Modeling and optimization for resilience enhancement of multi-energy systems

Abstract

Extreme natural disasters in recent years have significantly disrupted energy supplies, leading to enormous economic and social losses. Resilience can be described as a system's capacity to prepare for and adapt to changing conditions and withstand and recover rapidly from a disruption. As one of the most important techniques, this thesis focuses on the restoration aspects to enhance the system resilience. Specifically, considering the increasing interdependence among various energy systems, technical issues regarding resilience enhancement in integrated power-transportation systems and integrated power-heat-transportation systems are discussed in this thesis. As for the integrated power-transportation system, firstly, this thesis proposes an optimal service restoration method with the coordination of transportable power sources (TPSs) and repair crews (RCs), considering the uncertainty of traffic congestion. Besides, with the growing integrations of renewable energy sources (RES) and energy storage systems (ESS), this thesis also investigates the cooperation between these stationary resources (e.g., RES, ESS) and transportable resources (e.g., TPS, RC) for aiding in multi-energy system resilience considering the uncertainty impacts. An adaptive progressive hedging (A-PH) algorithm is developed to decompose and solve the proposed model efficiently. Then with further consideration of the heat energy in the integrated power-heat-transportation system, a joint network reconfiguration model is proposed for more flexible and efficient multi-energy system restoration. Via power-to-thermal conversion units, an on-emergency TPS scheduling model is proposed to offer supporting effects for both power and heat load recovery. Moreover, a further study considering coordination between the dynamic repair process and the joint network reconfigurations for resilience enhancement under multiple sources of uncertainties is performed. When the binary variables exist in both the first and second stages, the Progressive Hedging (PH) algorithm cannot always promise an optimum. Thus, a novel Penalty-Based Gauss-Seidel (PBGS) decomposition method is developed and utilized to efficiently solve the stochastic mixed integer program.