| Many applications in power system operations and planning need efficient optimization methods to solve large-scale problems within a short period of time. This requirement is even more pronounced for real-time controls where fast solution speed is most important. As a major on-line application, the OPF problem is concerned with using mathematical programming methods to determine a secure and economic operating condition of power systems. The main objective of this research is, therefore, to develop and systematically evaluate advanced interior point methods for the efficient and reliable OPF solutions.; In this thesis, the OPF problem is formulated as a constrained nonlinear program in terms of all control/state variables, considering both power balance equality and security inequality constraints. Two particular OPF cases are studied in detail, namely, the real and reactive power dispatch problems. The minimization of production cost is considered as the objective in real power dispatch problems; while for reactive power dispatch problems, the objective function is the transmission active power losses to be minimized during the optimization process.; Successive linear programming is used to deal with the nonlinearity of the underlying problems. Consequently, the nonlinear OPF problem is linearized as a sequence of linear sub-problems, which are in turn solved by using interior point methods. To better suit the application of interior point methods, the sparse linear formulations are derived for both real and reactive power dispatch problems, based on decouple and couple load flow models, respectively.; The study of interior point methods is concentrated on infeasible primal-dual path-following methods. The derivations of two variants in this class of methods are presented in detail, namely, the infeasible primal-dual and the predictor-corrector primal-dual algorithms. Both algorithms are extended for a more general linear programming problem, considering lower and upper bounds for special needs in our applications. The search directions produced by these algorithms are analyzed to better understand the characteristics of interior point methods under research.; To explore the full potential of interior point methods for power engineering problems, intensive study has focused on all issues that influence the algorithm performance, such as the adjustment of barrier parameter, the determination of Newton step length and the initial point, and the use of multiple corrector steps. Practical issues related to successive linearization procedure are also investigated, including the choice of the linear step size and the tolerances for linear programming as well as for OPF procedure. Their effects on OPF performance are evaluated.; As the results of these investigations, several heuristics are proposed to reduce the number of iterations and to serve computational work in every iteration. Extensive numerical experiments have demonstrated that the OPF solution speed can be significantly improved by customizing algorithm parameters to the specific applications under concern. Finally, the use of sparse techniques is investigated in developing fast and robust interior point codes. Test results on large-scale problems have confirmed the efficiency and reliability of the algorithms. |
