Adaptation of an overflow flume to study the sediment bedload threshold using image correlation

The bedload threshold of two fine sands S1 and S2 (D₅₀ = 328 and 210 µm) were investigated in two different flumes: a recirculating flume (CH1) and an overflow flume (CH2) (Figure 1). Table 1 summarizes the characteristics of both flumes. In CH1, the sandy layer has a thickness of δ = 0.02 m and is placed on the fixed bottom. In CH2, the sediments were deposited in a pit extending over a 1.0 m length with a thickness of δ = 0.05 m. The bed thickness and the slopes (1:3) are consistent with the study of Van Rijn et al, (2019). For both flumes, the sediment bed is located in a fully developed flow area, and the water depth was maintained at d = 0.15 m. The main objective of this study was to configure CH2 to enable to study the threshold of motion of sediment particles using the image correlation method developed at the LOMC laboratory by Vah et al. (2020). This non-intrusive technique provides a highly sensitive and objective measurement compared to traditional techniques.

The image correlation method identifies the bedload threshold (U_bl) by analyzing the decorrelation between a reference initial image of the sediment bed at rest and a sequence of images captured during a slow linear flow acceleration (1.3 mm.s^-2). As grains begin to move, the correlation drops linearly. Beyond the threshold, the slope of the decorrelation curve diminishes despite the increasing sediment motion. This observation is interpreted as a decorrelation saturation effect. As the bed surface undergoes important restructuring, the image loses its statistical similarity to the reference frame, reaching a correlation floor where further displacements no longer yield a linear decrease in the coefficient.

Generally, overflow flumes are tilted to analyze sediment motion. In this study, the flume slope was kept constant because the image correlation method requires a constant water depth. Consequently, an overflow flume (CH2) was adapted to allow the comparison of thresholds in two different setups. In CH1, the channel is filled until the target water depth is reached. Thus, no additional device is required to control the water height.

The filling process for CH2 differs significantly: it must be progressively filled over 15 minutes before each test to reach the desired water height (Figure 2a). This filling must be performed slowly to prevent premature sediment motion. Therefore, the spillway in CH2 was automated via LabView software to ensure slow filling and to maintain a constant water height. 

The main results are summarized below:

• For both S1 and S2, the U_bl values detected in CH2 are lower than those in CH1 (Figure 3). This finding is attributed to step 1 in CH1 which excites the sediment particles and triggers an earlier threshold of motion.
• An overflow flume can be effectively used to determine the threshold motion of sediment particles.