Estimating the heating energy demand of residential buildings in Switzerland using only public data

Abstract

Space heating and domestic hot water is responsible for 86 % of the residential heat consumption in Switzerland. Planning of future sustainable energy systems necessitates reliable estimations of these major demand contributors and their evolution in the future. In this contribution, we propose a procedure to estimate both the current and future annual and hourly space heating demand, relying solely on publicly available building properties. To assess the method’s accuracy, we validate and calibrate two simple white-box models using measured consumption data from the City of St. Gallen and the Canton of Basel-Landschaft. The validation results show that, for residential buildings in St. Gallen, the best-performing model achieves a mean absolute percentage error (MAPE) of 26.6 %, comparable to values reported for black-box models trained on measured data in the literature. However, performance decreases in the more heterogeneous Basel-Landschaft dataset, with a MAPE of 40.1 %. Calibration generally improves model accuracy, particularly when building characteristics are homogeneous, but has limited impact when high variability is present. Furthermore, the results confirm that simple models can produce robust aggregated heat demand estimates, even with minimal input data, while model suitability for individual building predictions remains more sensitive to dataset variability. The proposed method offers a simple, easily parameterizable approach for estimating space heating demand in cases where detailed measurement data is unavailable, making it suitable for both bottom-up regional energy system planning and building stock analyses.

Topics & Concepts