Inter-area Oscillations in Power Systems PowerElectronicsandPowerSystems SSeerriieessEEddiittoorrss:: M.A.Pai AlexStankovic UniversityofIllinoisatUrbana-Champaign NortheasternUniversity Urbana,Illinois Boston,Massachusetts Inter-areaOscillationsinPowerSystems:ANonlinearandNonstationaryPerspective ArturoRomanMessina,ed. ISBN978-0-387-89529-1 RobustPowerSystemFrequencyControl HassanBevrani ISBN978-0-387-84877-8 SynchronizedPhasorMeasurementsandTheirApplications A.G.PhadkeandJ.S.Thorp ISBN978-0-387-76535-8 DigitalControlofElecticalDrives SlobodanN. Vukosavic´ ISBN978-0-387-48598-0 Three-PhaseDiodeRectifierswithLowHarmonics PredragPejovic´ ISBN978-0-387-29310-3 ComputationalTechniquesforVoltageStabilityAssessmentandControl VenkataramanaAjjarapu ISBN978-0-387-26080-8 Real-TimeStabilityinPowerSystems:TechniquesforEarlyDetectionoftheRiskofBlackout SavuC.Savulesco,ed. ISBN978-0-387-25626-9 RobustControlinPowerSystems BikashPalandBalarkoChaudhuri ISBN978-0-387-25949-9 AppliedMathematicsforRestructuredElectricPowerSystems:Optimization,Control,andCom- putationalIntelligence JoeH.Chow,FelixF.Wu,andJamesA.Momoh,eds. ISBN978-0-387-23470-0 HVDCandFACTSControllers:ApplicationsofStaticConvertersinPowerSystems VijayK.Sood ISBN978-1-4020-7890-3 PowerQualityEnhancementUsingCustomPowerDevices ArindamGhoshandGerardLedwich ISBN978-1-4020-7180-5 ComputationalMethodsforLargeSparsePowerSystemsAnalysis:AnObjectOrientedApproach S.A.Soman,S.A.Khaparde,andShubhaPandit ISBN978-0-7923-7591-3 ContinuedafterIndex Arturo Roman Messina Editor Inter-area Oscillations in Power Systems A Nonlinear and Nonstationary Perspective 1 3 Editor ArturoRomanMessina CentrodeInvestigacio´ny deEstudiosAvanzados delIPN Guadalajara,Mexico [email protected] ISBN978-0-387-89529-1 e-ISBN978-0-387-89530-7 DOI10.1007/978-0-387-89530-7 LibraryofCongressControlNumber:2008939222 #SpringerScienceþBusinessMedia,LLC2009 Allrightsreserved.Thisworkmaynotbetranslatedorcopiedinwholeorinpartwithoutthewritten permissionofthepublisher(SpringerScienceþBusinessMedia,LLC,233SpringStreet,NewYork, NY10013,USA),exceptforbriefexcerptsinconnectionwithreviewsorscholarlyanalysis.Usein connectionwithanyformofinformationstorageandretrieval,electronicadaptation,computer software,orbysimilarordissimilarmethodologynowknownorhereafterdevelopedisforbidden. Theuseinthispublicationoftradenames,trademarks,servicemarks,andsimilarterms,evenifthey arenotidentifiedassuch,isnottobetakenasanexpressionofopinionastowhetherornottheyare subjecttoproprietaryrights. Printedonacid-freepaper springer.com Preface The study of complex dynamic processes governed by nonlinear and nonstationary characteristics is a problem of great importance in the analysis and control of power system oscillatory behavior. Power system dynamic processes are highly random, nonlinear to some extent, and intrinsically nonstationary even over short time intervals as in the case of severe transient oscillations in which switching events and control actions interactinacomplexmanner. Phenomenaobservedinpowersystemoscillatorydynamicsarediverseand complex. Measured ambient data are known to exhibit noisy, nonstationary fluctuationsresultingprimarilyfromsmallmagnitude,randomchangesinload, drivenbylow-scalemotionsornonlineartrendsoriginatingfromslowcontrol actionsorchangesinoperating conditions.Forcedoscillations resultingfrom majorcascadingevents,ontheotherhand,maycontainmotionswithabroad rangeofscalesandcanbehighlynonlinearandtime-varying. Predictionoftemporaldynamics,withtheultimateapplicationtoreal-time systemmonitoring,protectionandcontrol,remainsamajorresearchchallenge duetothecomplexityofthedrivingdynamicandcontrolprocessesoperating on various temporal scales that can become dynamically involved. An understanding of system dynamics is critical for reliable inference of the underlying mechanisms in the observed oscillations and is needed for the developmentofeffectivewide-areameasurementandcontrolsystems,andfor improvedoperationalreliability. Complex power system response data can contain nonlinear and possibly strong local trends, noise, and may exhibit sudden variations and other nonlineareffectsassociatedwithlargeandabruptchangesinsystemtopology or operating conditions that make the extraction of salient features difficult. Accounting for nonlinear and time-varying features can not only provide a better description of the data but can also reveal crucial information on system’s oscillatory behavior such as modal properties and moving patterns. By tracking theevolving dynamics of theunderlying oscillations, theonsetof system instability can be determined and the critical stages for analysis and controlcanbeidentified. v vi Preface Recentyearshaveseenaflourishingofactivityinvarioustechniquesforthe analysis of power system dynamic behavior. Foremost among linear analysis tools,Prony’smethodhasbeenwidelyappliedtoestimatesmall-signaldynamic propertiesfrommeasuredandsimulateddata.Applicationsoflineartechniques