Energy vulnerability and the exercise of a data-driven analysis protocol: A comparative assessment on power generation aspects for the non-interconnected islands of Greece

Islands often suffer from structural handicaps and considerable energy problems, with the definition of relevant energy vulnerability metrics being both essential and challenging. On that basis, we argue that commonly used Key Performance Indicators (KPIs) are not able to capture the special and diverse characteristics that describe different island regions. Against this backdrop, a flexible data-driven protocol is proposed, which is able to generate and also properly aggregate KPIs in a meaningful and explanatory way. The protocol's ability to assess vulnerability is evaluated using the area of the Aegean Sea in Greece, engaging the numerous non-interconnected island power systems of the region. Under this exercise, we use the generated KPIs to formulate new variables, using Principal Component Analysis (PCA). The emerging principal components, which describe the energy vulnerability space, are associated with Resilience, Efficiency and Dependence and constitute the basis for computing the vulnerability metric. The components are also exploited for clustering the individual power systems into groups of different vulnerability status. Moreover, we put effort in assessing the sensitivity and adjustability of the proposed protocol as a regional-level policy tool and provide extensive discussion of the interrelations of the examined islands and their reflections on energy vulnerability.