Comparison of Flow Characteristics in a Physical Model with Three Numerical Models in a River Reach, under two Different Training Schemes

Application of numerical models in river engineering projects is unavoidable. However, the reliability
of model, minimum requirement for field data, and fewer computational processes are of major
concern. The main purpose of the present study was to test the reliability of some well-known
numerical models in the simulation of flow characteristics in different river planforms. Comparison
were made between the flow characteristics in a river physical model and the simulation results from
three numerical models, under two different river training schemes, with identical initial and boundary
conditions.
In this study, a reach of the Nazloo River in Urmia, Iran, was selected. A fixed-bed physical model of
the River Reach (1200 m long, with horizontal scale of 1:100 and vertical scale of 1:20, and with the
Nazloo cross bridge included within the reach) was constructed, calibrated and verified. Two training
schemes were planned, using longitudinal and transversal structures, i.e, levees and groynes. Flow
parameters (such as,: depth and point velocity) were measured for four different flows. Three river
models: 1D model HEC-RAS, Quasi-2D model BRI-STARS, and 2D model FAST-2D were selected.
Six flow parameters (i.e. river flow capacity, water surface elevation, mean flow depth, mean velocity,
mean bed shear stress and Froude number) were compared between the physical model and the three
numerical models. Four different flow conditions were examined, and the results were compared in
three sub-reaches along the river reach (from downstream to upstream of the bridge).
The averaged predictive errors from these three models were determined for corresponding flow
parameters. The HEC-RAS, FAST-2D, and BRI-STARS models are considered to be the best fitted
models with the true physical model, respectively. Simulation results from the HEC-RAS model are
well adapted to the river flows confined between the two-sided levees, where the geometry of the river
reach is more uniformly defined, and in minor flood flows. The prediction from the FAST-2D model
is superior along the river reach with groynes involved, particularly in higher flow levels. The
application of each of the three models is recommended in river projects subject to the inclusion of the
order of certainties provided by the present study.