The main objective of the research was to develop a means of numerically predicting the turbulent flow field in a baffled agitated vessel. This has been achieved by way of the 'k-e' model of turbulence. An extensive review of the literature on turbulence modelling is given to support that choice. A range of experimental studies concerning the flow in an agitated vessel, which is driven by a turbine having flat vertical blades, is reviewed for the purposes of validating predictions, and setting boundary conditions. Qualitative agreement only is found between the predictions and experimental data for the three mean velocity components at places where such data is available. The vertical plane flow pattern is in qualitative agreement with observations. This will remain the case until more detailed experimental measurements are made. Swirl velocity predictions are much less certain than the vertical plane components, however. This is attributed to the assumption that gradients in the circumferential direction can be neglected. To account directly for swirl reducing baffles, it is shown that a specially constructed source term in the swirl momentum equation, may lead to a realistic simulation of their effect. Comparisons of turbulence predictions are difficult to make in view of a general lack of reliable data. The results of this work are encouraging however, in that impeller stream turbulence is shown to be highly non-homogeneous, while turbulence in the recirculation zones is homogeneous by comparison, except near walls. Furthermore, turbulence variable distributions seem to be qualitatively correct though magnitudes are much less certain. The integrated rate at which turbulence kinetic energy is dissipated into heat, is predicted to be equally divided between the impeller stream and recirculation zones. Finally, it is shown that there is much scope for improving the model through a closer examination of impeller boundary conditions, by adapting more appropriate models to simulate baffle effects, and from wider comparison with experimental data by varying the system geometry.
|Date of Award||1980|