Boosting the Power Grid Resilience to Extreme Weather Events Using Defensive Islanding

Mathaios Panteli, Dimitris N. Trakas, Pierluigi Mancarella, Nikos D. Hatziargyriou

Research output: Contribution to journalArticlepeer-review

126 Citations (Scopus)


Several catastrophic experiences of extreme weather events show that boosting the power grid resilience is becoming increasingly critical. This paper discusses a unified resilience evaluation and operational enhancement approach, which includes a procedure for assessing the impact of severe weather on power systems and a novel risk-based defensive islanding algorithm. This adaptive islanding algorithm aims to mitigate the cascading effects that may occur during weather emergencies. This goes beyond the infrastructure-based measures that are traditionally used as a defense to severe weather. The resilience assessment procedure relies on the concept of fragility curves, which express the weather-dependent failure probabilities of the components. A severity risk index is used to determine the application of defensive islanding, which considers the current network topology and the branches that are at higher risk of tripping due to the weather event. This preventive measure boosts the system resilience by splitting the network into stable and self-adequate islands in order to isolate the components with higher failure probability, whose tripping would trigger cascading events. The proposed approach is illustrated using a simplified version of the Great Britain transmission network, with focus on assessing and improving its resilience to severe windstorms.

Original languageEnglish
Article number7434044
Pages (from-to)2913-2922
Number of pages10
JournalIEEE Transactions on Smart Grid
Issue number6
Publication statusPublished - 1 Nov 2016


  • defensive islanding
  • extreme weather
  • Resilience
  • resiliency
  • smart grid
  • system splitting


Dive into the research topics of 'Boosting the Power Grid Resilience to Extreme Weather Events Using Defensive Islanding'. Together they form a unique fingerprint.

Cite this