The vast size of a modern interconnected power grid precludes controlling and operating it as a single object. Subdividing a power grid into a number of internally coherent areas allows to cope with its inherent complexity and to enable more efficient control structures. This thesis focuses on discovering the power system structure to facilitate the definition of control areas for wide-area monitoring, protection and control (WAMPAC) applications. Graph partitioning is a well-developed discipline whose potential is not fully recognized in the power system domain. Particularly, spectral graph partitioning methods are shown to be very promising. Their efficiency is first demonstrated by accurately selecting the number and extent of control zones for secondary voltage control (SVC). Next, it is shown that grouping generators with similar slow rotor angle dynamics can also be efficiently tackled through spectral graph partitioning. The final topic is constrained graph partitioning subject to node grouping constraints, which is related to intentional controlled islanding (ICI). As both solution time and accuracy are critical for ICI, a new polynomial-time heuristic algorithm is proposed that is more accurate than comparable state-of-the-art methods.
Original languageEnglish
QualificationDoctor of Philosophy
Awarding Institution
Award date31 Mar 2020
Print ISBNs978-94-6384-116-0
Publication statusPublished - 2020

    Research areas

  • Dynamic model reduction, generator aggregation, intentional controlled islanding, number of clusters, power network partitioning, secondary voltage control, slow coherency

ID: 71067330