PSI - Issue 62

Matteo Castellani et al. / Procedia Structural Integrity 62 (2024) 193–200 Matteo Castellani / Structural Integrity Procedia 00 (2019) 000 – 000

197

5

with reinforced concrete, form continuous solid walls with the deck and have a tapered rectangular cross-section narrowing from top to bottom.

a)

b)

Fig. 4. Case study bridge: (a) general view of the bridge; (b) detail of a pier.

4. Survey activities of the case study bridge 4.1. UAV Photogrammetry process

For this study, a professional quadrotor, DJI Mavic 3 Enterprise series drone, was employed to conduct the photogrammetric survey operations for the selected case study. This portable UAV has a built-in GPS that is used for way-point navigation and dataset geotagging. The Mavic 3 Enterprise, equipped with a 20MP camera and a 4/3 CMOS sensor, can capture images at extremely high resolution (5280px×3956px). Furthermore, the camera features adjustable shutter settings with an aperture range of f/2.8-f/11, providing excellent visibility even in dark environments. The vehicle and camera characteristics are summarized in Table 1.

Table 1. DJI Mavic 3 and its camera characteristics. UAV Specifics

DJI Mavic 3 Enterprise Series

Dimension Weight (Kg) Payload (Kg)

221×96.3×90.3 mm

0.915 0.680

Sensor Pixels

CMOS 4/3

20 Mpx

Image Resolution

5280px×3956px

The survey activities involved the central spans of the selected bridge useful to test the proposed workflow. Flights were carried out at different altitudes, both linear with respect to the axis of the bridge and rotational around the three piers, as indicated in Section 2.1. In total, 309 photos were taken and used for the model reconstruction process. The point cloud construction process was carried out using Metashape Pro software, following the several steps described in Section 3.1. The dense point cloud, representing the results of photogrammetric process, consists of a total of 78,710,651 points for the reinforced concrete structure. The total processing time is 10 minutes and 21 second for the alignment, and 48 minutes and 44 second for the dense point cloud generation. This aspect is significantly