Mars surface power is an ISRU load problem

What the work claims

Rucker, writing for a NASA study team, reports the results of a trade study on surface power for a conceptual crewed 500-day Mars surface stay.¹ Two claims carry the paper. First, Mars Ascent Vehicle (MAV) oxygen propellant production is the single largest driver of surface power: of a roughly 34 kWe peak requirement, the in-situ resource utilization (ISRU) plant that fills the MAV's tanks accounts for 32.76 kWe maximum including a 30 percent margin, more than everything else on the surface combined. Second, meeting that requirement with multiple small Kilopower fission units instead of one roughly seven metric ton, 40 kWe Fission Surface Power (FSP) system can save four to eight metric tons of landed mass depending on sparing strategy, while also cutting stowed volume by up to about 60 percent and enabling operations a single large reactor cannot support.¹

Weight the source correctly: this is a systems trade study, not a test report. Every number in it is a paper estimate built on assumed cable masses, assumed junction boxes, and an architecture the authors repeatedly call not well defined. It recommends nothing; its stated purpose is to feed a lander packaging study.

How it works

The study began from the baseline NASA had been carrying since the Design Reference Architecture: two 40 kWe FSP systems per mission, each about 7,000 kg, stowed seven meters tall, with radiators extending about 34 meters when deployed. The primary unit lands with the cargo lander, autonomously relocates at least one kilometer from the future habitat (a trip estimated at up to 40 sols), and powers the propellant plant; the crew later arrives with a spare reactor that is only deployed if the first fails. Four concerns motivated the study: the 40 kWe figure had never been validated bottom-up, seven tons of contingency reactor is expensive insurance, the 40-sol activation window forces the lander to be power self-sufficient for over a month, and a seven-meter-tall reactor dictates a dedicated surface mobility system.

Bottom-up sizing changed the picture. Stationary loads sort into ISRU, the MAV, the habitat, and a sample laboratory. For the largest MAV variant considered, a habitable six-crew vehicle ascending to a one-sol orbit, the propellant load is 23,533 kg of oxygen produced over 480 days, which sets ISRU power at 25.2 kWe average and 32.76 kWe with margin; a small two-crew taxi MAV would need only 19.63 kWe with margin.¹ The MAV's keep-alive power (heating avionics and propellant, health monitoring) adds about 1,157 W with margin. The six-crew habitat draws 17.258 kW in the most recent model cited, which assumes only partially closed air and water loops, so fully closed life support would push it higher. The total peaks at about 34 kWe during the uncrewed cargo phase and 33.6 kWe during the crewed phase, both including 30 percent margin and both explicitly preliminary.¹

The alternative is the Kilopower concept from NASA Glenn: a heat-pipe-cooled reactor directly coupled to Stirling converters, versus the FSP's pumped sodium-potassium loop, with a parasol-like deployable radiator on the 10 kWe version. Table 4 of the paper gives unit masses of 406 kg (1 kWe), 751 kg (3 kWe), 1,011 kg (5 kWe), 1,247 kg (7 kWe), and 1,545 kg (10 kWe), with crew separation distances for under 3 mR per hour running from 100 m for the smallest unit to 500 m for the 10 kWe unit, against 1,000 m for the 40 kWe FSP.¹ Cabling dominates the farm-level trade in miniature: high-voltage cable is estimated at 60 kg per kilometer, low-voltage DC at 1,028 kg for a basic 3:2 copper conductor up to 1,349 kg armored (the study assumed 1,100 kg), so the FSP's separate low-voltage auxiliary line costs nearly a metric ton more in cable alone. Startup shows the same asymmetry: a Kilopower starts on roughly the energy in two D-cell batteries, while starting the FSP needs 5 kWe from an estimated 424 kg of solar arrays.¹

Steelman: the case for the modular farm

The mass case is stronger than the headline. Four primary plus one contingency 10 kWe unit, 50 kWe total, comes in about 200 kg lighter than a single 40 kWe FSP with no contingency power at all, and a one-or-two-spare strategy saves up to eight metric tons; contingency power equal to the largest single unit is essentially free by construction, because losing one of four is an anomaly rather than a mission loss. Growth is cheap in both directions: a late requirement rise from 34 to 43 kWe costs 0.75 to 1.5 tons with an added Kilopower unit, against a second seven-ton FSP.¹

The operational case is the genuinely novel half. One unit activates immediately beside the lander, ending the 40-sol self-sufficiency burden and shrinking lander thermal, structural, and even descent propellant mass. Small units ride on ordinary pressurized rovers instead of demanding a dedicated transporter. And the numbers that jump off the page are about mobility, not megawatts: a solar pressurized rover spends about 80 percent of its time standing still recharging and covers 14 km per day; two redundant Kilopower charging stations raise that to 46 km per day, and four units arrayed in pairs 90 km apart extend maximum one-way range from the habitat from 98 km to 225 km, enough to sweep Jezero Crater, Nili Fossae (221 km), northeast Syrtis (81 km), and Nili Fossae Trough (246 km) from a single landing point.¹ The reactor stops being a building and becomes fuel infrastructure that happens to make rovers better.

