Numerical Study Of Time Scale Distortion In Sediment Flushing Model Experiments

Movable-bed physical models are widely used to understand fluvial processes, estimate the amount of reservoir sedimentation, and study sediment transport in estuary and coast areas. Time scale distortion is one of the chief factors affecting the accuracy and reliability of the test results of movable-bed physical models, particularly in the case of unsteady flows. For example it causes propagation of flood wave in the model lag behind the prototype. It can also result in dissimilar erosion or aggradations in the case of abrupt discharge changes.This paper presents a case study on problems caused by time scale distortion in the physical model experiments for sediment flushing in the navigation channels upstream of the Three Gorges Dam. A 2-D shallow water and sediment transport numerical model was adopted to simulate the sediment flushing process in both the prototype and the physical model. Because the big-sized physical model could not be easily operated, another small-sized conceptual physical model was used in the research. The numerical model was also adopted to simulate the experiments conducted in the conceptual physical model. Flow fields, water surface curves and sediment erosion volumes calculated by the numerical models were found to be similar with the corresponding data measured in the physical model experiments, showing that the numerical model can relatively accurately simulate the sediment flushing process in the prototype, the big-sized physical model and the conceptual physical model. The numerical simulation result showed that 4000m3/s is the optimal flushing discharge. Test results of conceptual model experiments with different geometrical distortion rate were also compared with the numerical model.Based on the test results from physical model and numerical simulations, two problems caused by time scale distortion in the sediment flushing experiments were distinguished: the dissimilarity of the proportion of steady flow duration between the physical model and the prototype, and the insufficiency of time length discharging high concentration flow through the flushing sluice and tunnels in the physical model. A new method to divide the duration of unsteady flow and steady flow in the physical model was proposed, using which the physical model experiment could better represent the situation in the prototype. The total duration of sediment flushing was extended to compensate for the loss of time length discharging high concentration flow in the physical model.The accuracy of conceptual model are not only affected by time scale distortion but also by dissimilarity of roughness. Some methods to improve the accuracy of physical model experiments were proposed, and validated with test data and numerical simulations. The effectiveness of extending the duration of sediment flushing for increase of the erosion volume in the big-sized and conceptual physical model were measured and calculated. It was suggested that the total flushing duration be extended to 422s to best represent the erosion volume in the prototype. The possibility of using model sands of greater density in the physical model was discussed, and the sediment erosion volume was calculated if these sands were used. Loosening the model sand can increase the erosion volume in the navigation channel which was reduced by the compaction of model sand after a long time of deposition. Its effectiveness was measured in the physical model and calculated in the numerical model. The conceptual physical models of different geometrical distortion rates was proposed to increase its roughness to different degree to better resemble the water surface curve measured in the big-sized physical model.