Abstract:
The morphological behaviour of bifurcation in rivers is a poorly understood problem
with which many river engineers are confronted. Bifurcation can be found in deltas,
in estuaries and in braided rivers. In one-dimensional mathematical simulation of a
river reach that contains bifurcation, some nodal point relation is needed to obtain its
sediment distribution in those branches. Determination of this nodal point relation is a
problematic task. While the sediment distribution at channel bifurcation is a threedimensional
phenomenon, a two-dimensional mathematical model for morphological
simulation with the inclusion of helical effect or secondary flow in the vicinity of
bifurcation omits the need of nodal point relation to obtain its sediment distribution in
the bifurcated channels.
The present study deal the two-dimensional mathematical modelling under the
morphological module of the sediment distribution at channel bifurcation with the use
of curvilinear grid along its main and bifurcated channel. For this study purpose,
software called MIKE21-C developed at Danish Hydraulic Institute, Denmark has
been used. In the model set up, an additional programme for the generation of grids
has been modified and adapted to fit the intrinsic curvatures occurring in the
bifurcation. A grid number of 75 in the x-direction and 25 in the y-direction give a
resolution of 0.16 m by 0.04 m in the specified directions, respectively.
The model has been calibrated against the available physical model data conducted by
Hannan (1995), Roosjen and Zwanenburg (1995) and Islam (1996). They conducted
test runs in a bifurcation model constructed in the laboratory of Water Resources
Engineering Department, Bangladesh University of Engineering and Technology.
They used both symmetrical and asymmetrical noses for three discharge conditions,
i.e., 20 lis, 30 lis and 40 lis. The present mathematical model reproduced the physical
model results for symmetrical nose satisfactorily. After the calibration, the model has
been applied for different sensitivity runs. The scenarios of the sensitivity analysis
include different Chezy's number (25 m1
/
2/s, 30 mI/2/s, 35 mI/2/s), different sediment
transport formulas as Engelund-Hansen and van Rijn, raised boundary water level in branch 2 by 5 mm and different grain size diameters (dso) such as 0.19 mm, 0.27 mm
and 0.35 mm.
From the two-dimensional mathematical model simulations, a set of data of discharge
ratio and sediment ratio have been calculated. This data have been set to the nodal
point relation 2= k(!iLr 'where k is the coefficient and m is the exponent. It has
been found Sz qz that the value of exponent m increase with the increase
of upstream discharges and is greater than 5/3 for all three upstream discharges which
explain well with the findings of Hannan (1995) and Roosjen and Zwanenburg
(1995). Sensitivity of the Chezy's number gives higher sediment transport with
increase in Chezy's number for Engelund-Hansen formula. Sensitivity of different
grain size has been tested using 0.19 mm, 0.27 mm and 0.35 mm diameter and it has
been found that the present setup is valid for grain size greater than 0.19 mm. When
testing sediment transport formula, with van Rijn sediment formula, a lower sediment
transport volume has been found compared to Engelund-Hansen. Influence of
boundary water on the sediment distribution at the bifurcation has been tested by
raising the boundary water level at branch 2 by 5 mm. Due to raised boundary
condition, flow through branch 2 is reduced and also the sediment transport and in the
branch 1 flow and sediment transport has been increase and thus increase the
discharge and sediment transport ratios compared to the same boundary level in the
two branches.
It is concluded from the study that sediment distribution at channel bifurcation is
independent of upstream discharge for one type of nose and two-dimensional
curvilinear modelling is a good tool for this type of study where several options and
scenarios could be tested easily.