| Deep bed filtration is an essential and widely used process in the treatment of drinking water.; The goal of this research was to further the understanding of the filtration mechanisms and to also further the efforts of adequate modeling of the filtration process. The macroscopic model used in this research included the detachment mechanism and thus, it was able to model the entire cycle of filtration. An assessment of the ability of the model to predict the filtration process for varying types of particles, mainly hydrophobic and hydrophilic particles, was also an important aspect of this research. The laboratory experiments were conducted under well-defined conditions, in which the filter operating parameters were varied. Those parameters were: filtration rate, filter media depth, influent particle concentration, and hydrophobic/hydrophilic nature of the particles. The results of the laboratory scale experiments illustrated the significant difference in removal efficiency and headloss development when hydrophobic and hydrophilic particles are presenting the influent water. When the influent suspension contains hydrophilic particles, the increase in headloss over the course of the filter run is minimal, while there is consistently little or no removal of particles during the experiment. On the other hand, hydrophobic particles are removed more readily during the filtration cycle, which leads to more rapid increase in headloss over the filter run and gradually improving removal efficiency.; In order to alleviate the difficulty in having to determine model parameter values through experimental data, empirical equations for the direct calculation of the ripening, detachment, and headloss parameters were determined. Literature and prior research has indicated that values of the three parameters are affected by filtration rate, filter media depth, diameter of the media grains, and diameter of the influent particles. Since the experiments conducted as part of this research incorporates a wide array of filtration conditions, it was possible to derive empirical equations for the three parameters. The empirical equations were able to accurately predict the parameter values generated by the model through the simultaneous fit of the experimental removal and headloss data. The empirical equations were most accurate for the ripening and headloss model parameters. The equation derived for the detachment parameter was less precise than the equations for the other two parameters. (Abstract shortened by UMI.)... |
