Risk-averse microgrid reconfiguration under demand uncertainty and storage degradation

Microgrids must decide where to place, how to configure, and how to operate distributed energy storage while facing uncertain demand and technology degradation; yet existing plans often ignore either risk or degradation, leading to shortages or overbuilt systems. This study tackles the joint planning–operation problem of resilient microgrid reconfiguration under uncertainty and energy decay, developing a risk-averse optimization model that captures demand variability and technology-specific, time-dependent storage losses. The study proposes a two-stage stochastic programming framework with Conditional Value-at-Risk (CVaR) applied to second-stage operational costs to hedge tail outcomes arising from extreme demand scenarios. The first stage selects infrastructure (i.e., storage site activation, technology assignment, and renewable-to-storage routing) subject to capacity and reconfiguration limits. The second stage, cast as a capacitated lot-sizing problem, optimizes replenishment and dispatch given scenario-based demand and degradation dynamics. To ensure tractability, a divide-and-conquer procedure is introduced that exploits structure in reconfiguration choices. A real-world case study on the Toronto Green Microgrid Initiative validates the approach. Results show that integrating storage strategically and operating it with risk aversion reduces unmet demand by up to 70% in high-risk scenarios, while storage supplies over 50% of total load in multiple cases and enables effective use of surplus renewables. These findings offer a practical pathway to configure and operate microgrids that are cost-efficient on average and robust to tail events, especially for small, remote, and off-grid communities operating amid multiple uncertainties.