Multivariate optimization of phosphate removal and recovery from aqueous solution by struvite crystallization in a fluidized-bed reactor

Abstract

Struvite crystallization has been widely studied for phosphate removal and recovery from aqueous systems. In this study, struvite crystallization was carried out in a fluidized-bed reactor. Multivariate optimization was conducted using Box–Behnken design (BBD) with influent pH, influent phosphate concentration, and Mg/P molar ratio as independent variables. The output variables comprised total and dissolved phosphate concentrations, ammonium and magnesium concentrations, and fines concentrations. Experimental values of the total phosphate and dissolved phosphate concentrations ranged from 25.6 to 109.4 mg/L and from 7.6 to 39.3 mg/L, respectively, while the fines concentration varied from 5.2 to 101.6 mg/L. Quadratic mathematical models describing the response behavior of experimental BBD data were generated for total phosphate, dissolved phosphate, and fines concentration. The model p-values ( <0.0001) were significant and their lack-of-fit p-values ( >0.05) were insignificant. Numerical optimization of process parameters was conducted to minimize total and dissolved phosphate, ammonium and magnesium concentrations, and fines concentration in the effluent. At influent phosphate concentration of 300 mg/L, the results converged to a set of operating conditions: pH 9.5 and Mg/P = 1.3. The close agreement between the data from the validation experiment and the model-predicted values (relative error < 10%) indicates the robustness of the models.