Firebreak is built on Distributed Order Network (DON) Theory — a mathematical framework where grid segments form nodes in an interaction network, and entropy emerges from physical measurements at each edge. The system identifies collapse risk when network coherence crosses a mathematically derived threshold. No neural networks. No training data. No black box.
The transmission grid is modeled as a network derived from California Energy Commission (CEC) public GIS data. Every node corresponds to a real substation or junction. Every edge corresponds to a real transmission segment. Edge entropy quantifies the disorder state of each segment from physical measurements — grid telemetry and environmental conditions — combined into a single continuous value representing segment instability. A high-entropy edge is one where the combined physical state has crossed into disorder. The system measures grid state, not weather.
A single elevated entropy reading is not enough to trigger an alert. Dangerous weather happens regularly without causing fires. Firebreak tracks how entropy builds and persists over time — distinguishing weather that looks dangerous from conditions that are building toward actual collapse. Several fires in the 14-fire backtest were only detectable through this temporal evolution, recovering signals that individual hourly readings missed.
The bifurcation threshold (θc = 0.42) is derived from DON Theory. It is not a tuned parameter and is not machine-learned. It is a property of the mathematical framework that describes when a network transitions from ordered to disordered behavior. When segments exceed this threshold with sufficient temporal persistence, the system routes power through lowest-risk alternative paths. In sparse grids where no safer path exists, the system flags risk for operator review. Validated against 14 historical California wildfires: 13/14 detected (covering 140/140 deaths) via DON substrate evolution, zero false positives across 300 non-fire topology-days.
Same inputs produce the same result. There is no stochastic variation, no random seed, no run-to-run divergence. The output is a direct mathematical consequence of the physical measurements.
The system was not trained on fires. It was not trained on anything. It calculates entropy from physical measurements using a fixed mathematical framework. The 14-fire backtest validated the framework — it did not train it.
The system cannot produce confident outputs from nonsensical inputs. Entropy is bounded by the physical measurements that produce it. If the inputs are noise, the entropy reflects noise — it does not fabricate a coherent answer.
Every routing decision traces to specific entropy values on specific edges, which trace to specific physical measurements at specific times. The full decision chain is logged and reproducible.
Firebreak operates on existing SCADA infrastructure and standard environmental data feeds. No new sensors, no new construction, no physical installation. The system connects to existing telemetry via OPC-UA or DNP3 and supplements with publicly available weather data. Deployment is a software integration — not a capital project.
DON Theory is protected by issued and pending patents of DON Systems LLC. Detailed technical briefings are available under NDA to qualified utility partners and regulatory bodies.
DON Systems provides detailed technical briefings under NDA for utility engineering teams, grid operators, and regulatory staff. The briefing covers the DON Theory mathematical framework, the entropy calculation pipeline, validation methodology, and deployment architecture.
Request Briefing