Co-Optimization of Damage Assessment and Restoration: A Resilience-Driven Dynamic Crew Allocation for Power Distribution Systems

TL;DR

This paper addresses rapid resilience-Driven restoration of power distribution networks after severe events by proposing a mixed-integer linear programming framework that co-optimizes feeder patrolling, damage assessment, fault isolation, repairs, and load re-energization. The core contribution is a dynamic, adaptive decision framework that updates crew dispatch as fault locations and repair times are revealed, paired with a conservative two-scenario power-flow formulation that bounds voltages and line flows across intermediate configurations. Key innovations include integrated damage assessment with patrol routing, explicit handling of manual switches, and a time-aware charging/restoration sequence that preserves radial network topology. Numerical experiments on IEEE 123-node and 8500-node feeders demonstrate the model’s scalability, reduced outage costs, and comparable accuracy to conventional power-flow approaches, suggesting practical applicability for large real-world systems with resilience objectives.

Abstract

This study introduces a mixed-integer linear programming (MILP) model, effectively co-optimizing patrolling, damage assessment, fault isolation, repair, and load re-energization processes. The model is designed to solve a vital operational conundrum: deciding between further network exploration to obtain more comprehensive data or addressing the repair of already identified faults. As information on the fault location and repair timelines becomes available, the model allows for dynamic adaptation of crew dispatch decisions. In addition, this study proposes a conservative power flow constraint set that considers two network loading scenarios within the final network configuration. This approach results in the determination of an upper and a lower bound for node voltage levels and an upper bound for power line flows. To underscore the practicality and scalability of the proposed model, we have demonstrated its application using IEEE 123-node and 8500-node test systems, where it delivered promising results.

Figures (16)

• Figure 1: Task distribution for repair crews
• Figure 2: Chronological description of the model
• Figure 3: High-level description of the constraints
• Figure 4: Passive loading condition
• Figure 5: Active loading condition