Scan to BIM for autobahn bridge in Austria

Our client specializes in capturing infrastructure objects and developing detailed digital models using real textures extracted from drone-captured images. These models serve as essential documentation, enabling the identification of damages such as cracks, corrosion, or structural wear. These digital representations facilitate the creation of comprehensive building inspection reports, which in turn help develop the best reconstruction strategies and maintenance plans.

To capture the infrastructure in detail, the team employs photogrammetry technology using drone-captured imagery. This approach was selected due to the limited accessibility of certain areas with traditional scanning devices or digital cameras.

Scope of work: Precise modeling of the concrete girder bridge and the adjacent ground surface.

Input: Full point cloud of the bridge and adjacent area obtained through photogrammetry (drone capture).

Output: 3D model of the bridge in LOD300.

Bridges present unique challenges compared to traditional buildings due to their non-orthogonal shapes, curvatures in plan view, and elevation changes in side views. As a result, standard Revit tools are insufficient for modeling such complex surfaces.

To address this, our team developed an approach involving the creation of cross-sectional outlines at intervals along the bridge. Using Revit’s mass tool, they generated smooth, curved surfaces that pass through these cross-sections, enabling accurate representation of curved elements. This approach simplifies the later application of real textures.

While drone-based photogrammetry offers many advantages, such as ease of access to remote areas and cost-efficiency, the data often requires significant refinement. Post-processing typically reveals noisy point cloud data, where points may be scattered or surfaces poorly defined, adding complexity to the team’s workflow as they determine the most effective way to structure the model.

Another frequent issue is data duplication, where point clouds may overlap slightly or be misaligned. While the preferred solution is to request a new dataset from the provider, when this isn’t possible, the team manually evaluates and selects the best data set, though this may compromise precision in some cases.