves about 2 kW/m^2. Assuming 50% efficient RF klystrons, the thermal load is similar to a large old school incandescent theater sign – so we won’t be melting down. Indeed, power stations often use cooling ponds or radiators that are smaller than this. An alternative approach is to use higher powered digital RF amplifiers, which would look similar to a field covered in Starlink antennas. Starlink phased arrays are significantly more complex, being able to transmit and receive data as well as raw microwave power, and cost about $6000/kW, compared to about $20/kW for a basic magnetron. At 1 kW per transmitter element, we’ll need millions of them so unit costs may fall as low as $200/kW, which compares favorably to the cost of energy generation, despite inherent inefficiencies.

Regardless of transmission technology, the station requires at least 500 MW of electrical power before amplifier inefficiencies and beam losses are accounted for. As a result, each station would require a dedicated power station. Modern combined cycle gas power plants can be built for just under $1000/kW, so capital expense would be around $1b per station for electricity generation, and around $200m for transmitter elements. Let’s throw in development costs, NASA overhead, and generous rounding and say we should be able to build a deep space power transmission network for $10b in a few years.

Within the beam, the marginal cost of adding additional power capture is the cost of the panel – perhaps $100/kW and 1 kg/kW, while the marginal cost of tweaking the beam to increase power flux density at a site of interest requires merely moving the ground-based transmitters a short distance to tweak the antenna shape, at a cost of perhaps a few weeks of labor, let’s say $100,000. So while $10b up front cost seems like a lot of money, it’s important to note that the actual cost of power delivered to the Moon on a per kWh basis is barely more expensive than power on Earth, and there is little marginal benefit to reducing power usage, which helps relax design constraints.

What are the alternatives?

Let’s say we have a baseline demand for 100 kW on the Moon, similar to the International Space Station. Currently available space-rated solar panels achieve around 100 W/kg, so 100 kW of solar PV would weigh 1 T, not including structures, connectors, power electronics, etc. State of the art batteries can achieve about 250 Wh/kg, so 100 kW of power over a five day shaded period would require 50 T of batteries.

What is the cost of delivering mass to the Moon? We don’t actually know. We’re going to spend more than $100 b developing the SLS moon rocket, which is currently on the launch pad awaiting its maiden flight. But the SLS can’t actually land anything on the Moon, or even get much mass into Lunar orbit. It’s kind of half of a moon rocket. The most recent NASA OIG report estimates its cost at $4.2b per launch. I think it’s safe to assume that’s a lower bound going forward, and it doesn’t include a lander.

Under the Artemis HLS baseline, 12 T of cargo capacity to the surface was sought. The 100 kW lunar power system would then require 4-5 cycles of the HLS mission operation, each requiring an SLS launch as well as 3-4 other launches, plus new landers and descent/ascent elements. $10b is a generous cost estimate, and several miracles would be required to exceed a cadence of two years between missions. At the end of the day, we’re looking at $50b and a decade just to deliver a power system that is slightly worse than the current ISS power system.

Regular readers will be shocked to learn that SLS fails to deliver value. Shocked, I say!

Does Starship change this much? Assuming $10m per launch and 10 launches per Lunar landing for LEO refilling, $1b could transport 2000 T of cargo to the Moon per year. Indeed, the 50 T power system could fit in a single Starship lander with room left over, for a total delivery cost only $25m.

But just how far does 100 kW go in a Starship world? Given that each Starship has double the internal volume of the ISS, and 2000 T/year could easily support a base with 1000 people, we’ve gone well beyond the default Artemis option of really expensive camping for a couple of people for a few weeks every few years. McMurdo antarctic station consumes up to 2 MW of power, which is probably a good baseline assumption for the Moon, assuming that no large scale electrolysis of water is occurring. Under our previous assumptions, a 2 MW power system would weigh 1000 T, consuming half of the total yearly Starship manifest assuming 100 launches per year!

Current space hardware development costs range from $100,000/kg for basic, boring LEO hardware up to >$1m/kg for deep space robots such as the Mars Science Laboratory rover. A 50 T lunar power system would then cost between $5b and $50b, depending on how long it took to make, while the 1000 T option breaks this cost model thoroughly.

