Protecting the quality and quantity of our water resources requires advanced treatment technologies capable of removing nutrients from wastewater. This research work investigated the capability of one such technology, a hollow-fiber membrane-aerated biofilm reactor (HFMBR), to achieve completely autotrophic nitrogen removal from a wastewater with high nitrogen content. Because the extent of oxygenation is a key parameter for controlling the metabolic processes that occur in a wastewater treatment system, the first part of the research investigated oxygen transfer characteristics of the HFMBR in clean water conditions and with actively growing biofilm. A mechanistic model for oxygen concentration and flux as a function of length along the non-porous membrane fibers that comprise the HFMBR was developed based on material properties and physical dimensions. This model reflects the diffusion mechanism of non-porous membranes; namely that oxygen follows a sorption-dissolution-diffusion mechanism. This is in contrast to microporous membranes in which oxygen is in the gas phase in the fiber pores up to the membrane surface, resulting in higher biofilm pore liquid dissolved oxygen concentrations. Compared to offgas oxygen analysis from the HFMBR while in operation with biofilm growing, the model overpredicted mass transfer by a factor of approximately 1.3. This was in contrast to empirical mass transfer coefficient-based methods, which were determined using either bulk aqueous phase dissolved oxygen (DO) concentration or the DO concentration at the membrane-liquid interface, measured with oxygen microsensors. The mass transfer coefficient determined with the DO measured at the interface was the best predictor of actual oxygen transfer under biofilm conditions, while the bulk liquid coefficient underpredicted by a factor of 3. The mechanistic model exhibited sensitivity to parameters such as the initial lumen oxygen concentration (at the entry to the fiber) and the diffusion coefficient and partitioning coefficients of oxygen in the silicone membrane material. The mechanistic model has several advantages over empirical-based methods. Namely, it does not require experimental determination of KL, it is relatively simple to solve without the use of advanced mathematical software, and it is based upon selection of the membrane-biofilm interfacial DO concentration. The last of these is of particular importance when designing and operating HFMBR systems with redox (aerobic/anoxic/anaerobic) stratification, because the DO concentration will determine the nature of the microenvironments, the microorganisms present, and the metabolisms that occur. During the second phase of the research, the coupling of two autotrophic metabolisms, partial nitrification to nitrite (nitritation) and anaerobic ammonium oxidation, was demonstrated in a single HFMBR. The system successfully treated a high-strength nitrogen wastewater intended to mimic a urine stream from such sources as extended space missions. For the last 250 days of operation, operating with an average oxygen to ammonia flux (J_O₂/J_NH₄⁺) of 3.0 resulted in an average nitrogen removal of 74%, with no external organic carbon added. Control of nitrite-oxidizing bacteria (NOB) presented a challenge that was addressed by maintaining the J_O₂/J_NH₄⁺ below the stoichiometric threshold for complete nitrification to nitrate (4.57 g O₂ / g NH₄⁺). The DO-limiting condition resulted in formation of harmful gaseous emissions of nitrogen oxides (NO, N2O), which could not be prevented by short-term control strategies. Controlling JO2/JNH4+ prevented NOB proliferation long enough to allow an anaerobic ammoniaoxidizing bacteria (AnaerAOB) population to develop and be retained for >250 days. Addition of a supplemental nutrient solution may have contributed to the growth of AnaerAOB by overcoming a possible micronutrient deficiency. Disappearance of the gaseous nitrogen oxide emissions coincided with the onset of anaerobic ammonium oxidation, demonstrating a benefit of coupling these two autotrophic metabolisms in one reactor. Obvious differences in biofilm density were evident across the biofilm depth, with a region of low density in the middle of the biofilm, suggesting that low cell density or exocellular polymeric substances were primarily present in this region, Microbial community analysis using fluorescence in situ hybridization (FISH) did not reveal consistent trends with respect to length along the fibers, but radial stratification of aerobic ammonia-oxidizing bacteria (AerAOB), NOB, and AnaerAOB were visible in biofilm section samples. AerAOB were largely found in the first 25% of the biofilm near the membrane, AnaerAOB were found in the outer 30%, and NOB were found most often in the mid-depth region of the biofilm. This community structure demonstrates the importance of oxygen availability as a determinant of how microbial groups spatially distribute within an HFMBR biofilm. The combination of these two aspects of the research, predictive oxygen transfer capability and the effect of oxygen control on performance and populations, provides a foundation for future application of HFMBR technology to a broad range of wastewaters and treatment scenarios. / Ph. D.

Mathematical models of a nitritation: anaerobic ammonia oxidation (anammox)-coupled biofilm in a counter-diffusion hollow fiber membrane bioreactor (HFMBR) and a nitritation: anammox-coupled biofilm in a co-diffusion rotating biological contactor (RBC) were developed and implemented using AQUASIM. Four different start-up scenarios on the nitritation: anammox-coupled biofilm in an HFMBR were investigated. The supply of oxygen was simulated with the flow through the lumen of the hollow fiber membrane. For the four scenarios, two scenarios investigated the start-up when nitrite was supplied in the feed while the other two scenarios investigated when the source of nitrite was through nitritation only. The results showed that the presence of nitrite in the feed facilitated the start-up of the reactor. In addition, the results also showed that increasing oxygen flux through the membrane up to a certain ratio of ammonia flux with oxygen flux affected reactor performance by improving nitrogen removal and reducing start up time. For the nitritation: anammox-coupled biofilm in an RBC, four different process options were investigated: the number of reactors, the initial anammox (AnAOB) biomass fraction, the bulk oxygen concentration and the maximum biofilm thickness. Modeling results revealed that the steady state total nitrogen removal in RBC reactors in series occurred primarily in the first and second reactors. It is concluded that the number of reactors in series dictates the effluent performance and, therefore, this number can be selected depending upon the desired total nitrogen removal. Simulation results also revealed that increasing the initial AnAOB biomass fraction from 0.01% to 1.0% had no effect in the steady state nitrogen removal but had an effect in the required time to reach the steady state total nitrogen removal and the maximum biofilm thickness. Modeling results of the third process option showed that increasing the bulk oxygen concentration in the reactor from 0.2 g/m3 to 5 g/m3 linearly increased the steady state total nitrogen removal and reduced the time to reach the maximum biofilm thickness. Beyond 5 g/m3, steady state total nitrogen removal decreased. In addition, simulation results revealed that the thicker biofilm clearly showed a more linear correlation between the increase in bulk oxygen concentration and the increase in the steady state total nitrogen removal within a range of bulk oxygen concentrations. The results showed that RBC performance could be controlled by several process options: the number of reactors in series, initial biomass fraction, the bulk oxygen concentration and the maximum biofilm thickness. The mathematical modeling results for the HFMBR and RBC have shown that both have potential as carriers for nitritation: anammox-coupled biofilms targeted at the removal of nitrogen in the wastewater. / Master of Science