PSI - Issue 64

Cedric Eisermann et al. / Procedia Structural Integrity 64 (2024) 1224–1231 Eisermann et al./ Structural Integrity Procedia 00 (2019) 000–000

1229

6

The as-is geometry of the bridge was captured using terrestrial laser scanning (Leica RTC 360, Leica P40, and Z+F Imager 5010X) for the pilot area. Prior to scanning, a control network was established for the entire structure with a standard deviation of less than 5mm, ensuring a high accuracy in current and future measurements. All scans were performed while stationary and georeferenced using at least four control points from the control network. This allowed a redundant and thus precise linkage of all scans. The point cloud obtained from the laser scanning has a point density of 2cm across the entire bridge. To convert the point cloud into a 3D model, a manual approach was used. First, a total of 278 cross-sections were derived from the point cloud, each extracted every 50 cm along the bridge axis and at stations where the bridge cross-section undergoes discontinuous changes (e.g., jumps). The points of each section were then traced in a CAD program and extruded to form a solid volume. The Gerber joint in the middle of the first main span was manually modeled based on existing technical drawings. The caps and surfacing were subtracted from the 3D model using their planned dimensions. No additional measures, such as test drilling, were taken to determine their exact geometry. 3.3. Geometric model comparison To calculate the coefficient of variation of the geometry V geo , a dataset of target-actual deviations is required. In this case, the cross-sections at the ends of the 3 m long cantilever sections were initially selected for comparison. This selection was made based on the assumption that the geometry within a construction section is correlated, e.g., due to the same formwork. However, this assumption needs validation in further studies. As a result, a total of 33 cross sections were evaluated by calculating the absolute deviations and relative deviations between the geometric values of the as-designed model X d and the equivalent values from the as-is model X i . The results of the comparison of the total cross-sectional area are demonstrated in Fig. 4. The consistently positive deviations indicate that the constructed cross-sectional area is larger than the originally planned area. The range of absolute deviations is from 0.073 m² to 0.358 m², which corresponds to relative deviations between 0.57% and 4.77%. The absolute and relative deviations slightly increase towards the Gerber joint. On average, the absolute and relative deviations are 0.201 m² and 2.06%. The trend of the constructed geometry being larger than the planned geometry is also reflected in the comparison of the total volume, with an absolute deviation between the actual (615.28m³) and the planned volume (601.63 m³) of 13.65 m³ (2.27%).

Fig. 4. Comparison of cross-section areas

Table 1 lists the deviations in dimensions of the individual cross-section components. The thickness of the webs and top slab exhibits only minor deviations at the measured points. The bottom slab thickness is on average 7.10% larger than the planned geometry, which is also reflected in the deviations of the corresponding area. The deviations