Group and Flow-Based Centralities for Power Grids

Speaker(s)

Marcin Waniek

Affiliation

University of Warsaw

Language of the talk

English

Date

Oct. 22, 2026, 12:15 p.m.

Room

room 4060

Seminar

Seminar Algorithmic Economics

Identifying the power-grid components whose failure can cause the greatest disruption is challenging, especially when several outages interact in ways that simple node-by-node analyses cannot capture. In this talk, I will discuss centrality-based approaches that go beyond classical topology-only measures by incorporating electrical information, groups of nodes, and generator-load relationships. We compare a broad range of classical and electrical centralities, introduce group variants of reactance- and current-flow-based measures, and consider source-target versions of betweenness restricted to generator-load pairs. We evaluate these methods on benchmark power grids under generator and node attacks. The results show that group centralities are particularly useful when removals change network topology, while source-target restrictions often improve attack effectiveness and can substantially reduce computational cost. Overall, we show how power-grid-specific adaptations of centrality measures can improve the identification of critical components and component sets.