Two-Equation Turbulence Modelling of Open Channels

The transport of sediment and pollutants in open channels essentially depend on the turbulence phenomenon among other factors. Being the most complex phenomenon to fully comprehend and estimate, turbulence computations have been the most challenging part of engineering calculations. Several analytical and empirical models have been in vogue, prior to the widespread availability of powerful computing facilities, to estimate turbulent boundary layer properties. However, this practice changed approximately four decades ago when a few turbulence models started to emerge and initially gained popularity among the researchers and then among the practicing engineers. Almost all the present-day commercial models, used for engineering calculations for rivers and estuaries include some of these turbulence models. Nonetheless, many engineering calculations are based on empirical models or a combination of turbulence and empirical models yet. As usual, turbulence models used in research are far more complex and computationally expensive than the ones used in the field applications. Several studies on open channels based on Direct Numerical Simulation (DNS), Large Eddy Simulation (LES) and Reynolds Stress models have been carried out in the past. However, two-equation turbulence models have gained popularity among researchers as well as practicing engineers because of their reasonable accuracy with computational economy. Many versions of such models are reported in the literature among them k-epsilon and k-omega have been the most popular two-equation models. Sana et al. (2009) and Sana and Tanaka (2000, 2010) have compared the performance of some of the popular versions of two-equation turbulence models in case of oscillatory boundary layers. In this paper, a few model versions are reviewed based on their predictive abilities and computational economy against the well-known bottom boundary layer properties in open channels. Qualitative and quantitative comparisons have been made to infer that the choice of model versions should be based on the field application. For example, the bottom shear stress is very well predicted by the k-omega model whereas the cross-stream velocity profile and turbulent kinetic energy are predicted more efficiently by k-epsilon model versions. Consequently k-omega model will be more appropriate for estimating bottom sediment transport whereas k-epsilon model is expected to yield better results in case of suspended sediment or pollutant mixing and transport in the field. This study may be useful for the researchers and practicing engineers in selecting a suitable two-equation model for calculating various bottom boundary layer properties.