Targeting household energy-efficiency measures using sensitivity analysis

The Community Domestic Energy Model (CDEM) has been developed to explore potential routes to reduce carbon dioxide (CO2) emissions and the model is used to predict the CO2 emissions of the existing English housing stock. The average dwelling CO2 emissions are estimated as 5827 kgCO(2) per year, of which space heating accounts for 53%, water heating for 20%, cooking for 5%, and lights and appliance for 22%. Local sensitivity analysis is undertaken for dwellings of different age and type to investigate the effect on predicted emissions of uncertainty in the model's inputs. High normalized sensitivity coefficients were calculated for parameters that affect the space heating energy use. The effects of the input uncertainties were linear and superposable, so the impact of multiple uncertainties could be easily determined. The results show that the accumulated impact on national CO2 emissions of the underperformance of energy-efficiency measures could be very large. Quality control of the complete energy system in new and refurbished dwellings is essential if national CO2 targets are to be met. Quality control needs to prioritize detached dwellings because their emissions are both the greatest and the most sensitive to all energy-efficiency measures. The work demonstrates that the uncertainty in the predictions of stock models can be large; a failure to acknowledge this can lead to a false sense of their reliability.