Let’s compare the two options over a 10 year program. 175 GWh of power are consumed on the Moon, assuming 2 MW consumption over that time frame. For Earth-based power, start up costs are $10b, and ongoing yearly costs are $880m for power (at $100/MWh), plus some cost for maintaining the antennas. Let’s round it up to $1b a year, meaning the program costs a total of $20b on the ground side. On the Moon, the panel hardware costs perhaps $200,000 to manufacture and, weighing just 2000 kg, only $2m to deliver assuming Starship HLS achieves $1000/kg landing costs. This could be much higher and it would barely affect the overall program cost. All up costs work out to about $1.14/kWh, which is barely 10x more than terrestrial power.

In contrast, a 1000 T power system that cost $1b to deliver and $100b to manufacture over 10 years works out to $5.76/kWh, not even including the programmatic costs of delays and the fully loaded cost of increasing power capacity. Indeed, for ground based power the $/kWh number actually decreases as more power is captured on the Moon, since it helps amortize fixed ground costs.

The bottom line is that generating power on the Moon is not only horrendously expensive, it also requires a bunch of dedicated mass and effort and schedule. I would go so far as to say that alternate approaches being examined presently require an unreasonable number of miracles to work at all, let alone work in finite time and budget. Either we need a beefy space-rated nuclear power option, or much better batteries, or the ability to mine and store tens of thousands of tonnes of Lunar water.

Let’s look at some other benefits to bathing the Lunar south pole region in coherent microwave radiation that can be readily converted into electrical power. As previously mentioned, microwave power conversion systems are super simple, and can be integrated into the outer surfaces of habs, scientific instruments, landers, rovers, and even space suits. There’s no need to hand out a bunch of the mass budget to huge batteries to drive rovers around, or to keep space suits working for hours. There’s no need to entrench miles of meticulously insulated power transmission cables between a bunch of outlying facilities. The power is literally in the (non-existent) air.

Let’s say we want to increase the size of the Lunar base by 50%. With Earth-based power transmission, the size increase only requires a few extra panels to grab power that was going to waste anyway. There is no marginal cost increase for putting out more panels, since the power is there anyway – use it or lose it. What resources there are can be devoted to building out the base and doing cool stuff with it.

Without ambient microwaves, any base growth places enormous demands on whatever existing systems produce power, or else require building and shipping a new nightmare power plant from the Earth, with (optimistically) 10 year lead times and >$10b total cost.

The moon is tidally locked, so except for low amplitude (~7°) orbital libration, the Earth is always in the same position in the Lunar sky. The power generating panels don’t need to track the sun, nor worry about shading at various times of day. Provided they’re set up to face the Earth once, they’re good forever.

There is, sadly, one major limitation. The “dark side of the Moon” never sees the Earth, so Earth-based power transmission won’t work there. Indeed, the polar region is on the terminator, so some parts of it never receive sunlight and somewhat larger parts of it never or rarely see the Earth. On the other hand, while even the sunniest parts of the lunar pole are shaded for days at a time during Lunar winter, roughly half of the Moon, including plenty of land near the poles, always has an uninterrupted view of the Earth.

Original source. This map shows which parts of the Lunar south pole have consistent views of the Earth.

Nevertheless, Earth-based Lunar microwave power transmission solves the Lunar power problem completely for finite money for half of the Moon, which is 50% more than existing alternative architectures. If we’re serious about building a sustainable lunar base, the least we can do is make pragmatic investments now that provision the program with extremely cheap, post-scarcity electric power.

36 thoughts on “Powering the lunar base, version 2

    1. I discuss that in the first lunar power blog. It would need 100s of km (at least) of cables plus half a dozen completely redundant solar power farms. This much surface infrastructure is not scalable and obviously incredibly expensive.


    2. Surely the simplest solution is a nuclear power plant. McMurdo has one. Space transportation didn’t kick off until
      SpaceX perfected reusable boosters, which the conventional wisdom said was impossible. Also impossible-but-probably-not-really is nuclear fuel in a package that can survive a worst-case scenario accident. Solar power *may* work to supply the power needed for the anticipated infrastructure in LEO, but not for a truly space-fairing society. Starship will force some commensurate engineering for the other technologies needed to leverage it.


  1. If this is implemented, then I guess the rebellion scenario becomes that much less likely – a side benefit for some people.


      1. I have no idea if a lunar colony could be self-sufficient in other things to consider independence.
        However, if power is supplied from Earth, any “rebellion” (I put that in quotes because a state of rebellion is in the minds of the leaders) could be answered by shutting off the power.
        That would kill the lunar settlers in short order of course, but that might be considered desirable.


