A knowledge graph-based framework to automate the generation of building energy models using geometric relation checking and HVAC topology establishment

Abstract

Building Energy Models (BEM) are widely utilized throughout all stages of a building's lifecycle to understand and enhance energy usage. However, creating these models demands significant effort, particularly for larger buildings or those with complex HVAC systems. While a substantial amount of information can be extracted from Building Information Models (BIM) — which are increasingly accessible and provide necessary data for geometric and HVAC contexts — this information is not readily usable in setting up BEM and typically requires manual translation. To address this challenge, this paper introduces a BIM-to-BEM (BIM2BEM) framework that focuses on automating the generation of HVAC parts of BEM models from BIM data. Core to the methodology is the extraction of HVAC system topologies from the BIM model and the creation of a knowledge graph with the HVAC topology. The topology transformation unfolds in three key stages: first, a geometry-induced knowledge graph is established by examining the geometric relationships among HVAC elements; second, this graph is converted into an informative HVAC topology with enhanced properties from additional data sources; and finally, the informative topology is simplified into a BEM-oriented HVAC topology compliant with BEM platforms such as EnergyPlus. A case study of a large university building with a complex HVAC system showcases that the proposed framework achieves automatic and precise generation of building performance simulation models. The model's predictions are then validated against actual measurements from the building.