Numerical Benchmarks in Fusion Modeling
Standardized problems and cross-code comparisons that fusion plasma codes use to verify equilibrium, transport, and stability solvers.
Shared Yardsticks for Plasma Codes
Fusion plasma modeling spans equilibrium, transport, stability, and heating codes, each solving hard coupled equations with no general exact solutions. The field relies on numerical benchmarks: standardized problems with agreed setups where independent codes are compared. These campaigns give the community a shared yardstick, so that a difference between codes reflects a difference in method rather than in problem definition.
Typical Benchmark Targets
- Magnetohydrodynamic equilibrium solvers, compared on defined boundary shapes and pressure profiles.
- Transport codes, compared on standardized profiles and sources.
- Stability codes, compared on growth rates for defined equilibria.
- Gyrokinetic turbulence codes, compared on agreed reference cases.
Verification and Validation Roles
Benchmarks serve verification when codes are compared to exact or manufactured solutions of the plasma equations, and they serve as cross-code corroboration when compared to each other. They approach validation only when compared to measurements from operating devices. The distinction matters: two plasma codes agreeing with each other confirms consistency, not correctness against nature, which requires experimental data.
Relevance to Kronos
The Hyperion breeder is modeled as a D-T spherical tokamak with negative triangularity, and the burner as a D-3He tandem-mirror generator; both are design and simulation, not built hardware. Their plasma physics claims rest on solvers that must be verified against exact solutions and benchmarked against independent codes, and validated against existing devices and experiments where the regime overlaps. Frozen design points, such as the breeder plasma current and field values, are outputs of such modeling, and their credibility depends on the verification and benchmarking behind the codes that produced them.
No hardware net-gain claim is made before first-of-a-kind operation precisely because full-device behavior lies beyond what code benchmarks and existing-device validation can establish; that final confirmation belongs to the machine itself.