Stochastic optimization of distributed generator location and sizing in an islanded utility microgrid during a large-scale grid disturbance

Abstract

Natural disaster-prone locations can experience widespread power outages when the disasters cause large-scale damage to the existing power grid. Such damages can hinder a utility company’s ability to provide power to essential public resources (e.g., hospitals, grocery stores, fire, police and gas stations) of the utility’s serviced area. Backup generators, which provide power to these essential public resources during the outage, have limited capacities and can suffer from such failures as well. Microgrids, defined as localized power grids that can operate independent of the main grid, can help utilities provide disaster relief power supply to the essential public resources to improve their resiliency during the outage. This research investigates a multi-source capacitated facility location coverage problem (MS-CFLCP) that models a utility-owned microgrid that is operating independent of the main grid due to a large-scale main grid disturbance. The developed MS-CFLCP optimizes the location, sizing, assignment and the number of distributed generators (DGs) within the utility-owned microgrid, and aims to minimize the following total costs of the microgrid: (1) investment costs; (2) operation and maintenance costs; (3) distance traveled for power supply costs; (4) the unmet demand penalty costs; and (5) excess DG penetration penalty costs. The MS-CFLCP is solved with two-stage stochastic programming while considering the uncertainty in DG power output and essential resource demand. A budget constraint is included to capture practical financial considerations of the utility company when establishing the microgrid. We apply the model to a case study, using solar/photovoltaic-based DGs, to show its effectiveness and benefit to utilities.