Consistency in early simulation inputs with statistical analyses of historical models: Reducing the impact of mental models and simulation settings

Simulation inputs, especially in early design phases, are associated with great uncertainty and, on top of that, dependent on the modeller. Minor setting variations can significantly impact the simulation output, making it difficult to evaluate the design quality and compare different Building Performance Simulations (BPS). In this study, we have made a statistical data analysis of design inputs and simulation settings for 1320 selected BPS files created within a single company in the period 2009-2020. This extensive historical dataset provides insight into modellers' behaviour, common practice and typical building design variations. The former can help streamline future BPS and make them less modeller dependent, whereas the latter provides realistic starting points in the early design phases.