      2. There is no possibility of a lunar settlement becoming self-sufficient, so it is a wholly idle concern. Much more likely is flatlanders losing interest, and cutting funding below what is needed to keep it alive. Then people would have to leave.

        But, with any luck, people can’t even live for long in 1/6 g. (We don’t know how much gravity is needed to keep bones from dissolving away. Or what other processes might be similarly affected.) Then there would be no such thing as a lunar settlement.


  2. In a Starship world, why not just take the existing power plant from a decommissioned nuke submarine, fit it in a SS, and send it off. It is proven tech, dirt cheap, and already has served an isolated habitat with populations in the hundreds.

    And when the plant itself reaches end of life, just tank up the parked SS and send it on a long slow trip to the sun.


    1. Naval nuclear reactors require an unlimited supply of freshwater. They’re also heavy, expensive and only reliable in the sense that they run at a fraction of nominal power most of the time so a given core doesn’t need very much uptime on a given cruise before the next shore overhaul. Additionally, the heat dissipation would be a significant challenge. But besides those two disqualifying reasons and the significant challenge it should be easy.


    2. US naval nukes are extremely badly adapted for lunar use. They depend on water, which there is none of.

      A nuke designed for the moon could entirely lack outer containment and shielding, so most of the weight typical of nukes would be left behind. It could also, being designed for use in the lunar night, count on an abundant supply of extreme cold, down to -250 F, to operate very efficiently. The only really bulky part would be the radiator panels the coolant would condense into, but they could be very light with no need to contain much pressure. Probably if the turbine or pump, the only two moving parts, needed service you would discard and replace it via robot.

      Liked by 1 person

  3. Regarding the thermal approach, why use water as the storage medium for the heat? One advantage solar thermal has on the Moon is that the sunlight is not diffuse; While the Sun is not a point source, it can be concentrated to an intensity comparable to the solar surface, and light can be mirrored around with relatively low losses.

    Drill a shaft, and direct the concentrated sunlight down it, and eventually you can have a substantial mass of regolith heated to near the melting point. At that point you can run a “geo” thermal powerplant between those heat wells and radiators located in permanently shaded spaces.

    Liked by 3 people

  4. To me beaming power via microwaves would be more practical with a set of solar arrays put into a near rectilinear polar orbit. One of the excuses NASA is using for creating the Gateway is that it will act as a communications relay. The orbit is easy to maintain (in sims) and spends most of it’s orbit in view of the south pole. A Starship or 2 could be loaded with solar panels and big antenna to be permanently parked at the gateway. Or you could place two starship/power systems into separate NRLO and always have at least one overhead. What kind of of antennal size do we need for 70K kms?


    1. Big gossamer mirrors in halo orbits around the L1 or L2 point, focusing sunlight onto solar panels on the moon, might work better. Or, big mirrors concentrating (a narrow wavelength range of) solar onto panels and using the power to drive a laser pointing at panels on the moon.

      You can make a mirror for a narrow band by stacking layers tuned to the wavelength you want to reflect, as in bird feathers and butterfly scales. The rest of the light goes on through. The panel then has no problem with cooling, as all the incoming light is converted, and you can focus a very great deal of light onto it without melting it.

      Probably such mirrors should be available for use on Earth. They are quite cheap to make, layered on mylar film, and a simple arrangement could triple the output of an ordinary panel. Probably a similar mirror layer to put on the panel to reflect away all but the productive photons could keep desert panels cooler, thus more efficient and longer lasting. (Maybe somebody does this already?)


  5. “and it is almost certainly cheaper per Watt than any other approach, because it circumvents the requirement for almost any fixed infrastructure on the Moon.”

    Cheaper then any other approach? Have you already forgotten the brilliance of sending diesel generators to the moon? Or the slightly more boring idea of pointing a mirror at a solar panel? https://www.nasa.gov/directorates/spacetech/niac/2021_Phase_I/Light_Bender/

    Sure a mirror is fixed infrastructure on the moon but it’s really simple fixed infrastructure.


      1. Why are there geometric problems with beaming light from a parabolic mirror but not microwaves? I’d expect diffraction to be more of a problem at longer wavelengths.


  6. Nuclear, people. Radiation is NOT a concern – get real. You need reliable sources of power, and at a minimum a TNG will fill the bill until larger plants can be built.


