À la Une

Soutenance de thèse Amin Abedi


M. Amine Abedi soutiendra en anglais, en vue de l'obtention du grade de docteur ès sciences, mention sciences de l'environnement, sa thèse intitulée:

Optimization‐based Frameworks for Systemic Vulnerability Assessment in the Electric Power Systems

Date: Mardi 30 mars 2021 à 10h00

Lieu: zoom (lien à venir)


Jury :

  • Professeur Bastien Chopard, Département d’Informatique, Thesis Director.
  • Professeure Costanza Bonadonna, Département des Sciences de la Terre, Thesis Co-director.
  • Dr Franco Romerio, Institute for Environmental Sciences (ISE), University of Geneva.
  • Dr Giorgio Tognola, Azienda Elettrica Ticinese, El Stradún 74 CH - 6513 Monte Carasso.
  • Dr. Ludovic Gaudard, Precourt Energy Efficiency Center, Stanford University, CA 94305 USA.
The purpose of this thesis is to identify the vulnerabilities of the power system to ensure its robustness and resilience. Like any other critical infrastructure (CI), power systems are subject to disruptions, either unintentional or deliberate, that may have a significant impact on their performance. Hence, to protect and mitigate such vulnerabilities when CIs suffered from an event, first, one has to explore advanced tools for modeling the electric power grid and its components with respect to its vulnerability to disruptions. Then, trying to guarantee the correct flow of electricity from generation facilities to consumers using appropriate countermeasures.

“Vulnerability analysis” in power systems is important for the first step, to determine how vulnerable a system is, and it is used to detect and rank the most critical elements of a power system under a variety of low-probability-high-consequence events such as multiple-components outages. This thesis aims to address two main issues, i.e. (i) the systemic vulnerabilities of the power system under multiple contingencies and different operational uncertainties; (ii) the critical components which must be protected or fortified when the protective resources are limited. These goals are achieved in three parts:

Part one introduces different definitions of the vulnerability concept and compares the state-of-the-art methods in this field. Then, it highlights the advantages and disadvantages of the standard methods in the vulnerability analysis. In this part, we conclude that each method possesses its own limitations and a perfect method does not exist for all circumstances. Then, we provide a guide to choose the best and the most relevant method for different power system hazards and different levels of acceptable accuracy, computational burden, and required input data.

Part two finds out the acceptable level of assumptions and available data to answer the reliability, vulnerability, and resilience questions. Afterward, the cascading failure is addressed, and a framework for the integration of security methods capable of viewing the problem from different perspectives, e.g. integrating reliability and vulnerability analyses, is also developed. Although traditionally reliability indices have been adopted as reference metrics, we show that they miss some of the key features of the security concept, especially when we have to deal with low-probability-high-consequence events. Hence, we conclude that the vulnerability analysis can complement the reliability analysis for these events. Moreover, the analysis of five different IEEE-RTS topologies shows that a system considered from the vulnerability viewpoint could behave differently compared to a system considered from the reliability viewpoint. Furthermore, we conclude that the assumptions in the power flow equations can significantly affect the final results and may lead to inaccurate predictions.

Part three introduces and develops a hierarchical leader-follower (bilevel) optimization problem where the upper level (leader) tries to maximize the damage, and the lower level (follower) tries to minimize the probable consequences. Thanks to this rational strategy, the critical components whose failures lead to the largest system loss can be determined. Afterward, our proposed model is extended to be used as a multi-period model, and as a model that immunizes the system analysis against the worst uncertainty. The proposed model is applied to the IEEE test systems and a real-life system i.e. modified Iran’s transmission network. The results show that our model is much more efficient than the previously reported one, where the approximated power flow equations are used in the lower level. Moreover, our model shows that Iran’s expanded network where only some lines are built is more robust in comparison with the existing network. At the end of this part, in order to guarantee the operational security of power systems with uncertainties, an adaptive robust trilevel optimization model for immunizing the system against the worst uncertainty has been carried out. The final one-level model has been applied to the IEEE 24-bus network and to modified Iran’s transmission network. Our simulation results show that the power system vulnerability assessment without considering uncertainties leads to optimistic results. We also observe two properties of our model and prove a lemma that improves the computational performance of our final mixed-integer linear program (MILP) model.

In summary, the focus of the present thesis is on modeling, simulation, and optimization of the power systems with respect to their vulnerability to disruptions and hazards. The outcomes of our research can provide valuable inputs and tools of analysis to public and private decision-makers and system operators.