Automated geometry-based analysis for object-level Scan-vs-BIM validation of TBE components

Technical building equipment (TBE) systems offer significant potential for energy savings that can be realized through digital twins and simulations. Digital twins require a precise virtual representation of the real object. Geometric-semantic models originating from Building Information Modeling (BIM)-based planning can be a valuable basis for this. However, due to unavoidable deviations between planning and construction, a high-quality comparison of as-planned and as-built data is crucial. This study presents a geometric and statistical analysis framework for component-specific validation of TBE systems using as-planned BIM models and as-built point clouds. The framework was evaluated on datasets of increasing complexity, from simulations to real-world projects. Results show that with sufficient data quality, the framework enables the validation of up to 88% of components, significantly reducing manual effort, cost, and time. However, as a geometry-based approach, its performance is affected by data quality issues such as point cloud noise and occlusions.