Prediction of solute distributions in an asymmetric compound channel is carried out using two turbulence models. Results of the numerical models are compared with turbulence data recently obtained using laser Doppler anemometer (LDA) and laser induced fluorescence (LIF) in a small laboratory flume. The predicted distributions of solute and Reynolds flux using a k-ε model and an algebraic stress model for various injection points near the water surface are used to identify different mixing mechanisms. A skewed distribution of solute on the floodplain observed in the experimental data is well predicted by the algebraic stress model but not by the k-ε model. The cause of the skewed distribution is examined through the variations of secondary flow and eddy diffusivity. The predicted eddy viscosity and diffusivity and the turbulent Schmidt number are discussed with the experimental data. As a result, solute concentration distribution can be well predicted by adjusting the turbulent Schmidt number even if the eddy viscosity was not correctly calculated by any particular model. An effect of secondary flow on peak concentration in the shear layer along the channel is also demonstrated.
SHIONO, K. - Department of Civil and Building Engineering, Loughborough University, UK
SCOTT, C.F. - Department of Civil and Building Engineering, Loughborough University, UK
KEARNEY, D. - Binnie Black & Veatch Ltd, 38 City Road, Chester, CH1 3AE, UK
Keywords: Numerical model; solute transport; compound channel; mixing coefficients; Reynolds fluxes.
This paper is an analysis of two different turbulence models, one based upon an algebraic stress model the other being a kinetic energy model. The aim was to determine which method was more accurate when compared to experimental data. Various injection points were used in the experimental procedure for corroboration of data before graphically plotting various aspects of the experiement against the two models.
The numerical predictions of transverse distribution of concentration are all lower and narrower than the experimental data which is said to indicate under-prediction of mixing. Overall, it was found that predictions with the algebraic model of local velocities were in good agreement with the experimental data, but not so good with the kinetic energy model, because it was not able to predict secondary flow. Both models over-predict vertical mixing compared with experimental data. Both models did not predict the distributions well for shallow flooded water conditions, however the algebraic model predicted a skewed distribution which was found to be the case for deep flooded water. In both cases the peak concentration was more than 20% different in the models with and without a secondary flow modelling capability.
It is clear from this paper that generally, the algebraic model forms a better basis to predict solute transport in a compound channel, however, values of peak concentration calculated cannot be solely relied on. This paper presents a good case for turbulence models as well as addressing many of their problems in terms of prediction capabilities. It also requires some previous knowledge of flow prediction to decide which model is best to use. The paper would be of good use to engineers using models (with caution) for justification and looking for models to get a good idea of what will happen in experiments rather than solely using a model as basis of an experiment.