Abstract/Summary

The residential sector accounts for 24% of electricity conswnption in Botswana. This is significant and warrants a comprehensive understanding of the energy characteristics of the sector. This would not only inform occupants, but it would also inform the utility providers and hence policymakers. The comprehensive understanding of household electricity conswnption requires detailed data which is virtually non-existent in Botswana. In this research, data that was collated to address this shortfall has been achieved through the use of questionnaires, appliance use diaries and electricity monitoring. The data collection was in two phases: Phase 1 and Phase 2. Phase 1 covered three location while Phase 2 focused on a locality in Gaborone. The questionnaire resulted in a comprehensive dataset of 73 households for Phase 1 and 310 households for Phase 2 while three and fourteen houses were monitored for Phase 1 and 2 respectively. The research, therefore, makes use of a mixed method (qualitative and quantitative) approach all of which complement each other. The data collated was used for statistical analysis to discover the existing trends in the data and to determine household typologies. Two statistical models; General Linear Model (GLM) and Linear Mixed Model (LMM) were developed to determine the most influential factors; dwelling, socio-economic and appliance on electricity conswnption. In addition, identification of household typologies was carried out using the k-means clustering method. From the analysis, it was observed that dwelling, socio-economic and appliance factors account for 44%, 33% and 45% of the variation in electricity conswnption respectively. A combination of all the factors, however, accounted for 57% of the variance in electricity conswnption, hence a better predictor of household electricity use. The analysis also indicated other significant factors such as the time of the day, the day of the week and occupancy trends amongst others have a significant influence on household electricity use. From the cluster analysis, four independent groups were identified based on selected characteristics (features) that is, dwelling type, tenure, the nwnber of rooms, the number bedrooms, annual electricity conswnption and the number of appliances. The clusters identified provide a potential to better understand the underlying electricity conswnption characteristics provided by different household segments. In this way, it is possible to ensure that interventions (such as the demand side management strategy already in place) will encompass as much of the population as possible, and certainly, those groups which offer the greatest potential for beneficial impact together with the building and appliance factors that underpin these potentials. Whereas at the onset of this research there were limited information/data, the fmdings presented in this research provides significant data and analyses for further research, which is underpinned by, detailed methods and modelling techniques.