  7. Casey

    I have enjoyed your blogs for a while now, and this one is very intriguing. I know that our local L5 chapter has been wrestling with this for some time now.

    The details of KWHs and $s are beyond me, being a science fiction writer, not a scientist. But I have an idea based on broad general principles, and I would rely on you to either poo poo it, or, if it holds any water, to flesh it out.

    My idea would use your lunar surface infrastructure, with some different elements in space. First, it involves satellites in lunar orbit. I’ve never understood why this concept gets so little attention. NASA, more and more risk averse, has come up with some celestial acrobatics to put Artemis into lunar polar orbit so there will always be line of sight communications with Earth. Why? When it would be so logical to put some communications satellites in lunar orbit, like the TDRS constellation for Earth orbital manned flights. And it would require less delta V for an already undersized capsule.

    My plan for lunar electricity hinges on two technologies.

     Can a satellite capture sunlight and convert it to microwave energy? It seems Japan has been working on this for Earth energy needs. 
     Can the microwave energy be stored and transmitted to another satellite before that one beams it to the surface? From what I understand about how Starlink and other low latency internet constellations work, it looks like this technology is coming along. 

    So, instead of the energy being beamed from Earth you can instead have the energy created by satellites in lunar orbit.

    There are a number of possible configurations. A handful of satellites in lunar equatorial orbit generate the power When over the receiving station it beams the power down. When in lunar night the satellite goes dormant while others in sunlight do the work.

    Or the satellites can be in a polar orbit, always in sunlight assuming the right, Artemis-like orbit. Power can be beamed down to almost any location on the moon, including the poles.

    This scheme avoids beaming power from Earth and the losses due to the atmosphere and the distance squared law. It also confines microwave beaming through space to the cislunar neighborhood. It avoids or solves the problem of the 14 Earth-day-long night.

    Other opportunities to consider are whether a Lagrange point could be helpful. Or a Lunar synchronous orbit?

    All this would make your lunar surface collectors more efficient since the microwaves originate hundred, not hundreds of thousands, of miles away.

    I’d be curious to hear from you if there’s and promise in this. Speculative, sure, in view of the current state of the relevant technologies. But, of course, “speculative” is what sf writers thrive on.

    Thanks for the thought provoking blogs.


    John Vester


    1. IMO the only efficiency that matters is capital efficiency. Even though the process I proposed here throws away 99.99% of the power, it’s still cheaper.

      Lunar orbits have stability issues.


      1. Can’t be too terribly unstable, since the Lunar Prospector for instance spent 19 months in lunar polar orbit. Beaming power from Earth is nice outside-the-box thinking, but I would like to see some real math comparing cost of solar + microwave in lunar orbit (including cost of extra propellent for course corrections) vs. Earth orbit.


  8. Solar panels mounted on poles on the rim of Shackleton Crater, rotating to face the sun, would surely be less bother.

    I think your figures on W/kg are scaled to terrestrial panels. Lunar panels, not exposed to oxidation, precipitation, or wind, could be gossamer perovskite film, much lighter, and could approach 50% efficient.

    With a tall enough pole, which could be very spindly in the total lack of weather, it could spend 100% of the time in sunshine. But anyway if you sited a bunch of poles around the rim, at any time most would not be shaded. Then you would need a transmission cable a few tens of km long. Unfortunately the bottom of Shackleton crater is not quite cold enough for superconducting cabling.

    The poles could just have big steerable mirrors on them, instead, reflecting sunlight to a fixed panel array. Then you don’t need the long cable.


  9. My imaginary lunar city still has its solar panels at the top of its elevator, because it looks cool. We don’t know what the best material to make an elevator from will be, so I can say that mine conducts electricity at no extra cost.


  10. If the battery storage system is going to be indoors in a radiation hardened enclosure, I don’t see why you can’t use a COTS battery storage system like the Tesla mega pack.


    A 3Mwh module weighs 23 tons. It costs about $1.5 million. 300 Mwh of storage would cost $150 million and weigh 2300 tons. Triple that for some lightweighting and minor customization and you are still under $500 million.

    Assuming 1 Mwh per person, that should be enough for a 300 person station.

    It will be a very long time before SpaceX sells starship flights at $10 million. I would go with at least $50 million.

    That would be about $8 billion in launch costs.

    If starship truly delivers and that is a big if, you can build a 300Mwh battery storage system that could support a 300 person base for around $10 billion.

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s