WSP09211 - Laserfiche WebLink

Home Browse Search

364 Computational Fluid Dynamics vicinity of the reaches in this study. They made their calculation by determining the skin friction required to match measured transport rates dUling periods in which the sand supply was stable. This calculation yielded a skin friction that was 15% of the total shear stresS. This low value is consistent with the e.xtremely large channel roughness and associated form drag and was used in the calculations in this study in area, of the reach where the sand depth is less than the bed roughne>s. The procedure used in this study for determining the skin friction in portions of the channel where the sand is sufliciently thick to cover the bed roughness is different from the one used by Wiele el (II. (1996). Sand thickness tends to be greatest in recirculation zones. which are isolated from the tumultuous !low in the main chan- nel. and where the llow and sand transport more closely resemble that of alluvial streams. In this region. the local channel resistance and skin friction were calculated as functions of local llow. depth and sand size. This procedure used the methods described by Bennett (1995) to estimate bedform dimensions and form drag. Bennett drew on the work of van Rijn (1984) who used /I. to distinguish between ripples. dunes and upper plane hed and to estimate the dimensions of the bedfornls. if present. Given bedform height and wavelength. the local friction and skin friction are determined in Bennell's algorithm using the relations of Smith and McLean (1977) and I\elson and Smith (1989b). Relating local !low resistance and skin fliction to bedforms is an improvement over the use of values derived only from the local hydraulics, but errors may be induced by uncertainties in the relations used and in the assumption of equilibrium between the local !low and the bedforms. The model has been used to examine depositional processes and rates in the Colorado River main stem during a llood on the Little Colorado River (Wiele el (II.. 1996). to examine the dfect of sand supply on depositional patterns and magnitudes, and to compare the elfects of natural tributary flooding with flooding caused by increased dam releases (Wiele, 1997; Wiele etal.. 1999). The model has shown good agreement with cross-section measurements from before and after the LillIe Color- ado River !lood in 1993 (Wide elal.. 1996) and has replicated depositional patterns during Ihe 1996 test flow (Wiele elu/.. 1999). In one of the study reaches during the Little Colorado River flood, high sand concentrations led to massive deposition in the main channel (up to 12m) and formed a large bar along the left side of the main channel in the recirculation zone. In contrast. with lower concentrations and higher water discharge during the test !low, (he main channel scoured. and the focus of the deposition was near the recirculation zone reattachment point. The model replicated Ihese differences in depositional pattern with no calibration. The accuracy of the model without calibration in these studies supports the use of the model to predict results for the hypothetical cases in this study. . . 14.3 Study site selection and morphology Four modelling sites (Figure 14.1) were selected. which are within the study area of Hereford elal. 11991. 1993). Each of these reaches also contains gullies studied b.\ Thompson and Potochnik (2000). The modelling site farthest downstream. the Upper Lnkar reach. contains an especially sensitive archaeological site with abun- dant artefacts concentrated in a small area. -- .