Estimation of maximum demand on a British electricity-board system. Forecast periods of 1-3 years

The paper is concerned with the forecasting of maxium demand (m.d.) on an area board (or Central Electricity Generating Board Region) system where forecast periods of 1-3 years are important for planning the amount of, and the timing of, capital expenditure. A causal model is sought that gives a good fit to the historical data of recorded m.d. and a satisfactory performance in prediction. The primary parameters in the causal model were found to be the gross domestic product (g.d.p.), temperature and load factor. The system maximum demand (m.d.) is forecast for average-cold-spell (a.c.s.) conditions with variances due to: (a) standard deviation (s.d.) of model estimating errors (b) s.d. of the National Institute estimating errors for g.d.p. (c) s.d. of departure of temperature for a.c.s. conditions The mathematical formulation of the problem enables an estimate to be given, with a precise probability, of a given load being exceeded under any assumed set of adverse conditions. The principle that the system m.d. can be estimated from a limited number of parameters can be judged by the performance and precision of the model. The model gives a very good fit to historical data. The standard error (s.e.) is of the order of 0.6% of recorded m.d. The parameters were shown to be well chosen by their high mathematical significance, and the performance of the model in the 1973 prediction was consistent with its performance on the historical data. Confidence in the performance of the model, in prediction, was strengthened by a retrospective examination of the predictive performance of the chosen model when based on earlier data.