Author(s): Andre Paquier; Claire Beraud; Jerome Le Coz; Benoit Camenen
Linked Author(s): Jérôme Le Coz, André Paquier, Benoît Camenen
Keywords: Bed load; Armoured bed; Field experiment; 1-D numerical model; 2-D numerical model
Abstract: To restore a more active sediment transport in the Rhine River, the introduction of coarse sediment from the floodplain to the river main channel is planned. A real-scale test was performed in 2010 by building a 22, 000 m3 gravel bar. The monitoring of this gravel bar allowed to better understand the main processes involved during floods: the sediments are first eroded from the gravel bar to the main channel and then transported along the main channel. In order to be able to extrapolate experimental results to other injections of sediments, the numerical modelling of the first flood event was performed. The grain size distributions are described by the median diameter and the standard deviation only (21 mm and 3. 7 respectively for the gravel bar). For the simulated flood, the armoured bed that constitutes most part of the main channel bottom did not move similarly to what was observed in the field. To produce realistic results, the critical bed shear stress was corrected using the Ikeda formula accounting for slope effects. If the sediment is deposited into horizontal layers and the shear stress is calculated from the average flow velocity using the Merged Perpendicular Method, the 1-D model produces results that are in close agreement with the measurements: about 70% of the sediments constituting the bar at the beginning of the experiment are eroded and 90% of them are immediately deposited in the main channel in the vicinity of the bar or within 200 metres downstream. The 2-D model computes bed shear stress locally, which, - compared to 1-D model - generally provides higher values in the main channel and lower values next to the gravel bar. Thus, the 2-D model produces lower erosion (about 30% of the initial volume) and the eroded sediments are transported over a longer distance in the main channel. Finally, although the accuracy of the results depends on the values of the parameters, the main processes are acceptably taken into account by both models.