| Summary: | In this study, the workability of a newly designed, scaled up pilot-scale reactor was examined. The work was carried out at Lulu Island Wastewater Treatment Plant (LIWWTP) in Richmond, BC. One of the main objectives of the study was to demonstrate the ability of the reactor to remove at least 70% of the phosphate present in the centrate. Results showed that the reactor was capable of removing, under controlled conditions, over 90 % of phosphate and 4% of ammonia-nitrogen. Phosphate concentration in the effluent could be lowered to 5 mg/L. More than 85% of the phosphate removed was recovered as harvestable struvite pellets. The desired degree of phosphate removal was achieved by controlling the reactor supersaturation ratio. This ratio was in turn, controlled by varying the magnesium input and/or operating pH. Data collected indicated that it was possible to achieve over 90% P-removal at a pH of 7.5; this is contrary to the information found in literature, which recommends that a higher pH (8.2~9.0) is required. Factors that affected phosphate removal were the operating pH, the reactor SSR, the N:P and Mg:P molar ratios. The determination of struvite solubility product with centrate and distilled water gave different values. The solubility product was dependent on the water tested, the pH of the solution and the temperature. The temperature coefficient and enthalpy at 25°C were 1.21 and 137.5 kJ/mol, respectively. Analysis of the harvested product showed that the pellets were composed of nearly pure struvite (96% by weight), with small amounts of calcium and traces of aluminum and iron. ICP/MS testing of struvite samples found lower heavy metal content than that present in P-rock. The reactor SSR was determined to be the significant factor for pellet size; Mg:P molar ratio and upflow velocity determining the pellet hardness. Through the process, it was possible to grow pellets larger than 4.75 mm. Two struvite models, developed by Potts and Britton at UBC, were used to predict the process performance along with the phosphate and ammonia effluent concentrations. The models were validated by comparing the predicted values with the actual operational data. Comparison of P-removal efficiency, effluent phosphate and ammonia concentrations showed that the former was more accurate in the prediction. A trial was made by using an artificial neural network, Neusciences Neuframe® 4.0 model, to predict the effluent phosphate concentration. This model was found to be inefficient in its prediction, largely due to the small number of data used to train the model. === Applied Science, Faculty of === Civil Engineering, Department of === Graduate
|
|---|
