Design misdirection: how BIM modeling granularity compromises energy performance assessment in integrated BIM-to-BEM workflows

Building Information Modeling (BIM) to Building Energy Modeling (BEM) integration offers a promising pathway to enhance energy performance assessment, yet the variability in BIM modeling granularity remains a significant source of uncertainty, contributing to the performance gap. This study systematically investigates how different facets of modeling granularity impact BEM outcomes. Using a university building archetype, a parametric matrix of 62 scenarios was generated by varying four domains of modeling granularity: spatial configuration, construction assemblies, thermal zoning, and internal conditions. An automated BIM-BEM workflow was employed to process and simulate each scenario. The analysis reveals a significant performance spread across the scenarios, with the predicted Energy Use Intensity (EUI) varying by up to 68 %. The results identify internal conditions, a form of semantic granularity, as the dominant factor driving this variability, organizing outcomes into distinct, non-overlapping performance clusters that far outweigh the influence of spatial or construction detail. Furthermore, the study demonstrates that simplified semantic models can significantly distort the building’s energy balance, in some cases overestimating total cooling energy use by 18%. This distortion, in turn, can lead to design misdirection by masking localized load peaks and altering the identification of critical spaces for system sizing. Based on these findings, the study articulates a conceptual model that distinguishes between spatial and semantic granularity. This model serves as a heuristic guide to facilitate the shift from a paradigm of “as detailed as possible” to a more resilient strategy of “as detailed as useful”, thereby promoting a more conscious and purpose-driven approach to modeling granularity.