Local and time-dependent phase distribution phenomena in bubble columns

In this project experiments were performed and models were developed to describe the hydrodynamics and phase distribution in tall air-water bubble columns without liquid throughput. The experiments were performed in a two meters high, 23 cm internal diameter column with a porous plate air distributor. The superficial gas velocity varied between 1 and 9 cm/s, covering both homogeneous bubble flow and turbulent bubble flow. Three techniques of measuring the time averaged gas fraction are compared: using a glass fibre probe, pressure transducers, and the bed expansion technique. The glass fibre probe is also used to measure radial gas fraction distributions about 115 cm above the air distributor, at several superficial gas velocities. The glass fibre probe appeared to suffer from a serious anisotropy. Two methods to correct for this anisotropy are discussed. This results in far more uniform gas fraction distributions than measured by other researchers, who did not correct the anisotropy. Various models regarding the flow in bubble columns are presented. A Reynolds number criterion is suggested to predict the transition from homogeneous to turbulent bubble flow. Another model predicts the radial non-uniformity of the gas fraction distribution. It assumes the presence of a radial pressure gradient due to the centrifugal force of turbulent liquid velocity fluctuations. A momentum balance method based on the model of Geary and Rice (1992) is used to calculate liquid recirculation velocities. The structure and time-dependent behaviour of gas fraction fluctuations are examined by analysing the signals from one or two glass fibre probes with correlation techniques. A stochastic model is presented to describe the structures of these fluctuations. This model takes into account the characteristic properties of these structures such as their size, intensity, velocity, and lifetime. The values of these properties are estimated by fitting the measured covariance functions of the glass fibre probe signals to the model predictions. These fluctuations are also examined by analysing signals from pressure transducers mounted flush with the column wall. These pressure fluctuations are compared to the predictions of the gas fraction fluctuation model. However, they seem to originate from more than one effect.