Automatic Detection of Geometric Errors in Space Boundaries of IFC-BIM Models Using Monte Carlo Ray Tracing Approach

Abstract

In Industry Foundation Classes (IFC) building information modeling (BIM), the objectified concept of a space boundary (SB) provides a means to define building space geometries with surface entities. Such building-geometry definitions are widely used for various engineering applications such as energy simulation, lighting analysis, and facility management. However, quality issues (i.e., geometric and nongeometric issues) of SBs have been widely reported, which makes it necessary to validate the SBs before retrieving them from IFC models for relevant applications. Unfortunately, there is still a lack of reliable mechanisms/tools to automatically evaluate the quality of SBs, especially the geometric quality. This study proposes a Monte Carlo ray tracing approach to automatically detect geometric errors in SBs. The approach checks SBs space by space in terms of whether each space is correctly bounded by its SBs. The geometric errors in the set of SBs of a space that the approach can detect include gaps, overhangs, and overlaps between SBs as well as incorrect surface normal directions of SBs. To accelerate the ray tracing process in the approach, the axis-aligned bounding box (AABB) tree is implemented to spatially index SBs of each space. The approach is evaluated with extensive performance tests in terms of robustness and efficiency. The results show that the approach can robustly and efficiently detect all four types of geometric errors even in extreme cases and that the AABB tree helps speed up the approach significantly for large-scale IFC models with many complex spaces.