Motion detection of sand size mixtures 12.26

Coastal nourishments are performed to manage coastal systems against threats like sea level rise, extreme weather conditions, and increasing population demands. Sand used in shoreface nourishments is neither uniform in size nor has the same size distribution as its surrounding environment (see Huisman et al. (2016) and Guillén & Hoekstra (1997)). Sand transport models cannot accurately estimate the critical bed shear stress for various particle sizes, especially when they interact with finer or coarser grains. According toVan Rijn (2007c), sand particles behave differently when they are in a size mixture. Existing studies have focused on incipient motion processes for the grain size range of sand-gravel mixtures, and not for sand-sand mixtures, as a result, existing formulations cannot reliably predict the critical shear stress of the coarse and the fine sand fractions.  

As part of the SOURCE project, Initiation of Motion (IoM) experiments were conducted in the laboratory flume of WaterProof in Lelystad (from mid-November 2025 to December 2025). Two types of bimodal sand mixtures were used: one set consisting of a finer fraction of 0.1 mm and a coarser fraction of 0.4 mm, and the other a finer fraction of 0.1 mm with a colored coarser fraction of 0.6 mm. Different size compositions of this bimodal sand were placed in a test bed in the middle of the flume (see Figure 1). Flow velocity increased slowly, and respectively, more sand grains moved by increasing velocity, until all sand grains started to move, and general transport stage exceeded (see Figure 1.2 ofthe report of Van Rijn (2020)). Corresponding flow velocities were recorded every moment using an ADV in the upstream part of the test bed. The process of this increasing number of moving sand grains (with different sizes) was recorded by an 8K camera from top view (see Figure 1). Three spotlights were used to improve the lighting conditions (see Figure 2). To reduce the effects of vibrations of water surface and the flume walls on the video quality, a clean plexiglass (methyl methacrylate) plate was placed and fixed on the water surface. As shown in Figure 2, stands were used to ensure this plate and the camera position were fixed in place.

                                 Figure 1 Schematic representation of the flume at WaterProof

                           Figure 2 A Photo of experimental set-up at the WaterProof flume

To derive the incipient motion of the 2 size fractions in the mixtures, the recorded videos will be analyzed. The two fractions should be distinguished (by e.g., the pixel color or pixel clusters), and it should be quantified how many grains (of each fraction), and what percentage of each fraction are moving at a specific time in the movie. This time corresponds to a recorded velocity by the ADV. In this way, we will derive what percentage of each size fraction is moving at a specific flow velocity. Finally, the results from the various size compositions will be compared to quantify how much being in contact with another sand size affects the initiation of motion. Based on this, empirical formulations on sand mixture incipient motion will be improved.

The objective of this research is to observe, understand and quantify sand grain size effects on the initiation of motion of mixed sands. To achieve this goal, there are three steps:

1.       Conducting experiments in WaterProof laboratory in Lelystad (already done).
2.       Processing and analyzing experimental data (mostly videos) collected from the flume
3.       Comparing the experimental results with the results from empirical formulations

    We are seeking a student who is excited about performing advanced image/video processing techniques, and drawing insights from the results. Using machine learning or AI techniques in the image processing can be an addition (it is not mandatory) to the more traditional coding techniques. This depends on the interests of the student and his/her creativity. The student will gain practical experience in motion detection techniques of objects in a video under the direct supervision of a PhD candidate. Coding skills are required and essential, but background in hydraulics, and sediment transport is a plus (not necessary).