Idaho National Laboratory has built a reactor-physics simulation stack called Griffin that INL says is being used in NASA-related nuclear power modeling work and to help accelerate the path toward lower-cost, factory-built microreactors rated below 20 megawatts thermal. The software, rooted in the open-source MOOSE framework, couples neutronics, thermal mechanics, and heat-pipe modeling into a single digital test bed. With the first fueled microreactor experiments expected inside a repurposed Cold War-era containment dome as early as 2026, Griffin is positioned as part of a DOE effort to reduce the cost and time needed to move small reactors from concept to deployment.
What Griffin Actually Does
Griffin is not a single code but a MOOSE-based simulation stack that stitches together separate physics solvers so they run as one coordinated model. Neutronics handled by Griffin itself can be coupled with Bison for fuel-performance thermal analysis, Sockeye for heat-pipe behavior, and other sub-applications through MOOSE’s MultiApps system. That architecture lets engineers simulate a reactor’s full operating envelope, from steady-state power generation to transient upset scenarios, without building physical hardware first. An INL technical report published through the DOE’s Nuclear Energy Advanced Modeling and Simulation program frames microreactors as systems producing less than 20 megawatts thermal and lays out the programmatic case for building these simulation capabilities before prototypes reach the test stand.
The Virtual Test Bed, maintained by INL’s National Reactor Innovation Center, hosts open-access models that demonstrate this approach in practice. One model couples Griffin with Bison to reproduce the physics of the SNAP-8 experimental core, a space-power system from the 1960s, using historical data archived with OSTI. Another provides geometry, material parameters, and boundary conditions for a fictitious heat-pipe-cooled microreactor assembly, exercising Bison, Sockeye, and DireWolf in tandem. These shared models are explicitly intended for industry stakeholders, giving private developers a head start on designs they can later validate with physical testing and, eventually, with fueled campaigns in INL’s microreactor facilities.
Digital Twins on a Non-Nuclear Test Stand
Simulation accuracy means little without real-world feedback loops, which is where INL’s Microreactor Agile Non-nuclear Experimental Testbed, known as MAGNET, enters the picture. MAGNET is an integrated thermal testing capability that replicates heat-pipe and power-conversion conditions without fissile fuel. INL demonstrated the first digital twin of a simulated microreactor on this platform, pairing a virtual model with live sensor data, machine learning forecasting, and autonomous control algorithms. The stated rationale is direct: lower operating costs and improve safety by catching problems in software before they appear in hardware, while also building a body of evidence that multiphysics codes like Griffin can track real systems in real time.
The MAGNET digital twin architecture breaks into five layers: data acquisition from physical sensors, multiphysics simulation through MOOSE, machine learning forecasting for predictive maintenance, asset control integration, and a dashboarding interface that operators can monitor in real time. A separate INL report on sensor architectures describes how this digital twin can be assembled during construction itself, comparing as-built conditions against design intent to reduce risk and keep costs and schedules under control. The U.S. Nuclear Regulatory Commission has also formalized what counts as a digital twin for nuclear applications, defining the system elements as the physical plant, the twin model, bidirectional data flows, and resulting actions and recommendations. That regulatory definition matters because it sets the bar for any developer hoping to use simulation evidence in a licensing case, effectively turning digital twins from a buzzword into a framework regulators can interrogate.
From Simulation to Fueled Hardware in DOME
The gap between a validated digital model and a commercial reactor narrows considerably once physical test data enters the loop. The Demonstration of Microreactor Experiments facility, or DOME, is a fueled testbed at INL with capacity up to 20 megawatts thermal. It reuses the EBR-II containment dome, a structure originally built for the Experimental Breeder Reactor-II during the Cold War, and repurposes it as a modern microreactor arena. The DOE has stated that developers can use DOME test campaigns to verify their analysis codes and software digital twins before moving to commercial deployment, turning what was once a demonstration breeder site into a proving ground for compact, factory-built reactors.
Peer-reviewed work already supports the credibility of the simulation tools heading into those tests. A study published in Nuclear Engineering and Design validated MOOSE-based multiphysics simulations against experimental data from KRUSTY, the NASA and DOE heat-pipe reactor prototype, showing that coupled neutronics and thermal models can reproduce power and temperature histories within experimental uncertainty. A separate paper in the same journal used MOOSE MultiApps to evaluate transient scenarios and failure cases for a molten-metal-fuel microreactor concept, tying the results to off-grid and space applications including lunar and Mars bases. These are not abstract exercises: when DOME fires up, the simulation predictions will face their sharpest test yet, and the results will help determine how much regulators and investors trust the digital-first design approach that Griffin represents.
Griffin’s Role in Space Power
Griffin’s reach extends well beyond terrestrial grids. The software has already been used to help design nuclear systems for NASA, building on the same multiphysics capabilities that underpin microreactor work on Earth. In practice, this means Griffin can model compact, low-maintenance reactors that operate in vacuum, endure wide thermal swings, and reject heat through radiators rather than cooling towers. For space missions, where mass and reliability are paramount, the ability to iterate designs in software before committing to hardware is especially valuable, and the KRUSTY validation work provides an experimental anchor for those simulations.
NASA and DOE analysts have also examined how microreactor-style systems could support lunar surface operations and deep-space infrastructure, drawing on the kind of modeling discussed in an OSTI report on advanced microreactor concepts. In that context, Griffin’s coupling of neutronics with fuel performance and heat-pipe behavior allows designers to explore long-duration operation without refueling, autonomous load-following, and fault-tolerant configurations that can ride through off-normal events without human intervention. As DOME begins delivering fueled test data and MAGNET continues to refine digital twin workflows, the same toolchain that accelerates commercial microreactor development could also inform future space-power reactor designs discussed in DOE and NASA research, linking terrestrial innovation with off-world energy concepts.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
