Application of Statistical Wind Models for System Impacts

The development and characterisation of wind power time series is important for the operation and development of the UK power system in the context of high wind power penetration. There is a real and pressing need to assess the impacts of increasing amounts of wind power on the UK power system. Statistical models are presented to characterise in a more comprehensive way the temporal and spatial nature of windspeeds in the UK. Auto-Regressive Moving Average models (ARMA) , often used for predictive purposes on shorter time-scales, are developed to explore differences and similarities in windspeed and wind power data from sites across the UK, based on a categorisation by terrain type. Analysis of model parameters describing the short term memory (or persistence) and diurnal and seasonal effects has revealed differences between categories of wind farm. Vector auto-regressive (VAR) type models extend the work to allow for spatio-temporal correlations between sites. The spatial diversity that the models seek to capture is important because of the smoothing of aggregate wind power generation and this has significant implications for power system operation and in particular the calculation of loss of load probability and capacity credit. This work aims to increase the understanding of how a substantial UK wind penetration will impact on grid operation, providing a powerful tool for operational and planning purposes.