Can reach-calibrated resistance factors sustain three-dimensional flow patterns in an acute river-bend?

MORELL MI; TASSI PA; VIONNET CA

Oxford, Wallingford

Workshop; XIXth Telemac-Mascaret User Club Workshop; 2012

HR Wallingford

P { margin-bottom: 0.08in; direction: ltr; color: rgb(0, 0, 0); }P.western { font-family: "Liberation Serif","MS PMincho",serif; font-size: 12pt; }P.cjk { font-family: "WenQuanYi Micro Hei","MS Mincho"; font-size: 12pt; }P.ctl { font-family: "Lohit Hindi"; font-size: 12pt; }A.western:link { }A.ctl:link { } The purpose of this study is to analyse if reach-averaged values of the hydraulic resistance coefficient, calibrated to match observed water levels, can sustain the simulation of three-dimensional (3D) helical motion on a pseudo-river bend, where the flow is prone to inertial effects by centripetal forces. This pseudo-river bend is located at the outlet of the low-gradient Colastiné River, Argentina, where the flow diverts in an almost T-shaped difluence. Two drogues equipped with Blue-tooth technology were used to identify depth averaged particle paths at the sharp turn of the river difluence, showing a clear separation between stagnant water and curve flow, which indeed behaves as a river-bend. Due to the irregular bed topography and the changing river planform on the pseudo-river bend the flow is particularly susceptible to convection effects such as acceleration, deflection, stagnation, and flow separation, making the zone extremely computational demanding for capturing flow gradients. Accordingly, field data was collected on the pseudo-river bend and further upstream along 10Km using an acoustic Doppler current profiler (aDcp) coupled to a satellite position system. Then, a computational study using Telemac 2D & 3D is performed first to demonstrate how reach-averaged hydraulic resistance factor varies as more physical details are brought into play. Secondly, the resistance factor calibrated to match the observed hydraulic gradient along the reach is tested against 3D field data of cross-circulation captured with the aDcp. In more detail, flow in meandering open channels is characterized by the generation of the pressure- driven secondary flow on the channel bends that triggers scour along the outer bank and sediment accumulation along the inner bank. Results from simulations on meandering streams using different turbulence models have shown little difference. This result could indicate that turbulence modelling is restricted to provide the correct amount of mixing, which in turn depends on the behaviour of flow gradients. Furthermore, well-known solutions of secondary currents on river-bends show an explicit dependence on hydraulic resistance and turbulence mixing coefficient values. But since the turbulence mixing coefficient can be assumed to be proportional to the hydraulic resistance factor, a better understanding of the role of boundary generated flow resistance is required for further understanding the mechanisms responsible for shaping the bed topography in river bends. Unfortunately, loose-boundary hydraulic suffers from a lack of generally accepted relationships between bed-form morphology and flow resistance, which complicates loose-bed modelling. This is because there is a strong interrelationship between resistance to flow, bed configuration, and rate of sediment transport. Resistance to flow parameters are normally written either in terms of the friction factor CF or the Manning coefficient n, whereas the Chézy coefficient CZ is a discharge coefficient that varies inversely with resistance factors. It is common practice in hydraulic simulation models to use the resistance factor as the calibration parameter to match observed water levels. Nevertheless, and in spite that several elements in hydraulic modelling were improved over the past decades, the resistance coefficient still has large uncertainties because it depends primarily on the bed-form configuration which may change from plane bed to ripple and dunes. Consequently, the bed-form configuration of the river reach is treated here as a sub-grid scale problem embedded in the overall value of the reach-average resistance factor, where the turbulence mixing coefficient is adjusted to fine-tuning, whenever possible, vertical profiles of the stream-wise and cross-wise velocity components. Preliminary computations will be discussed during the meeting.