Bedload sediment transport interdependency in unsteady flows

Abstract

Our understanding of bedload transport and the reciprocal interaction with its surrounding physical environment, despite recent studies showing increased interest in the problem, remains weak. The main aim of the current study was to improve understanding of bedload transport under unsteady flow conditions and then investigate how this transport responds - and feeds-back to - near-bed turbulence and a changing bed surface roughness (transport-turbulence-roughness, TTR, interdependency). Experiments were carried out in an 8m long, 0.3m wide by 0.3m deep tilting (Armfield) flume. Two experimental bed mixtures were used, one over a fine graded bed mixture (d50 = 2.93mm) and a second over a coarse graded bed mixture (d50 = 4.47mm). Data were gathered across various parametrically defined unsteady flow hydrographs using a sediment trap; ADV; and DSLR camera which facilitated measurements of bedload transport, near-bed instantaneous point velocities and bed surface roughness, respectively. The bedload transport data identified a hysteresis effect across the unsteady flow hydrographs tested. A clockwise hysteresis was found across the first hydrograph in sequence tested, and a reversal to a counter-clockwise hysteresis was seen across the second hydrograph in sequence. The quadrant analysis technique was used to identify the proportion of specific types of turbulent event occurring near the bed surface. From this, an evolution of turbulent ejection events was found, which from an initial high reduced moving towards hydrograph peak. An opposite and commensurate effect within the sweep events was seen at the same time. A reversal of this trend, in both turbulent ejection and sweep event data, occurred on the hydrograph falling limb. Analysis of the bed surface roughness data found coarsening on the hydrograph rising limb and fining on the falling limb – with the fining found never to return the bed surface roughness to its original state, thus leaving it coarser than before the hydrograph passage. A further set of experimental runs were carried out, replacing the DSLR camera with a particle image velocimetry (PIV) system capable of capturing instantaneous vector flow fields (streamwise and lateral directions) and images of grain motion concurrently. Grain-scale analysis of this data found that the streamwise fluctuation from the mean (u’) velocity at the time of grain motion suggested two mechanisms by which turbulence induces bedload transport (a positive mechanism and a negative mechanism). These mechanisms were found to likely be associated with turbulent ejection and sweep events, respectively. Investigation of turbulence-induced momentary peak forces (impulse) found that the positive mechanism (i.e. turbulent sweep events) was generally more capable of transferring greater momentum to grains by inducing larger magnitude impulse events on grains. Calculation of a representative impulse using ADV data provided a continuous dataset describing the prevailing fluid impulse on a typical grain across the whole hydrograph. This captured the evolution of turbulence across the hydrograph. Separately, the PIV data permitted the calculation of the range of impulse resulting in grain motions that were recorded in motion. Comparison of the representative impulse (ADV) and known impulse required for motion (PIV) showed an overlap in their respective histogram distributions. This finding was used to propose an approach using joint-probability that may be expanded upon to help reduce the uncertainties inherent in bedload transport prediction.