This research report is concerned with the development of a system for forecasting residential energy demand in the Queensland electricity industry. The system will be used to improve the ease of preparation and quality of residential energy demand forecasts and to answer questions on the impacts of marketing and demand-side management activities.
The main objective of the current research work is to develop a residential end-use energy demand model for Queensland based on an analysis of the 1985/86 National Energy Survey data. This work extends the previous experience gained in developing a similar model based on the 1983 National Energy Survey.
The report presents a literature review of the various methodologies and models considered to be suitable for forecasting residential energy demand and concludes that an end-use model should be used. The review covers the theoretical and practical aspects of developing residential end-use models, ~ and the overseas and Australian experience in developing these models.
A practical residential end-use model for energy demand in Queensland has been developed. Estimates of the main parameters of the model (appliance penetration rates and appliance consumption coefficients), for 1985/86, have been produced. The appliance consumption coefficients were estimated by an econometric technique known as conditional demand analysis. The model has significant explanatory power and has been validated against comparable data obtained by other techniques and from other studies.
Difficulties were experienced with the formulation and interpretation of the survey data, with specifying the form of the model and in handling some of the statistical problems associated with using the conditional demand analysis technique. A number of outliers were removed and the model was corrected for heteroscedasticity. The final 1985/86 model has a simpler and more practical form and has greater statistical significance (R 2 = 0.73) than the 1983 model (R2 = 0.61).
Further work is required to fully explore the best use of household size and household income in the model and to integrate the model into an overall forecasting system for residential energy demand in Queensland. The additional experience and confidence gained from the current research work should be applied to redeveloping the model based on the 1983 National Energy Survey. The integration of the two surveys to create a pooled timeseries cross-section model, suitable for extrapolating appliance and household energy consumptions for load forecasting purposes, should be investigated.
The major conclusion of the research work is that conditional demand analysis is a very effective method for estimating appliance consumption coefficients. Further research work should be conducted into refining the estimates of appliance consumption coefficients and into developing an appliance penetration model for forecasting appliance penetrations. The model developed in this research study can be used as the starting point for developing an overall system for forecasting and simulating residential energy demand in Queensland.