The kinetics of ozone disinfection was investigated using a completely mixed reactor equipped with a fast sampling apparatus to eliminate the interference from decreasing ozone residual and minimize the additional inactivation from sampling. The experimental data were then used to fit five models, representing the three basic phenomena of inactivation. Both the Chick-Watson model and the presently proposed model were capable of predicting the survival of laboratory incubated bacteria. However, the latter model is more applicable for naturally isolated microorganisms that may exhibit the tailing effects. The empirical coefficient of dilution was found to be around 3, strongly suggesting that ozone concentration was much more important than contact time in determining the efficiency of the disinfection process. The overall activation energy was 85 kJ/mol, indicating the rate of bacterial inactivation with ozone is possibly reaction- controlled.

Both the axial dispersion model (ADM) and the back flow cell model (BFCM) were derived to describe the fundamental relationship between the contactor hydrodynamics, gas-liquid mass transfer, ozone decay and susceptibility of microorganisms. It was shown that the countercurrent flow and cocurrent flow bubble column contactors exhibited different patterns of dissolved ozone profiles. For countercurrent flow, the dissolved ozone was monotonically increasing along the depth of contactors, while for cocurrent flow it was much flatter, possibly with a maximum somewhat in the middle region. For reactive contacting chambers, the dissolved ozone was gradually decreasing as water passes through the contactors. In both countercurrent and cocurrent flow contactors, there was an occurrence of dissolved ozone concentration jump at water inlet due to backmixing. Consequently, different approaches were recommended to relate their characteristic concentrations to readily observed measurements for predicting the disinfection efficiency.

As compared to the ADM, the BFCM is easier to be solved, simpler to be formulated and more flexible to be applied for the complicated flow systems. Its applicability to predicting the performance of ozone contactors was verified experimentally using the pilot scale tests. It was found that for clean waters, the predictions were consistent with the experimental observations, based on the assumption that the ozone decay in water is a first-order rate process. However, this assumption was inadequate for most natural waters. Not only the raw water quality, but also the extent of ozonation should be considered. Accordingly, a variable defined as specific ozone utilization rate w was suggested. It was found that the w decreases exponentially with the amount of utilized ozone with an asymptotical minimum.

It was found that the backmixing in the contactor was affected by both the gas flowrate and water flowrates, with the latter dependence being stronger. However, the volumetric mass transfer coefficient ka is mainly determined by the gas flowrate, while independent of the water flowrate and flow directions. A power law relationship for the tested pilot column can be expressed as:​

k1a, s-1 = 1.19 x u60.82 0.35 m/min < u1 < 0.87 m/min 0.038 m/min < ug < 0.27 m/min

where u_G and u_L are the superficial gas and water velocity, respectively.