Understanding key factors affecting power systems resilience. (August 2021)

• Record Type: Journal Article
• Authors: Shen, Lijuan; Tang, Yanlin; Tang, Loon Ching
• Is Part Of: Reliability engineering & system safety. Volume 212(2021)
• Publication Date: 2021-08
• Subjects: Feature selection -- Power systems -- Severe weather -- Performance loss -- Recovery time; Reliability (Engineering) -- Periodicals; System safety -- Periodicals; Industrial safety -- Periodicals; Fiabilité -- Périodiques; Sécurité des systèmes -- Périodiques; Sécurité du travail -- Périodiques; 620.00452
• DOI: 10.1016/j.ress.2021.107621
• Languages: English
• ISSNs: 0951-8320

Abstract

Abstract: In this paper, we study the key factors that impact on power systems resilience under severe weather-induced disruptions from three dimensions: the extrinsic disruptions, the intrinsic capacities of a system and the effectiveness of recovery. Using 12 years of historical blackout data from 2007 to 2018 in the U.S., we apply various group selection and bi-level selection methods to identify the key predictor groups as well as factors within-group that affect power system resilience. After deleting the predictors which are fully or highly correlated with others, we split the remaining 39 candidate predictors into 8 natural groups and consider the number of customers affected and the recovery time as response variables. To ensure stability of the selection process, we adopt the random subsampling method to rank the importance of the groups and key predictors. It is found that the disruption types from the extrinsic disruptions dimension have a significant impact on the resilience of power systems, especially for the hurricanes with high scales. From the intrinsic capabilities dimension, the demographic group has a large impact on the number of customers affected. The number of customers affected tends to be large in highly urbanized areas with large population. From the effectiveness of recovery dimension, the group of economics is top selected for the recovery time. It is found that the power system tends to be more resilient with a better economic health. Feature selection under quantile regression is also conducted as the histograms show that the distributions of the responses are skewed and heavy-tailed. It is found that the recovery time is also greatly affected by the investment on the compliance and enforcement program from the North American Electric Reliability Corporation. In summary, our analysis provides interesting insights for understanding power system resilience and developing strategies to enhance the resilience. Highlights: Look into extrinsic disruptions, intrinsic capacities and effectiveness of recovery. Analyse 12 years of historical blackout data from 2007 to 2018 in the U.S.. Apply group selection and bi-level selection to identify key groups and predictors. Use the random subsampling to rank the importance of the groups and predictors. Provide insights for understanding and enhancing power system resilience.