More large blackouts in power systems have been encountered in the last few years than in any other period in history. Although the evolution of a large blackout is a complex process, it usually starts with a small triggering event that by successive events can cascade and evolve into a complete blackout. Therefore, hardening power systems against such events and improving power system reliability are of necessity. Several reliability indices for composite systems have been introduced such as loss of load probability, loss of load frequency, loss of load expectation, expected energy not supplied, expected demand not supplied, etc. Although these indices are of a great importance in both planning and operational power system reliability evaluation, they lack the ability to identify the influence of each area or equipment on system reliability. Moreover, these indices cannot determine the most effective way to improve the system in order to keep it reliable. Due to these reasons, significant efforts have been devoted to this area in recent years to evaluate what the system's reliability justifications are and to determine the best location in which to invest.One of the promising methods that can identify the most vulnerable component(s), bus(es), or area(s) is performing sensitivity analysis of the possible load shedding with respect to component parameters and operating limits. The sensitivity analysis provides a measure of the amount of load to be curtailed in response to the violation of the operating limits or component characteristics.The task of performing sensitivity analysis is amply used in power system reliability planning and operational studies. However, the application of sensitivity analysis in the changing operating environment such as the newly imposed emission constraints, penetration of the renewable energy sources and cascading failures in power systems has not been given much attention. Sensitivity analyses of the reliability indices can be conducted either analytically such as using state space enumeration or by means of simulations. Therefore, performing sensitivity analyses over a large set of variables for large systems is computationally expensive and often intractable.The work presented in thesis develops and proposes new expressions for sensitivity analysis that can be used to identify the most vulnerable components in power systems. Also, this work proposes a heuristic technique in conjunction with the population-based intelligent search methods (PIS) to reduce the computation burden. The work in this thesis is divided into three parts. Part I presents the concept of the sensitivity analyses and provides the mathematical derivations of the proposed expressions. Also, this part introduces new techniques to reduce the computational time and burden based on PIS methods and their applications in power system reliability evaluation. Furthermore, it introduces a new technique to classify the search space into success, failure and unclassified subspaces. Part II presents the application of the derived expressions on power system reliability planning studies such as the effect of each parameter on system reliability and the effect of emission constraints on power system reliability. Also, this part examines the proposed algorithms to speed up the computational time against the existing techniques. Part III conducts the application of the sensitivity analysis in power system reliability operation such as mitigation of cascading failures and prevention of the unfolding events. The expressions developed in part I were successfully applied on power system planning studies. The results were validated using different methods. Also, the proposed techniques to reduce the computation time and burden have shown very promising results. The results of the sensitivity analysis of the planning studies of composite power systems as well as the efficiency of the proposed computation reduction techniques have been published in several research papers.
Read
