1. Modelling and validation techniques for bottom-up housing stock modelling of non-heating end-use energy in England
- Author
-
Lorimer, S. W., Lowe, R. J., Steadman, J. P., and Bruhns, H. R.
- Subjects
697 - Abstract
This thesis engages with different methods and validation techniques for bottom-up stock modelling of non-heating end-use energy of the residential sector. These end-uses are not the primary focus of current domestic energy models, and there is a unique opportunity to use actual electricity use data to build and validate models as electricity becomes exclusively used for these end-uses in England. The first contribution to knowledge is the creation of a validation set from aggregated electricity use data that has become available from small census areas of around 600 households using only areas with minimal estimated rates of electric heating. The second contribution is a method for using partial data from recent housing and energy surveys to update complete, but dated surveys by using household size and seasonal distributions. This enables a yearly updated model validated against actual aggregate energy use. This led to an annually updateable single-level model of non-heating end-use energy based on the predictors of household size measured by the number of rooms and the number of occupants. This uses linear regression on a square-root transformation of energy instead of the current natural logarithm transformation. The model is found to have a slight over-prediction (1.5%) of energy use when validated. The final contribution is an alternative approach where the model was allowed to vary on the household’s area. A hierarchical linear model of domestic energy was built based on 20 area classifications. There is a weak, but significant effect of additional energy use in households located inside area classifications with higher mean household sizes. This effect is highly significant when building age is taken into account. Although validation was difficult because building age data is limited, this result points to a neighbourhood-level influence that explains energy use beyond individual household size if precise location data can be made available.
- Published
- 2012