Pilot scale phosphorus recovery from anaerobic digester supernatant

In this study, the pilot-scale struvite crystallization process developed at UBC was operated continuously for recovering phosphate, in the form of struvite, from anaerobic digester supernatants from the Annacis Island Wastewater Treatment Plant and the Lulu Island Wastewater Treatment Plant. In addition, the process performance for a synthetic supernatant with high phosphate concentration (100-190 mg/L) was verified. Study results showed that the process was capable of removing more than 90 % of ortho-phosphate from both the synthetic supernatant and the digester supernatants. Approximately 90 % of removed phosphate was recovered as harvestable struvite crystals. The desired phosphate removal efficiency was achieved through controlling the inlet supersaturation (SS) ratio, operational pH and magnesium dosage in the supernatant. It was possible to reduce the effluent ortho-phosphate concentration to less than 5 mg/L through choosing optimal operational conditions. Chemical analysis of the recovered crystals showed very pure struvite (91.2 % and 94.1 % by weight for crystals produced from the Annacis supernatant and the Lulu supernatant, respectively.) with small amounts of calcium and carbonate, and traces of iron and aluminum. Most of recovered crystals were round, hard and larger than 2 mm in mean diameter over the course of the study. The crystal retention time in the reactor and the magnesium dosage in the supernatant were identified as two major factors affecting the size, density, hardness and morphology of recovered struvite crystals. Determination of the struvite solubility product (Ksp) showed that there were significant differences among Ksp values for the synthetic, Annacis and Lulu supernatants, due to combined effects from impurity ions and suspended solids, as well as other unknown factors. In this scenario, the conditional solubility product (Ps) was considered to be more useful than Ksp in operating the struvite crystallization process and predicting process performance. A cost analysis showed that the application of air stripping or a high magnesium dosage in the supernatant could not reduce the total chemical costs, since both Annacis and Lulu supernatants used in this study had a high pH value around 8.0. However, a further study on other anaerobic digester supernatants with low pH was recommended. Two struvite models developed by Britton and Potts, respectively, were used to predict the process performance and determine the operational parameters. Both were validated through comparing the predicted results with the actual operational data. The comparison showed that the former predicted the actual results with relatively high accuracy; however, the latter demonstrated a large deviation from the real results, probably caused by the Ksp value used in the model.Applied Science, Faculty ofCivil Engineering, Department ofGraduat