On the Efficiency and Accuracy of Simulation Methods for Optimal Power System Operation: Convex Optimization Models for Power System Analysis, Optimal Utilization of VSC-type DC Wind Farm Grids and FACTS Devices
2015 (English)Doctoral thesis, comprehensive summary (Other academic)
Recently, significant changes in electric power systems such as rapid developmentof smart grid and electricity market and integration of non-dispatchablesources have added more complexity to the Power Flow Scheduling (PFS) andPower Balancing (PB) models. For instance, non-dispatchable sources introducean increasing level of uncertainty in the electricity market and power system operation.One of the solutions for handling these uncertainties in the power systemoperation is the improvement of system flexibility through a more efficient operationof power systems. On the other hand, efficient operation can be achieved bywell capturing variable behavior of uncertain sources such as wind power sourceswhich in turn demands efficient and robust PFS/PB models. This way, a moreflexible system, capable of efficiently accommodating higher levels of wind powerchanges, can be achieved. All these factors increase a need for PFS/PB models suchas Power Flow (PF) and Optimal Power Flow (OPF) models which can addressthese new challenges in an efficient, reliable, and economic way while supportingthe power system operation and control. In this regard, various solution methodshave been developed for solving different forms of PF/OPF formulation. The difficultyof solving OPF problems increases significantly with increasing network sizeand complexity. One of these complexities is how to model advanced controllable devices such as HVDC grids and Flexible AC Transmission Systems (FACTS) devices.Accurate handling of these complexities has limited the use of OPF in manyreal-world applications mainly because of its associated computational challenges.The main reasons behind computational challenges are nonlinearity and especiallynon-convexity of constraints representing power system and its components. Inthis regard, OPF problems are classified into two main groups. In the first group,researchers adopt Nonlinear programming (NLP) approach to fully represent thenonlinearity of the power system for the sake of accuracy but with the cost of complexityin the model. Computational and theoretical challenges associated withNLP approaches are then used as a motivation towards developing a more simplifiedOPF model, leading to the second group of OPF models known as LinearProgramming (LP) based OPF models. LP approaches are fast, reliable, and especiallyconvex, and therefore guarantee a global optimum to the simplified OPFproblem. The problem of LP approach to OPF is that the LP solution of OPF may not even be a feasible solution of original nonlinear OPF at all. Another issueassociated with LP models is that complex power system devices such as HVDClinks are difficult to be incorporated. These limitations have restricted the applicationof LP approaches for many OPF problems. According to the mentionedadvantages and disadvantages of NLP and LP based OPF models, what we seeks isan OPF model which can have main advantages of both LP OPF models (Efficientnumerical solvers) and full AC OPF models (Results accuracy). In this thesis, wedevelop convex optimization problems which can be adopted as both PF and OPFmodels which are capable of catching the nonlinear nature of power systems asmuch as possible while can be solved by efficient solution methods such as InteriorPoint Methods (IPMs). These OPF models can incorporate HVDC links, windfarm Multi Terminal HVDC (MTDC) grids, and shunt FACTS devices.
Place, publisher, year, edition, pages
Stockholm: KTH Royal Institute of Technology, 2015. , xiv, 97 p.
TRITA-EE, ISSN 1653-5146 ; 2015:022
Engineering and Technology
Research subject Electrical Engineering
IdentifiersURN: urn:nbn:se:kth:diva-166383ISBN: 978-91-7595-573-5OAI: oai:DiVA.org:kth-166383DiVA: diva2:810658
2015-06-09, H1, Teknikringen 33, KTH, Stockholm, 10:00 (English)
Ernst, Damien, Professor
Hesamzadeh, Mohammad Reza, Associate Professor
QC 201505212015-05-212015-05-072015-05-21Bibliographically approved
List of papers