Where a skeptic should push

The load-bearing assumption is the load model itself. Every total carries a 30 percent margin on top of habitat figures that shifted from 12.1 kW (2007, four crew) to 14.2 kW (2011) to 17.258 kW (2013) within eight years, and the authors state plainly that the architecture is not well defined and the results are preliminary.¹ If closed-loop life support adds power, or MAV cryogenic keep-alive proves heavier than the 1,157 W placeholder, the 34 kWe floor moves and every mass comparison moves with it.

Then there is the farm's dirty secret: connectors. The baseline FSP needs two cables and four connections; a four-unit Kilopower farm needs ten, and the worst-case twelve-unit 3 kWe farm needs twenty-six field connections, each a robotically mated dusty failure surface. The paper itself cites a U.S. Air Force study identifying connectors as a leading cause of avionics unreliability, and concedes the system-level weak link is the wiring between units, not the reactors.¹ Enrichment cuts the other way: four 10 kWe units need 200 kg of highly enriched uranium against 110 kg for one FSP, 250 kg with a spare, and more than double the baseline in the worst 3 kWe case. The estimated LEU escape route more than doubles system mass at 1 kWe (406 kg to over 1,000 kg) and extrapolates past three tons at 10 kWe, erasing the advantage entirely. Finally, Kilopower is stated to scale to 10 kWe and no further, so the farm is a ceiling as well as a floor.¹ None of this hardware existed as a tested system when the paper was written; these are component-concept comparisons, not measurements.

What it changes for Mars surface power

Implication: the non-obvious point is that surface power architecture is decided by the supply chain it feeds, not by the crew it shelters. The paper's Figure 2 makes this visible: ISRU alone sets the cargo-phase peak, so the power system of a Mars base is best understood as an industrial plant with a habitat attached. That inverts the usual design conversation, which starts from crew needs and treats propellant production as a bolt-on. Size the reactor to the chemistry and the chemistry is unforgiving: 23,533 kg of oxygen in 480 days is a 25 kWe anchor load with a 100 percent duty cycle, and no duty cycling or phasing trick shrinks it much.¹

Threat: the genuine threat is that granularity quietly relocates risk from the reactor to the wiring. A single large unit concentrates radiation, enrichment, and mobility problems but keeps the electrical interface trivial; a farm diffuses them into dozens of field-mated connectors that must survive Martian dust, thermal cycling, and robotic handling. The paper's own reliability reasoning, plus its 1,100 kg-per-kilometer low-voltage cable assumption, means a farm that skimps on connector qualification could lose its entire mass advantage to the harness it created.¹

Opportunity: the under-exploited opportunity is power as range-extending infrastructure. The 225 km figure comes from treating reactors as waypoints rather than as a base utility, and that logic generalizes: the same charging-station pattern applies to drilling rigs, far-side relays, or any surface asset whose productivity is today bounded by how fast a battery can refill. A power unit that redeploys in a rover's cargo volume converts a fixed capital asset into a transportable one, which is a business model, not just a mass saving.¹

The bottom line

As a study of where the watts go, this is durable: the finding that ascent propellant production dwarfs crew power needs follows directly from the chemistry and has only grown more relevant as Mars architectures lean on ISRU. As a verdict on modular fission, it is a conditional one: four-to-eight tons of savings is a paper result resting on assumed cable, junction-box, and connector masses, balanced against real enrichment, safety, and reliability costs that the authors enumerate honestly rather than hide. Confidence is high that the load model (roughly 34 kWe peak, ISRU-dominated, 30 percent margin) is the right shape for this mission class; confidence is moderate that a Kilopower farm beats a monolithic FSP on true lifecycle mass, because the answer lives in connector and deployment data this study could only assume. What would confirm it: long-duration dusty-environment connector testing and a flight-scale cable and junction-box build. What would break it: a requirement for well over 40 kWe, or a closed-loop habitat power draw that inflates the non-ISRU half of the budget.¹