Improved Predictions of Contaminant Degradation in Water Treatment Reactors

The efficacy of fundamental water treatment processes depends on reactor hydraulics. Despite the importance of reactor hydraulics, oversimplified hydraulic models have been used for the design, operation, and regulation of water treatment reactors. The most commonly used model assumes plug flow reactor (PFR) behavior with residence time equal to the time for the first 10 percent of flow to leave the reactor (PFR t10). This simplification is overly conservative when targeting low log reductions of contaminants, and may also overestimate treatment efficacy when targeting higher log reductions, such as in water reuse applications. The overall goal of this dissertation was to improve the prediction of contaminant degradation in water treatment reactors by accurately modeling reactor hydraulics. This goal was met through the following three objectives: (i) development of accurate models for residence time distribution (RTD) (i.e., macromixing); (ii) assessment of flow segregation and earliness of mixing (i.e., micromixing) in full-scale water treatment reactors; and (iii) quantitative evaluation of the effect of RTD model selection on predictions of contaminant degradation. This work generated a number of major conclusions and contributions to the modeling of contaminant degradation. (i) Reactor network (RN) models accurately represented observed RTD using fewer fitting parameters than alternative models and (ii) an open-source tool was created to fit RN models to tracer data. (iii) Micromixing was observed to be prevalent in full-scale reactors, and (iv) the tanks-in-series (TIS) and certain RN models most accurately represented micromixing. (v) Micromixing had the greatest impact on predictions of pathogen disinfection when specific lethality coefficient and disinfectant decay rate were high. (vi) The PFR t10 model may cease to be conservative when predicting contaminant reductions >2-log. (vi) Designing reactors for 1-log reduction using the PFR t10 model may increase capital costs by 10-80% relative to an accurate RTD model like the RN model; (vii) at 6-log reduction, properly sizing oxidation processes using the RN model may increase costs by over 100% relative to the PFR t10 model. Overall, this work provides a fundamental basis for the rational design, operation, and regulation of water treatment processes using the TIS or RN models.Doctor of Philosoph