Skip to content
Technology How it works Breeder — Hyperion Burner — Aegis Burner — MetroVolt AI-Native Architecture Magnets Fuel cycle Safety Roadmap
Solutions AI & Data Centers Defense & Government Grid & Baseload Neutron Detection Quantum
Learn Technical Library
Proof Publications Whitepapers Technical Library Open Science & Reproducibility The Honest Gates
Company About / Mission Leadership Environment Health & Safety Investors Careers Press Contact
3D Model
AI Architecture › L3 · Twin Modeling & AI
L3 · Twin Modeling & AI

L3 for the Breeder: Disruption Avoidance

The breeder's highest-consequence event is a disruption; L3 is architected to avoid disruptions early rather than mitigate them late.

THE STACK · click to jumpL7Ecosystem & StrategyL6Experience & VisualizationL5Applications & CopilotsL4OrchestrationL3Twin Modeling & AIL2Data FabricL1Control PlaneL0Foundation▲tlmctl▼L3 · TWIN MODELING & AIThe KRONOS-CTRL digital twin and its predictive shadow.1KRONOS-CTRL Twinlive plant state2GNNscoupled subsystems3PINNsphysics-constrained4Anomaly Ensemblesdrift & fault detection5MPCreceding-horizon control6Predictive Shadowruns seconds aheadMACHINE TIEState estimate descends to L1 control; alerts rise to L4 / L5.KRONOS FUSION ENERGYAI-NATIVE S.M.A.R.T. GENERATORTWIN MODELING & AISHEET 05REV. 2026-08L3 · AI-NATIVE STACK
L3 · Twin Modeling & AI — its place in the stack (left, click any layer) and its internal components (right). Telemetry rises; control descends.

Avoidance over mitigation

A disruption suddenly terminates the plasma, dumping thermal and magnetic energy into the structure, the event a 9.66 MA spherical tokamak most needs to avoid. Kronos's L3 strategy is avoidance-first: use the twin's stability margins and the anomaly ensemble's precursors to steer away from disruptive boundaries before they are reached, keeping mitigated shutdown as a fallback rather than a routine tool.

The avoidance chain

Because disruption precursors give short lead times (milliseconds to hundreds of milliseconds), Kronos pre-computes avoidance maneuvers per mechanism, current/shape adjustment for an approaching locked mode, fueling reduction for a density-limit approach, vertical-position action for an incipient VDE, so the controller triggers a validated response instantly rather than planning under time pressure. The lead-time analysis tells the system which responses are feasible for a given precursor.

Layered fallback

If avoidance cannot succeed in the time available, L3 hands off to a controlled, mitigated shutdown that spreads the energy release as safely as possible; and beneath that, the L1 hardware protection path remains, independent of any model. The layering means the AI buys the lead time to avoid, the mitigation limits damage if avoidance fails, and hardware guarantees the machine is protected regardless.

Pre-FOAK, the avoidance logic is trained and validated on disruption simulations and legacy-device disruption data, then recalibrated on real events after FOAK. The design goal is a machine that spends its operational life avoiding disruptions early, protecting the CrMoNbV vessel and preserving the ELM-free regime.

Content reviewed August 2026 · design-and-simulation stage