Abstract
Urban Building Energy Modelling (UBEM) has become an essential tool for analysis and planning renovation strategies as well as predicting future energy use patterns both at building and district levels. To aid urban planning effectively, digital solutions such as UBEM should support informed urban planning decisions by analysing various renovation scenarios, including energy use, occupant comfort, and climate resilience, while ensuring accuracy without excessive complexity. A critical input for physics-based UBEM tools is building geometry, which can be represented at different levels of detail (LoD). Striking a balance between accuracy and the costs and time required for data generation necessitates examining the influence of LoD in building geometry. This research aims to investigate the extent to which the LoD in building geometry can influence the accuracy of results. To achieve this, four LoD of building geometry were defined for 9 different detached single-family houses in Sweden, and their effect on model performance was assessed with and without calibration using the Monte Carlo Markov Chain (MCMC) algorithm. The calibration used simulated heating demand and utility bill data, optimising for Coefficient of Variation of the Root Mean Square Error (CVRMSE) and Normalised Mean Bias Error (NMBE) indices. The results indicate that, without calibration, models with differing geometric LoDs can exhibit significant performance discrepancies, with variations of up to 19.7% in the CVRMSE and 22.5% in NMBE observed between high and low geometric detail levels. However with calibration, the differences between models with varying geometric LoDs were substantially reduced, with average CVRMSE and NMBE decreasing to 2.9% and 0.9%, respectively, well within ASHRAE Guideline thresholds. The findings offer useful guidance for academic and societal stakeholders working to improve energy modeling in urban planning. By identifying effective LoDs and calibration methods, this research enables cost-efficient UBEM applications, supporting better decisions for sustainable urban development.