Sectionalization and Fast-Trip Planning for Ignition Risk Mitigation
Utility-caused wildfires pose severe risks, and operational interventions like PSPS and fast-trip settings reduce ignition likelihood at the cost of service reliability. This visualization highlights how hyperparameter choices shape the tradeoffs between ignition risk mitigation and reliability in our adaptive robust optimization framework for coordinated sectionalization and operational planning.
See our paper for the detailed model and case study results.
Understanding the key parameters
We do optimal planning under operational and physical constraints, with statistical control in prediction. Below is a guide to the main modeling assumptions and parameters.
Main assumptions
Uncertainty set validity
Future ignition risk is uncertain, but its range can be estimated from historical ignition data and weather features through a data-driven uncertainty set.
Effectiveness of PSPS and fast-trip
PSPS eliminates ignition risk on a de-energized circuit, while fast-trip only reduces a percentage of ignition risk on a protected circuit.
Optimal PSPS reaction
PSPS decisions are made optimally after ignition risk is realized (or when it can be well predicted), but must conform to a system-level reliability constraint shared with fast-trip devices deployed proactively.
Planning budgets
Limits on how much of the system can adopt sectionalization or fast-trip infrastructure.
Sect. budget (% of circuits)
The maximum share of all circuits that can be sectionalized—prepared so they can be shut off in segments during a PSPS event instead of all at once. A higher budget allows more granular shutoffs (fewer customers affected per event) but requires more upfront investment and coordination.
Fast-trip budget (% of circuits)
The maximum share of all circuits that can be assigned fast-trip settings—operational settings that reduce ignition risk by tripping more quickly under fault conditions. More circuits with fast-trip can reduce ignition risk but may increase momentary outages.
Reliability constraint and mitigation impact
Parameters that determine how PSPS and fast-trip actions count toward the system-wide reliability cap.
SAIFI Cap
A cap on the average number of interruptions per customer from PSPS and fast-trip actions combined. For example, SAIFI Cap = 0.2 means a California customer should not experience more than 0.2 wildfire-mitigation outages in the planning year. Annual system SAIFI is ~1 by industry standard, so we consider a range of [0.1, 0.3] for mitigation-only interruptions.
Reliability Impact of PSPS
If circuit i serves hi customers, one PSPS event contributes (hi / total system customers) to SAIFI. Since one PSPS action causes a single outage to all customers on that circuit, we use 1 to represent its relative per-event contribution to SAIFI.
Reliability Impact of Fast-Trip
Fast-trip contributes (hi × expected activations per circuit-year / total customers) to SAIFI—discounted relative to PSPS because activations per year are typically less than 1. Based on 2024 PG&E data (~2,500 EPSS events across ~3,000 circuits), this averages ~0.83 per circuit-year, so we use a range of [0.8, 0.9, 0.95].
Physical constraints
Parameters that link mitigation choices to their ignition risk reduction.
Effectiveness of fast-trip (% of mitigation)
How effective fast-trip (EPSS) is at controlling ignition risk—the degree to which enabling fast-trip on a circuit reduces its ignition probability. Higher effectiveness means the model treats fast-trip as a stronger mitigation lever.
Hierarchical uncertainty set and coverage guarantee
For a given planning window, we build a hierarchical polyhedral uncertainty set that encodes plausible future circuit-level ignition counts. The set imposes bounds on each circuit and on the sum of counts within each group (e.g., districts or climate zones).
For a prescribed level α ∈ (0, 1), we construct the set so that the probability the realized outcome falls inside the set is at least 1 − α—controlling the chance that the true outcome lies outside the predicted set to at most α. In practice, our method controls the error rate of the predicted uncertainty set for all circuits and groups to (at most) 0.5.
Key insights
Coordinated planning outperforms separation
Coordinating long-term infrastructure planning with short-term operational decisions can outperform treating them as separate problems.
Flexibility enables targeted interventions
Infrastructure configurations that improve operational flexibility enable more targeted interventions during high-risk conditions, reducing unnecessary shutoffs.
Risk reduction within reliability limits
The coordinated approach can reduce wildfire ignition risk while still respecting system-wide reliability constraints on customer interruptions.
Calibrated uncertainty avoids conservatism
Better-calibrated uncertainty sets avoid unnecessary conservatism, leading to more effective plans that act where risk is truly elevated.
Optimization Results
Evaluation
Key insights
Coordinated planning outperforms separation
Coordinating long-term infrastructure planning with short-term operational decisions can outperform treating them as separate problems.
Flexibility enables targeted interventions
Infrastructure configurations that improve operational flexibility enable more targeted interventions during high-risk conditions, reducing unnecessary shutoffs.
Risk reduction within reliability limits
The coordinated approach can reduce wildfire ignition risk while still respecting system-wide reliability constraints on customer interruptions.
Calibrated uncertainty avoids conservatism
Better-calibrated uncertainty sets avoid unnecessary conservatism, leading to more effective plans that act where risk is truly elevated.