Floating objects, particularly those of larger dimensions, often accumulate on hydraulic structures, reducing their flow capacity and increasing the risk of clogging and overtopping of bridges, culverts, and other installations. The retention of floating debris, combined with transported bedload at bridge openings and culvert inlets, can create unstable accumulations that raise upstream water levels and heighten the likelihood of flooding, overtopping, and local scouring of hydraulic structures. Understanding the quantity, characteristics, movement, and transport of floating debris is therefore essential for effective management and mitigation of debris-related hydraulic impacts. With the advancement of modern measurement technologies and methodologies, new approaches for detecting and analysing floating objects have been proposed. The choice of an appropriate combination of sensors and analytical methods depends on the specific problem under investigation. In this doctoral dissertation, a measurement system and methodology were developed for the detection, classification, and volumetric estimation of floating debris, applicable to both individual samples and clusters composed of multiple pieces. The proposed approach is based on non-contact measurement techniques. In laboratory experiments, measurements were carried out using a 2D laser scanner and digital industrial cameras. The volume of each sample was determined using the RANSAC method, along with a combination of above-water volume measurements and assigned specific material density. The point clouds obtained from LiDAR and photogrammetric measurements were compared and validated against reference values derived from Archimedes’ principle. In the fieldwork part of the doctoral dissertation, we aimed to simulate the conditions in the areas of debris boom installed at hydraulic structures, as well as floating object jams forming at individual bridge structures, where the accumulation and retention of floating objects occur. Field experiments were performed using an unmanned aerial vehicle equipped with either a camera or LiDAR system. The total volume of floating objects was determined with machine learning algorithms, which identified and quantified material types within the point cloud. Based on the assigned specific material density and the known above-water volume, the total floating objects volume was calculated.
|
