Resilience analysis of multistate energy system with time-dependent behaviors
Résumé
Resilience of modern energy systems has attracted more and more attentions in recent years, as modern energy systems are vulnerable to various disruptive events (e.g., hurricanes, windstorms, earthquakes, intentional cyber attacks). Most of the existing resilience quantification methods assume that the performance level of the system is continuous. A lot of practical energy systems, however, are multistate in nature or needs to be modeled by multistate models to control the modelling/computational complexity (Levitin, 2011). There are only a few works regarding quantifying resilience using multistate models. Thekdi and Chatterjee (2019) developed a multistate hidden Markov model to quantify resilience of infrastructure systems and discussed decision making based on the model. Nuss et al. (2017) used a continuous time discrete state Markov model to quantify the resilience against disruptions considering the multistate engineering system behaviors of unmanned surface vessels to support system design selection during trade studies. In a previous research of the authors (Du et al., 2019), a Markov reward process-based model is developed for quantifying resistant, absorption, recovery and overall resilience. All the above mentioned works are based on Markov (or hidden Markov) models. A significant drawback of these Markov model-based methods is that they have to assume that the transition times among states follow exponential distributions. Practical systems, however, exhibit a variety of time-dependent behaviors, which cannot be adequately described using only exponential distributions. In this paper, we develop a Semi-Markov reward process-based approach to quantify resilience of multistate systems, considering the time-dependent behaviors of both the pre-disruption protection processes and the post-disruption recovery processes. Four numerical metrics are defined to quantify resilience based on the developed model. The model is applied to analyze the resilience of a crude oil storage tank farm.