Forty kilowatts on the Moon: the design trades that matter

What the work claims

The paper claims that a 40 kWe fission surface power system for the Moon is technically feasible using near-term technology, and it presents the key trades that govern the design space: power conversion technology (Stirling versus Brayton), heat rejection approach (sodium-potassium liquid metal versus water heat pipe radiators), reactor shielding geometry (shadow shield versus conical), and power management and distribution architecture.¹ The government team developed its concept to adhere to the same NASA requirements posed to the three contractor teams, with the stated goal of producing representative concepts for NASA architecture studies, guiding government technology investments, and informing future requirements definition.

The 40 kWe target is not arbitrary. It reflects the power level NASA has identified as sufficient for initial lunar surface operations, including habitat life support, in-situ resource utilization demonstrations, and surface mobility charging. The fission approach is favored over solar because it provides continuous power through the two-week lunar night without requiring massive energy storage, and it is scalable to higher power levels for Mars missions.

How it works

A fission surface power system has four major subsystems, and the design trades live in the intersections between them. The reactor core produces heat from sustained nuclear fission of a uranium fuel, typically highly enriched uranium-235 in a refractory metal or ceramic matrix. That heat must be converted to electricity, which is the power conversion subsystem. The waste heat from conversion must be rejected to space, which is the heat rejection subsystem. And the electrical output must be conditioned and distributed to the surface loads, which is PMAD.¹

The conversion trade is Stirling versus Brayton. Stirling converters use a sealed working gas (typically helium) oscillating between hot and cold heat exchangers, driving a piston that generates electricity through a linear alternator. They offer higher theoretical conversion efficiency (up to 30 percent or more) and can be modularized into multiple small convertors, so a single convertor failure degrades output rather than killing the system. Brayton converters use a turbine-compressor-alternator on a single shaft with an inert gas working fluid, offering higher power density and a mature industrial heritage but requiring high-rotation-speed bearings and larger radiator area for the lower-efficiency heat rejection. The paper evaluates both against the 40 kWe requirement.

The heat rejection trade is sodium-potassium (NaK) liquid metal loop versus water heat pipe radiator. NaK has flight heritage on space reactor programs like SNAP-10A and remains liquid at room temperature, simplifying system startup. Water heat pipes operate passively with no moving parts but require freeze protection during the lunar night. The radiator geometry must be designed to reject heat at the conversion subsystem's cold-side temperature, which is typically 300 to 400 K for Stirling and higher for Brayton.

Shielding geometry is a mass driver. A shadow shield places material between the reactor and the payload, creating a cone of reduced radiation. A conical shield surrounds the reactor more fully but adds mass. The trade is between shield mass and the usable volume of the payload zone, because a narrower shadow cone means the payload must be farther from the reactor on a boom, increasing structure mass.

The strongest case

The strongest case for this architecture is that every major subsystem has either flight heritage or demonstrated ground testing at relevant scale. Stirling converters have been tested at NASA Glenn for tens of thousands of hours. NaK coolant loops flew on SNAP-10A in 1965. Brayton conversion has been demonstrated at multi-kilowatt scale in vacuum chambers. The reactor designs draw on decades of Los Alamos and Idaho National Laboratory experience with compact fast reactors. No single technology in the trade space requires a fundamental breakthrough; the challenge is integration, qualification, and the manufacturing of a flight unit within a specific mass and schedule envelope.

The 40 kWe level is also well-matched to the Artemis architecture's stated surface power needs. It is large enough to run a habitat and charge rovers, but small enough that the reactor and shield fit within a single lander payload. The scalability of fission means the same core technology, with more fuel and more convertors, can reach 100 kWe or more for Mars missions without a redesign from scratch.

Where a skeptic should push

The paper is a conceptual design study, and the TRL honesty rule requires stating that explicitly. None of the three contractor designs or the government reference concept has been built as hardware at flight scale. The last time the United States flew a fission reactor in space was SNAP-10A in 1965, which operated for 43 days at 500 We. The gap between a 500 We flight heritage from 60 years ago and a 40 kWe system today is enormous, and it spans not just reactor technology but also the entire qualification, launch approval, and safety infrastructure.

The conversion efficiency numbers in the study are design projections, not measured values from an integrated system. A Stirling convertor tested at NASA Glenn achieves high efficiency in a controlled bench environment, but the system-level efficiency after accounting for parasitic loads, controller losses, thermal losses in the heat transport loop, and radiator back-loading will be lower. The paper does not present an integrated system test, because no such test has been done.

The shielding mass trade has a hidden assumption: that the payload can be placed at a specific distance from the reactor on a boom or lander deck. That assumption interacts with the launch vehicle payload fairing and lander geometry in ways that a conceptual study can defer but a flight program cannot. A design that looks mass-efficient on paper may become mass-punitive when the boom must fold into a specific fairing and survive launch loads.

Finally, the study assumes a specific set of NASA requirements (40 kWe, 10-year design life, lunar surface). If those requirements change, the optimal trade space shifts. A Mars surface mission at higher ambient temperature and dust storm conditions would favor a different conversion and radiator architecture than a lunar polar mission.

What it means for power in space

The non-obvious implication is that the conversion technology choice (Stirling versus Brayton) determines the radiator area and thus the stowed volume, which determines the launch vehicle class, which determines cost. The conversion trade is not just about efficiency. It is about the entire mission packaging chain. A Stirling system with 30 percent conversion efficiency rejects less waste heat per kilowatt-electric than a Brayton system at 20 percent, so the radiator can be smaller, but Stirling convertors have lower power density per unit, so more convertors are needed, increasing the PMAD complexity. This coupling between conversion efficiency and packaging is the real design driver, and it is visible in the trade study's evaluation of radiator area against convertor count.

The genuine threat is schedule. The paper states that the government reference concept was developed to guide technology investments and inform requirements definition. That is the language of a program that has not yet committed to building hardware. The three Phase 1 contractor designs were conceptual. Phase 2, if it happens, would produce preliminary designs. A flight unit would follow. At each stage, the trade space narrows, and the mass and performance margins shrink. If the Artemis timeline demands surface power by a specific date, the trade space will be frozen at whatever maturity level the program has reached, which may be below the optimum.

The opportunity is that the government reference concept provides a baseline against which contractor designs can be compared. That comparison function is valuable because it prevents a single contractor from claiming that its design choices are the only feasible ones. The paper's stated goal of informing requirements definition means NASA can set performance targets that are grounded in an independent analysis rather than in a vendor's proposal. For power in space, that is how a government buys a reactor: not by trusting the contractor's numbers, but by having its own.

The bottom line

What is solid is the architecture: a 40 kWe fission surface power system using near-term conversion and heat rejection technology is technically plausible, and the key trades are well understood from decades of analysis. What is not established is any integrated hardware demonstration at flight-relevant scale. The Stirling versus Brayton trade, the NaK versus water heat pipe trade, and the shielding geometry trade are all resolved at the concept level, not at the qualified hardware level. Confidence is moderate that a system matching this reference architecture could be built and qualified within a decade, contingent on sustained funding and a committed flight program. The reading would be strengthened by a Phase 2 preliminary design with mass and performance margins verified against integrated test data. It would be undercut if the launch approval process for a surface fission system proves slower than the technology development, which is the historical pattern for space nuclear power.