Code Benchmarking and Validation Databases
Curated databases of well-characterized cases give codes a common yardstick, turning validation from anecdote into systematic evidence.
A shared yardstick
Individual validation exercises are hard to compare when each uses different cases and metrics. Benchmarking and validation databases collect well-diagnosed reference cases, with documented conditions, measurements, and uncertainties, so that many codes can be tested against the same data and their agreement compared on a common footing. This turns validation into cumulative, systematic evidence rather than scattered one-off claims.
Such databases also underpin the empirical scaling laws that systems codes depend on, since a scaling is only as good as the multi-machine dataset behind it.
What a good entry contains
A useful database entry documents the full state needed to reproduce a simulation, geometry, profiles, heating, and diagnostics, together with the measured quantities and their error bars. Without documented uncertainties a comparison cannot be judged, so uncertainty accompanies every measurement in a well-built database.
From cases to metrics
Beyond storing cases, mature validation practice defines quantitative metrics of agreement between simulation and data that fold in both measurement and model uncertainty, so that codes can be ranked objectively rather than by eye. This connects directly to validation workflows.
Design relevance
The codes used to design the Hyperion breeder and the Aegis and MetroVolt burner are validated against such databases of existing experiments. This is the basis for treating the design predictions as credible simulation while honestly reserving hardware net-gain claims until FOAK first tritium around 2030.
- Common reference cases for many codes
- Documented conditions and measurement uncertainties
- Underpin empirical scaling laws
- Enable objective agreement metrics