X Simpósio De Especialistas Em Planejamento Da Operação E Expansão Elétrica X Symposium of Specialists in Electric Operational and Expansion Planning

Load modelling is a perennial source of consternation. Postmortem analyses of disturbances frequently conclude that load models must be adjusted to achieve an adequate match between measured and simulated behaviour. However the development of load models that are accurate over a wide range of system conditions requires, 1) testing that is undesirable for customers, and/or 2) measurement of behaviour during infrequent disturbances. Furthermore, the stochastic nature of many loads implies load models that are accurate at one time may be inaccurate a short time later. In terms of quantitative analysis, for example matching simulations with measurements, it is absolutely clear that accurate load modelling is vitally important. But for qualitative investigations, where the aim is to assess the likelihood of a certain disturbance scenario remaining stable or causing instability, then the need for accurate load modelling is much reduced. The paper explores this conjecture through the use of trajectory sensitivity concepts. It is shown that for qualitative assessment, it is more important to know the sensitivity of behaviour to load parameters, than to precisely know the parameter values. However most power system failures are not initiated by instability, but rather by reactionary (unanticipated) protection operation. The ability to predict operation of protection devices, and hence potential protection-induced cascading failures, may be greatly affected by load model fidelity.