[en] We developed a physics based bottom-up model to estimate annual housing stock energy consumption at a local community level (Census Collection District—CCD) with an hourly resolution. Total energy consumption, including space heating and cooling, water heating, lighting and other household appliances, was simulated by considering building construction and materials, equipment and appliances, local climates and occupancy patterns. The model was used to analyse energy use by private dwellings in more than five thousand CCDs in the state of New South Wales (NSW), Australia. The predicted results focus on electricity consumption (natural gas and other fuel sources were excluded as the data are not available) and track the actual electricity consumption at CCD level with an error of 9.2% when summed to state level. For NSW and Victoria 2006, the predicted state electricity consumption is close to the published model (within 6%) and statistical data (within 10%). A key feature of the model is that it can be used to predict hourly electricity consumption and peak demand at fine geographic scales, which is important for grid planning and designing local energy efficiency or demand response strategies. - Highlights: ► We developed a physics-based model to estimate housing stock energy consumption. ► House type and vintage, family type and occupancy time were considered. ► The model results are close to actual energy consumption at local community level. ► Its’ results agree well with the published model and statistical data at state level. ► It shows the model could provide from hourly to annual residential energy consumption.