Building Energy Simulation (BES) offers valuable virtual assessment of decarbonisation strategies but faces implementation challenges due to the complex balance between accuracy, modelling detail, and data availability. There is a significant lack of understanding in how these factors interact across different building characteristics and how they impact the reliability of energy conservation measure (ECM) modelling. This study proposes a novel integrated framework that correlates input data quality, modelling complexity, and ECM reliability across diverse building types and sizes, and introduces a quantitative scoring system to objectively balance modelling effort against prediction accuracy. The methodology employs three levels of modelling detail (LOD), progressively simplifying input parameters while verifying calibration accuracy through one-at-a-time modifications. A systematic evaluation of fifteen ECMs is then conducted on each LOD model. The framework is validated using three University of Oxford case study buildings of different sizes, types, and ages. Results reveal larger buildings permit greater simplification while maintaining accuracy with goodness of fit values averaging 7.1 % for simplified models in the largest case study building compared to 8.6 % in the smallest. Process-driven buildings show greater resilience to schedule-related simplifications with sensitivity values averaging 0.08 compared to 0.17 for occupancy-driven buildings. Deep retrofit and renewable measures maintain high reliability with simplified models, scoring highest on the developed scoring system, while control-based ECMs require higher modelling fidelity. This comprehensive framework provides a novel methodology for optimising BES implementation based on building characteristics and analysis objectives, facilitating broader adoption of BES tools for net-zero strategy development and policy support.
