Optimization of Wastewater Treatment Process in a Bioreactor Through Hydrodynamic-biokinetic Modeling and Experimental Studies

Abstract

Membrane bioreactor has emerged as one of the leading technologies for treating municipal and industrial wastewater due to its efficiency in producing high-quality effluents. One of the significant challenges in bioreactors is the high energy and operating costs. The diffused aeration process of a bioreactor is the most energy-intensive operation amounting to 45-75% of the plant energy costs. This study attempts to optimize the wastewater treatment (WWT) process in a bioreactor through modeling and experimental studies. The overall aim is to develop efficient models which can be used to reduce the treatment costs of the WWT process while increasing the treatment efficiency. As a first objective, a multiphase mixture computational fluid dynamics (CFD) model was developed using k-𝜺 turbulence closure equations and a discrete population balance model (PBM) add-on with specific bubble classes to predict the oxygen mass transfer in synthetic water. The validated model was extended for sensitivity analysis for a diffused aeration system in a bench-scale aeration tank. Results show that the volumetric oxygen mass transfer coefficient increased by 15 %, with a decrease in air bubble size by 10 %. In a diffuse aeration system, the air bubbles had a wider distribution, with a larger diameter near the bottom of the bioreactor, and narrow distribution, with a smaller bubble size at the top of the bioreactor. As a second objective, an integrated model was developed by combining the multiphase CFD model, the PBM sub-model, an activated sludge submodel, and a combined extracellular polymeric substance (EPS) – soluble microbial product (SMP) (CES) submodel to investigate the oxygen uptake rate, the aeration efficiency, and treatment efficiency in bioreactors. Three different scale bioreactors, namely, i) case 1- laboratory, ii) case 2 – pilot, and iii) case 3- full-scale system, were studied. The model predictions on water quality were validated well with the experimental results. The validated model was used for sensitivity analysis to identify optimum conditions. The maximum percentage reduction in chemical oxygen demand and total nitrogen were 17 % and 18 %, respectively, for case 3. Also, a reduction of 32 % in the cost of aeration was observed when the bubble size was reduced to 5 mm (from the current value of 7 mm). The third objective focused on developing a multiphase CFD – porous- CES model to investigate the effect of hydrodynamics on biofouling and the effect of the EPS and SMP on the cake layer formation. The developed model was validated with experimental observations from the laboratory-scale ultrafiltration hollow fiber membrane setup. Observations showed that as the filtration time increased, the transmembrane pressure (TMP) increased, and the permeate flux decreased. Furthermore, in experimental set 2 (synthetic wastewater with sludge seeding), the effect of cake deposition on TMP and permeate flux was 17% and 1.5% higher, respectively, compared to experimental set 1 (synthetic wastewater with yeast sludge). The validated model was then used to investigate the sensitivity of the CES submodel by comparing it with the sectional resistance submodel. It was observed that the sectional resistance model underpredicted the mass of cake deposited by 13 % and overpredicted the limiting flux by 4 %. The results suggest the importance of accounting for the influence of EPS and SMP on the cake layer formation and biofouling. The fourth objective of this thesis reports a BioWin©- ASM for optimizing the biological nutrient removal (BNR) in a 55 million liters per day sewage treatment plant (STP). The proposed modification was to incorporate an intermediate virtual anoxic zone to achieve simultaneous nitrification-denitrification and total dissolved phosphorus (TDP as PO4) removal. The hydraulic residence time (HRT), dissolved oxygen (DO), and mixed liquor suspended solids (MLSS) of the bioreactor were varied to identify the optimum operating conditions. The optimum DO and MLSS levels were identified as 4 mg/L and 4000 mg/L, respectively, and the optimum HRT was found to be 2 h. in the aeration zone, 1 h. in anoxic, and 3 h. in the reaeration zone. Implementing these modifications in the STP, with minimal operational interventions and no capital costs, improved its performance as predicted by the model. The total nitrogen and TDP (as PO4) reduced from 20 mg/L to 8 mg/L and 3.5 mg/L to 0.9 mg/L, respectively, and met the revised discharge standards. This intervention gave a cost saving of approximately 5.6 million USD. This work has demonstrated that the numerical models can be successfully used to optimize the treatment efficiency while reducing the capital (membrane replacement) costs and operating (aeration) costs of a bioreactor. The time and efforts required for identifying the optimum conditions through numerical modeling are significantly less than physically characterizing the bioreactor (and varying the conditions to optimize them).