Topology-informed risk assessment in a power grid microgrid network

Thesis Abstract

This study addresses the escalating risk exposure of contemporary microgrid networks within urban distribution systems, where topology-driven connectivity and operational constraints amplify vulnerability to weather events, equipment failures, and cyber-physical threats. The aim is to develop a topology-informed risk assessment framework that integrates network structure with dynamic operational data to quantify and mitigate resilience gaps in microgrid configurations. Specific objectives include (1) mapping the microgrid’s topology to identify critical links and nodes using spectral graph measures and betweenness centrality; (2) developing a composite risk index that fuses probabilistic failure models with topology-derived vulnerability metrics; (3) evaluating the impact of islanding strategies and reconfiguration plans on systemic risk under stochastic fault scenarios; (4) validating the framework against real-world microgrid data to demonstrate predictive accuracy and decision-support utility; and (5) deriving strategic recommendations for grid operators to enhance resilience while maintaining reliability and economic efficiency. Methodologically, the study adopts a positivist research design employing quantitative analysis. The population comprises operational microgrids within a metropolitan distribution network, with a purposive sample of 12 microgrids that provide high-resolution topology data, reliability histories, and weather-correlated outage records. Data sources include SCADA logs, protective relay event archives, weather station feeds, and asset-level maintenance records from the utility operator, supplemented by publicly available feeder topology and load profile datasets. Instruments consist of a structured data extraction protocol, a topology profiling toolkit, and a Risk of Catastrophic Failure (RCF) scoring spreadsheet calibrated to the IEEE 1547 framework. The study ensures instrument validity through content validation with domain experts and reliability via test-retest procedures, achieving a composite reliability score above 0.85. Data analysis proceeds in four stages. First, topology characterization uses spectral clustering, Laplacian eigenmaps, and centrality measures to identify critical components and potential single points of failure. Second, a probabilistic failure model is constructed, integrating component failure rates, repair times, and cascading failure propagation probabilities, to simulate outage scenarios via Monte Carlo sampling (10,000 iterations per microgrid). Third, a composite topology-informed risk index (TIRI) is derived by combining scenario-based outage severity with topology-derived vulnerability weights through a hierarchical Bayesian framework, enabling probabilistic risk estimation under varying islanding and reconfiguration strategies. Fourth, results are interpreted in light of resilience theory, particularly the concepts of robustness, redundancy, and adaptive capacity, with model performance evaluated against historical disturbances using RMSE and AUC metrics for predictive discrimination. Expected findings indicate that topology-aware risk measures will outperform conventional reliability indices in forecasting cascading outages and identifying marginally resilient network configurations. It is anticipated that identifying high-betweenness links and poorly connected islands will reveal critical upgrade candidates, and that optimized islanding strategies will significantly reduce expected loss-of-load and recovery times under extreme weather scenarios. The study also expects to show that modest reconfiguration, informed by topology, can achieve substantial risk reductions at a lower cost than large-scale hardware reinforcement. The contribution to knowledge lies in operationalizing a replicable, topology-informed risk framework tailored for microgrids, bridging network science with reliability engineering, and providing a scalable decision-support tool for utilities. The research advances theoretical integration of spectral graph theory with probabilistic risk modeling in energy systems and offers practical guidance on prioritizing protective measures, maintenance, and reconfiguration policies. The study concludes that incorporating topology into risk assessment yields actionable insights for enhancing microgrid resilience without compromising service quality, and recommends the adoption of topology-driven maintenance scheduling, targeted infrastructure upgrades at critical links, and the development of adaptive islanding policies to bolster future grid security.