Modelling extreme peak electricity demand during a heatwave period: a case study

A frequency analysis of the occurrence of extreme peak electricity demand during the non-winter period using South African data for the period 2007–2013 is discussed in this paper. It is during the non-winter period when Eskom, South Africa’s power utility company usually carries out its maintenance of power plants. Extreme peak loads together with high unplanned and planned outages usually cause disruptions to the power plants’ maintenance programmes during this period. A detailed analysis of peak electricity demand is useful to maintain the stability of the power grid. If not properly managed this may result in negative economic impacts as electricity is one of the major drivers of any economy. Initially the data is detrended using a time-varying threshold. A sufficiently high threshold is obtained by using bounded corrected extremal mixture models. We then decluster the exceedances and fit a peaks-over-threshold model to cluster maxima. Empirical results show that demand of electricity significantly increases as a result of heat which builds up with consecutive days of extreme high temperatures. The main contribution of this paper is in the application of extreme value theory models in assessing the frequency of occurrence of extreme peak loads to assist power utilities in the planning of maintenance of generating units. The other contribution is in the development of a modelling framework which shows that as temperature converges to its upper bound, the marginal increase of electricity demand also converges. Most of the cluster maxima are experienced during the months October, November and February, suggesting that these may not be good months for scheduling maintenance of generating plants during the non-winter period in South Africa. This modelling approach may help system operators in scheduling and dispatching of electricity during the heatwave period.