Resilience-Oriented Intelligent and Decentralised Restoration Framework for Modern Power Distribution Networks

Abstract

Modern power distribution networks (MPDNs) are increasingly exposed to rare events such as extreme weather, which challenge traditional restoration strategies based on centralised control and limited mobile power sources (MPSs). These disruptions, combined with rising electrification and dependence on digital infrastructures, intensify the urgency for adaptive, scalable, and resilient restoration frameworks. Existing approaches are constrained by reliability criteria, insufficient modelling of interdependencies with transport and charging systems, and the limited availability of utility-owned MPSs. Addressing these challenges requires innovative solutions that harness distributed, individually owned MPSs such as residential electric vehicles (EVs) and, for the first time, electric boats (E-Boats).

This thesis developed a series of data-driven frameworks that significantly advance the resilience of MPDNs under extreme disruptions by coordinating residential EVs and E-Boats for post- disaster restoration. First, a spatial–electrical modelling and resilience assessment framework uniquely coupled a modernised IEEE 123 Test Feeder with real-world electric vehicle charging points (EVCPs) and synthetic onshore charging points (OSCPs), providing a realistic environment to evaluate MPDN restoration under extreme disruptions. Second, a two-stage restoration framework was proposed for active residential EV coordination, combining advanced proactive prepositioning with adaptive spatiotemporal dispatch, supported by a novel bidirectional geographic graph (BGG) and an integrated information system. Third, the framework was extended into a behaviour-aware, scalable coordination model, the Individually Owned EV Integration Framework (IIIF), which incorporated probabilistic user participation modelling, multi-layer dynamic clustering, and unified prepositioning–dispatch optimisation to manage large, non- predetermined residential EVs. Fourth, a multimodal restoration framework integrated EVs with E- Boats through a novel selection–pairing algorithm, EV routing, advanced E-Boat voyage and navigation models using maritime data, and advanced scheduling methods. This represented the first demonstration of utilising residential EVs and E-Boats exclusively as active assets for grid restoration, while ensuring effective power delivery when road systems were damaged or congested.

Simulation results across these frameworks demonstrated accelerated restoration and enhanced resilience: first, proactive prepositioning supplied critical loads as-soon-as-practical after events and reduced restoration time compared with utility-owned MPS baselines. Second, congestion- and accessibility-aware BGG routing shortened travel distance, increased delivered energy per trip, and raised EVCP utilisation by reducing queuing and idle time. Third, the behaviour-aware IIIF increased effective participation of eligible EVs and scaled to large non-predetermined EV fleets through multi-layer dynamic clustering with rapid optimisation convergence. Fourth, multimodal EV/E-Boat coordination restored road-inaccessible load areas via waterways, increased coverage under severe damage, and lowered restoration cost and energy consumption by minimising unnecessary travel and maximising use of stored battery energy. Across all comparative scenarios, these contributions delivered faster full-load restoration, higher restored load trajectories, and more balanced power allocation than existing methods.

Collectively, the contributions establish a new paradigm in resilience-oriented restoration by uniting geospatial modelling, user behaviour considerations, scalable EV clustering, and the first integration of E-Boat deployment into MPDN resilience studies. They provide operators and planners with effective tools to strengthen restoration capability, shorten outage durations, and align restoration with Net Zero 2050 decarbonisation goals. By bridging technical innovation with practical application, this thesis sets a benchmark for adaptive, decentralised, and multimodal restoration strategies, advancing both knowledge and practice in resilient MPDNs.