inthecontextofpoweroscillationsinclude,forexample,modalextractionfrom ringdowns, the analysis of dynamic tests, and the identification of transfer functions.Ongoingresearch into the study of modal behavior in the presence of high noise levels and possibly nonstationary situations has resulted in variations to these approaches that extend their practical use to the realm of near-real-time stability assessment and control, and has stimulated the development of enhanced monitoring systems. This in turn, has sparked a resurgence of interest in the development of new algorithms that use the availableonlineinformationtoestimatemodalproperties. Advancesinsignalprocessingalgorithms,alongwithcontinuouslygrowing computational resources and monitoring systems are beginning to make feasible the analysis and characterization of transient processes using real- timeinformation.Muchoftherecentworkhasbeendrivenbyinterestinnear real-time estimation of electromechanical modal properties from measured ambient data. This effort has resulted in various signal processing methods withthecapabilityoftrackingtheevolvingdynamicsofcriticalsystemmodes. Complementary, time–frequency analysis techniques that explicitly acknowledge and incorporate nonlinearity or nonstationarity in both the time and frequency domain are emerging as subjects of research and application in engineering investigations. Adaptive, nonlinear time-varying methods with the abilitytocapturethetemporalevolutionofcriticalmodalparameters,promiseto enhance our understanding of the physical mechanisms that underlie system oscillatory dynamics and have the potential to be applied to more general transientoscillations,governedbymultiscale,time-varyingprocesses. A significant element of this major thrust is the development of wide-area measurementsystems.Extractingthesalientfeaturesofinterestfromawidely dispersed and usually large number of system observations is a complex problem. In the analysis of large models, where a significant amount of observational data is available, the development of data-based statistical models with the capacity to process the vast wealth of information and extract relevant, physically independent patterns is appealing. For many of theabovedevelopments,acompleteframeworkfortemporalcharacterization ofsystembehavior,however,isstillevolving. The combined utilization of temporal, modal information and advanced measurement and control techniquesholds alsoenormous potential toprovide criticalinformationforearlydetection,mitigation,andavoidanceoflarge-scale cascading failures and could form the basis of smart, wide-area automated analysis and control systems. Analysis and characterization of time- synchronized system measurements requires mathematical tools that are adaptable to the varying system conditions, accurate and fast, while reducing thecomplexityofthedatatomakethemcomprehensibleandusefulforcontrol Preface vii and real-time decisions. Experience with the analysis of complex inter-area oscillationsfrommeasureddata,showsthatissuessuchasnoise,time-varying behavior,datameasurementerrors,andnonlineareffectshavetobeaddressed if these tools are to be of practical use. Further, the applicability of these techniquestoboth,ambientfromonlinesystemmeasurementsandlarge-scale transient oscillations has to be fully investigated because some techniques are bettersuitedforaspecifictypeofbehavior. This book deals with the development and application of advanced measurement-basedsignalprocessingtechniquestothestudy,characterization, andcontrolofcomplextransientprocessesinpowersystems.Recentadvancesin understanding, modeling and controlling system oscillations are reviewed. Specificattentionisgiventothemodelingandcontrolofcomplextime-varying (andpossiblynonlinear)powersystemtransientprocesseswhichhavenotbeen present in previous work. Techniques that explicitly address and treat nonlinearity and nonstationarity are given and efficient methods to generate time-varying system approximations from both measured and simulated data arediscussed.Attentionisalsogiventothevitalnewideasofdynamicsecurity assessment in real-time implementations and the development of smart, wide- area measurement and control systems incorporating FACTS (flexible AC transmission system) technology. Application examples include the analysis of real data collected on grids in western North America, Australia, Italy, and Mexico. These studies are expected to stimulate the interest of other researchers, toward the investigation of complex nonstationary power system oscillationsandmayformthebasisofmoreadvancedcomputationalalgorithms. Thebookisorganizedintoeightchapterswrittenbyleadingresearcherswho aremajorcontributorstoknowledgeinthisfield. Chapter1demonstratesandexaminestheperformanceofseveralmethodsfor estimating small-signal dynamic properties from measured responses. The theoretical basis for these methods is described as well as application, properties, and performance. Examples