Fission Surface Power TDU: measured system-level performance

What the work claims

The report presents measured system-level performance of a single-leg Fission Surface Power (FSP) Technology Demonstration Unit (TDU).¹ At nominal operating conditions, the power conversion unit (PCU) produced 8.9 kW of electrical power at 21.7% PCU efficiency, yielding 8.1 kW net system power at 17.2% system efficiency. At the maximum test point, the PCU produced 10.4 kW at 23.3% PCU efficiency, yielding 9.7 kW net system power at 18.7% system efficiency. The authors conclude that the descoped TDU demonstrated system-level performance close to the original requirement of about 10 kW net output at roughly 18% system efficiency, while also identifying specific hardware improvements needed for a flight-like system.

How it works

The baseline FSP system studied by NASA couples a 190 kWt uranium-dioxide fast-spectrum reactor to a primary pumped liquid-metal (NaK) loop. Heat is transferred through intermediate liquid-metal loops to four Stirling PCUs, each nominally rated at 12 kWe, for a gross electrical capacity of 48 kWe. Waste heat is rejected through pumped water loops to titanium-water heat-pipe radiators. The Affordable Fission Surface Power Study baselined this architecture for a roughly 40 kWe surface power system for Mars missions.

The TDU simulated one leg of that baseline, but with key simplifications. It used an electrically heated core simulator, a single NaK loop, one PCU, and a pumped water loop that rejected heat to a facility cooling system rather than radiators.¹ The original plan also included a secondary NaK loop, heat-pipe radiators, and power conditioning; budget and organizational changes removed those elements. As a result, the as-tested TDU had lower parasitic losses than the originally intended configuration.

The measurement boundaries are important. PCU power is the measured electrical output of the Stirling convertor. PCU efficiency is that electrical output divided by the enthalpy difference of the NaK measured across the PCU. System power is PCU power minus the parasitic electrical loads of the Annular Linear Induction Pump (ALIP) that drives the NaK and the water pump. System efficiency is system power divided by the total electrical heat input to the core simulator, so it includes insulation losses between components as well as the pump parasitics.

At the nominal operating point (850 K hot-end temperature, 375 K cold-end temperature, 1.0 kg/s NaK flow, 0.375 kg/s water flow, 15 mm piston amplitude), the PCU produced 8.9 kW at 21.7% efficiency and the system delivered 8.1 kW at 17.2% efficiency. The maximum-power point (875 K hot end, 350 K cold end, 1.1 kg/s NaK flow, 0.39 kg/s water flow, 15 mm amplitude) produced 10.4 kW at 23.3% PCU efficiency and 9.7 kW net system power at 18.7% system efficiency. The same paragraph in the body gives 9.6 kW, and Table I gives 9,567 W at 18.4%, so the reported maximum system value is consistent within about 150 W and 0.3 percentage points.

Steelman: why the result matters

This is one of the few published system-level tests of a liquid-metal-coupled dynamic power conversion system sized for planetary surface power. It shows that a Stirling PCU integrated with a pumped NaK loop and a water-cooled rejector can convert heat to net electricity at better than 18% system efficiency in a relevant thermal environment. That efficiency is substantially above the roughly 7% typical of thermoelectric radioisotope generators and in the range needed for fission surface power to be mass-competitive with solar-plus-storage for long-duration missions.

The test also did what a good demonstration should do: it exposed a real interface problem. The PCU output during NaK-coupled TDU testing was consistently below the output measured during electrically heated testing at Sunpower, and the authors traced the difference to increased thermal resistance between the NaK heat exchanger and the Stirling acceptor. They updated their model with a "tuned prediction" that matched the measurements, turning a hardware discrepancy into a quantified design input for the next build. The same report identifies the ALIP as another concrete improvement target: the as-built pump achieved 5% efficiency versus an expected 10%, cutting NaK flow to 1.2 kg/s and raising parasitic losses.

Where a skeptic should push

The first and largest caveat is that the TDU was not a fission system. There was no reactor, no nuclear fuel, no secondary NaK loop, no heat-pipe radiators, and no power management and distribution hardware. The heat source was an electrical core simulator, and the rejector was a facility cooling loop. The measured system efficiency therefore applies to a deliberately simplified boundary, not to a flight FSP.

The second caveat is that the test was cut short by a helium breach from the PCU working gas into the water cooling loop. The authors suspect a water manifold elbow with documented fabrication issues, but the failure occurred while the PCU was operating within specifications and no repair funds were available. A single hardware failure that ends the test campaign is exactly the kind of reliability issue that matters for a surface reactor that must run for years without maintenance.

Third, the performance shortfall relative to the component-level Sunpower tests is a warning sign for dynamic conversion in integrated systems. The NaK-to-acceptor thermal resistance reduced PCU power across the board, and piston seal clearances appeared to change when the electric head was replaced with the NaK head, causing drift and limiting amplitude to 15 mm. These are integration problems, not component problems, and they are harder to predict analytically.

Finally, the original program aimed to demonstrate the balance of plant to Technology Readiness Level 6, but scope reductions removed transients, radiators, power conditioning, and the secondary loop. The TDU should be read as a subsystem integration test, not as a TRL-6 qualification of a flight-like FSP.

What it means for fission surface power

Implication: The non-obvious implication is that the critical path for fission surface power may be the mechanical and thermal interfaces around the converter, not the reactor itself. The 190 kWt reactor and the Stirling PCU were both understood technologies; what the TDU revealed was that coupling them through a NaK heat exchanger and acceptor can eat enough performance to drop system efficiency by several percentage points. For a surface power architecture, the design of those interfaces is as load-bearing as the core or the radiators.

Threat: Dynamic conversion with moving parts and multiple fluid loops introduces integration failure modes that short ground tests may miss. The helium breach ended the campaign early and pointed to fabrication quality at the water-helium boundary. On the Moon or Mars, a similar seal or manifold failure would not be repairable, and the loss of working gas would shut down the power system. That reliability gap is the main threat to using Stirling-based FSP for long-duration autonomous habitats.

Opportunity: If the NaK-to-acceptor thermal resistance and the ALIP efficiency are solved, the FSP architecture can deliver system efficiencies near 20% at the tens-of-kilowatt scale. That is roughly three times the efficiency of a thermoelectric RTG and, combined with Sun independence, makes fission surface power a credible baseline for polar lunar outposts and Mars surface habitats. The TDU gives the first measured evidence that the liquid-metal/Stirling integration can work, and a clear list of what to fix next.

The bottom line

The 2016 FSP TDU test is a measured hardware-in-the-loop result, but it is bounded by a simplified system configuration and a shortened test campaign. The measured maximum of about 9.6 to 9.7 kW net system power at about 18.4% to 18.7% system efficiency is real for the as-tested boundary, and it is close to the program's original target. It is not, however, a flight system demonstration. The most durable findings are the identification of the NaK-to-acceptor thermal resistance and the underperforming ALIP as concrete design problems, and the helium breach as a warning about dynamic-conversion reliability. Calibrated reading: the TDU raises confidence that a NaK-coupled Stirling power system can reach the required efficiency, but it also shows that the integration details will decide whether that efficiency survives in a long-life surface reactor.