Development of a methodology of validation for urban building energy models and application to French residential consumption

International audience; Cities need to drastically reduce their energy consumption in order to fight climate change. To help them implement efficient energy saving measures, urban building energy models (UBEM) have been developed. A large amount of input data is required to model buildings at an urban scale and that information is often missing. To overcome this lack of model input data, data enrichment approaches including archetypes and calibration processes have been used, resulting in errors in the generated data which are not always well quantified. The aim of this paper is to develop a methodology of validation for UBEM that takes into account a data enrichment process. The proposed approach is to split the validation process into several parts based on UBEM use cases. UBEM outputs and enriched data to be validated are selected for each use case and the fitness for purpose of the model is then verified against measured data. This validation method has been tested on a specific use case: the prediction of yearly electricity and gas consumptions of dwellings aggregated at the district scale. It has been applied to two different UBEM tools: Smart-E and Dimosim. Smart-E uses its own hypothesis for data completion while a specific tool, Qiriel, is used to infer missing data for Dimosim. The simulation of one hundred districts (90 000 dwellings) located in the French department of Seine-et-Marne and the comparison to measured data enabled to study the error distribution of both UBEM tools. Further work should be led at smaller spatial and time scales.