include computer simulations and actual system experiments from the western North American power grid. Analysis goals center on estimating the modal properties of the system includingmodalfrequency,damping,andshape. Chapter 2 revisits some of the fundamental assumptions of the recently introduced Hilbert–Huang transform. The ability of empirical mode decomposition (EMD) to yield monocomponent intrinsic mode functions is examined in the context of power system oscillations. Some enhancements to the EMD are proposed to enhance its ability to better discriminate between closelyspacedfrequencycomponents.Additionally,frequencydemodulationis suggested, to extract physically relevant instantaneous frequency from the Hilbert transform. Synthetic data as well as real life data are used to demonstratethevalidityoftheenhancements. Chapter 3 discusses some refinements to the Hilbert–Huang technique to analyze time-varying multicomponent oscillations. Improved masking signal techniques for the EMD are proposed and tested on measured data of a real viii Preface event in northern Mexico. Based on this framework, a novel approach to the computationofinstantaneousdampingissuggestedandalocalimplementation of the Hilbert transform is also described. The accuracy of the method is demonstratedbycomparisonstoPronyandFourieranalysis. Chapter4investigatestheapplicabilityofHilbert–Huanganalysistechnique toextractmodalinformationinthepresenceofnoiseandpossiblynonstationary situations. Application of Hilbert analysis is examined relative to the more established Prony analysis, with particular reference to the considerable structural differences which exist between the two methods. Factors affecting theperformanceofthetechniquesincludingnoisetolerance,performanceinthe case of closely spaced frequency components and changes in the underlying system dynamics are discussed and investigated using synthetic and measured data. In Chapter 5 a real-time centralized controller for addressing small-signal instabilityrelatedeventsinlargeelectricpowersystemsisproposed.Usingwide- area monitoring schemes to identify the emergence of growing or undamped oscillations related to interarea and/or local modes, rules are developed for increasing multi-Prony method’s observability and dependability. This information is then utilized to initiate static VAR compensation controls to enhancethedampingofacriticalmode;thealgorithmsaretestedinatwo-area powersystemandinalarge-scalesimulationexample. Chapter6discusestheuseofmultivariatedataanalysistechniquestoextract and identify dynamically independent spatiotemporal patterns from time- synchronized data. By seeing the snapshots of system data as a realization of random fields generated by some kind of stochastic process, a statistical approachtoinvestigatepropagatingphenomenaofdifferentspatialscalesand temporalfrequenciesisproposedandtestedonrealnoisymeasurementsfrom theMexicansystem.Themethodprovidesaccurateestimationofnonstationary effects,modalfrequency,time-varyingshapes,andtimeinstantsofintermittent transientbehavior. Chapter 7 proposes new techniques for detection and estimation of nonstationary power transients. Attention is focused on two aspects of small signalmodels:thedetectionofchangeinthesystemandtheidentificationofthe new operating parameters. Techniques to detect significant changes in system dynamicsbyanalyzingthedynamicresponsetocontinualloadchangesbased on detection theory are proposed. Approaches based on time–frequency analysis techniques are then used to yield improved modal estimates in nonstationary environments. Applications to measurement data from the Australianconnectedsystemarepresented. Finally, Chapter 8 discusses the development of advanced monitoring and control approaches for enhancing power system security. The monitoring structure is based on wavelet analysis of wide-area measurements systems targeted to extract the critical damping of critical oscillation modes. A hierarchical response-based control strategy that may incorporate FACTS technologies and special protection systems is developed and tested on a Preface ix dynamic model of the Italian interconnected system to provide effective stabilizationofcriticalmodes. Thebookisthefirstcomprehensive,systematicaccountofcurrentanalysis methods in power system oscillatory dynamics in both time and frequency domains ranging from modal analysis, to data-driven time-series models and statisticalapproaches.Theprocedurescanbeusedinvariousdisciplinesother than power engineering, including signal and time analysis, process identificationandcontrol,anddatacompressionandhaswideapplicationsto many important problems covering engineering, biomedical, physical, geophysical,andclimatedata. Thisisabookintendedforadvancedundergraduateandgraduatecourses, aswellasforresearchers,utilityengineers,andadvancedteachinginthefields ofpowerengineering,signalprocessing,andidentificationandappliedcontrol. Guadalajara,Mexico A.R.Messina