No abstract

AbdractThis paper describes an approach for experlmental determlnationof aggregatedynamlcloadslnpowersystems.Theworkismotlvated by the Importance of accurate load modelllng In voltage stability analysis. The models can be expressed In general as nonlinear dlfferential equations or eqihalentty realised In block dlagram form as lnterconnectlons of nonlinear (memoryless) functions and Pnear dynamic blocks. These components are parameterised by load Indexes and time constants. Experlmental results from tests in Southern Sweden on the ldentlflcatlon of these parameters are described. K.ywords Load modefllng, power systems. dynamlcs. voltage stablllty. InhoducWonModels for dynamlcal analyds of power systems typically have a conslstency problem. while lt issclentlflcally posjible to glve quite detaliedmodels for generators, Ines,transformers and control devices. load modelling can often only be treated on an ad hoc bads. In stabllltyanalysis for Instance. we need a representation of effectlve power demand at high voltage buses. Thl s may Include the aggregate effect of numerous load devices such as lighting. heating and motors plus some levels of transformer tap-changing and other control devices. Buildlng up the aggregate effect by combining device characterlstlcs may not be possible. In many cases, quite slmpllfled aggregate load representations like lmpedances are used alongside detalled generator models. Thi s seems related to three research questions: .To what extent are accurate load models Important In power system stability analysis; 2.Given that derivation of aggregate models from component characteristics is not feasible. what are appropriate mathematical structures to represent high voltage effecWe load; 3.How can the aggregate load models (from question 2) be determined in practice.Briefly. we refer to these questions as model justification. structure determination and ldentiflcation respectlvely. The present paperIs primarily concerned with model structure and ldentiflcation from SVJtem tests.The study of bad model justification In dynamlcal anaws of power systems seems to demand more attention.The usual aggregate load model expresses the real and reactive load powers as nonlinear functlons of voltage; for instance, the form P O ( 6 ) ' wlth a dngle Index a, for real load power Is popular (with a similar form for reactive power) [ 1-31. Frequency dependence can be Included. but is usually Ignored. For translent (angle) stablllty, there Is stlll some debate about whether acceptable nrst swing assessment can be donewlth lmpedanceload models.i.e. a = 2 ratherthan speclflcally determlned values [4d]. The present work Is motivated more by vdtage stablllty analyds [a. Here It Is widely acceptedthatload characterlstlcs at lowvottage play an Important role; further,dynamlcs of loadsls Important [7.8]. However, the dynamlcal descrlptlon of loads at HV buses certainly requlred more attention. For static (load flow) voltage stability analyds. It Is often assumed that the load powers are In fact constant, I.e. a = 0. I...

No abstract

No abstract

No abstract