How BIM and 3D Point Cloud Data Work Together
BIM (Building Information Modelling) works on the assumption that the digital model accurately represents the physical facility. In new construction, this is generally true. In existing buildings, especially industrial facilities that have been modified over years of production changes, the model and reality often diverge significantly.
3D point cloud scanning bridges that gap. It captures the facility as it actually is, providing a spatial reference that the BIM model can be compared against, updated from, or built upon.
The Scanning Side
For BIM-related work, we typically use the Leica RTC360 terrestrial scanner. It captures the entire visible environment from each scan position as a dense point cloud. Multiple scan positions are registered together to create a single, unified dataset covering the survey area.
Scan density and coverage depend on the purpose. For general BIM comparison, scan positions at 10 to 15 metre intervals usually provide sufficient coverage of open areas. For detailed as-built modelling, closer spacing and additional scans of complex areas (plant rooms, service risers, overhead structures) are necessary.
The registered point cloud is exported in formats compatible with BIM software. For Autodesk users, ReCap (.rcp/.rcs) is the standard exchange format. For Revit workflows, the point cloud can be linked directly into the project environment.
BIM Comparison: Scan vs Model
The most straightforward application is comparing the point cloud against an existing BIM model to identify deviations. This is useful in several scenarios:
Construction verification. Checking that what has been built matches the design model. Structural steelwork, service routes, and floor levels can all be verified against the BIM model with the point cloud as reference.
Change documentation. In facilities where modifications have been made without updating the BIM model, the point cloud highlights where the model is out of date. This is common in manufacturing environments where changes are made by the operational team without updating design records.
Clash detection. When new equipment or services are being planned in an existing facility, the point cloud provides the actual spatial constraints. Running a clash detection between the proposed design and the point cloud identifies conflicts before installation begins.
Software tools for this comparison include Autodesk Navisworks, CloudCompare (open source), and Leica Cyclone 3DR. Each offers different levels of automation and reporting capability.
From Point Cloud to BIM Model
When no existing BIM model exists, one can be created from the point cloud. This is referred to as scan-to-BIM or as-built modelling.
The process involves the modeller working in Revit (or equivalent) with the point cloud loaded as a reference. Structural elements, walls, floors, services, and equipment are modelled by tracing the point cloud geometry. The result is a BIM model that represents the facility as it currently exists.
The level of detail (LOD) varies by requirement. LOD 200 provides basic geometry and approximate dimensions. LOD 300 adds accurate dimensions and spatial relationships. LOD 400 and above include fabrication-level detail. Higher LOD means more modelling time and cost, so the level should be specified based on the intended use.
Practical Considerations
Point cloud data files are large. A typical industrial facility survey generates 10 to 50 GB of raw data. BIM software handles this through indexing and level-of-detail rendering, but workstation hardware specifications matter. Underpowered machines will struggle.
Coordinate systems must be agreed at the outset. The scan data and the BIM model need to use the same coordinate origin and orientation. Misalignment here creates false deviations and wastes time.
The value of combining BIM and point cloud data is simple: you get a model you can trust, based on measured reality rather than assumptions. For any facility modification, upgrade, or new installation in an existing building, that trust has direct value in reduced rework and fewer surprises on site.