Nutrient recovery from source separated urine: modelling and experimental study

Abstract

Phosphorus recovery from urine can make an important contribution towards global food security and environmental sustainability. Phosphorus can be recovered through formation of struvite, a slow-release fertiliser, which forms by precipitation. Its precipitation was conducted by feeding magnesium and urine solutions to a continuous, well-mixed reactor. The resultant struvite particles were retained within the reactor, while the depleted nutrient solution exits. Most of previous research had arbitrary selected operational conditions. These studies did not consider any assessment of struvite productivity and neither thermodynamics. Struvite productivity comprises the quantity of struvite produced per volume of reactor. Thermodynamics simulations assisted to select a suitable conditions close to equilibrium. This work aimed to increase our process understanding through mathematical modelling, computer simulation and directed experimentation. Model Development: This thesis implemented a mathematical model comprising chemical thermodynamics, rate kinetics and mass balance relations. Thermodynamics includes chemical speciation where the activities of complexes are estimated from the total concentration of every element. Thermodynamic simulations can predict the saturation index at non-equilibrium conditions, and the quantity of struvite mass when the saturation index is set to zero at equilibrium. Kinetic models of the linear growth rate of struvite particles (GL) as a function of the struvite saturation index and parameters (GL = k ∙ Slⁿ) were evaluated. The mass balance considered eight elements present in urine composition (Mg, N, P, C, Na, K, S and Cl). This model considered a two-zone reactor with a mixing and reacting zone (bottom) and a settling zone (top). The model equations were solved in Engineering Equation Solver and gPROMS to design experimental struvite precipitation at laboratory and bench-scale. Model Simulations: The model was applied to batch, fed-batch and continuous struvite precipitation scenarios. Simulations predicted the struvite saturation index (SIMAP) at non-equilibrium conditions, which drove the mass transfer, and the resultant struvite mass and element concentrations of P and Mg in solution. The continuous reactor system consisted of a stirred vessel into which synthetic urine and MGSO₄.7H₂O solutions were fed. Simulations assessed the effect of the Mg quantity added, seed quantity and hydraulic residence time within the liquid volume of the reactor. Simulations showed that Mg/P molar feed ratio greater unit does not increase the struvite production rate. Close to equilibrium continuous precipitation was achieved by increasing the seed mass and the hydraulic residence time. These simulations also showed that kinetic parameters are insensitive when conditions approach equilibrium. The rate coefficient k = 1e − 6 m/h, and the order n = 1 were suitable for process design. Experimentation: Struvite precipitation in batch experiments validated the nominated thermodynamics model by comparing the predicted P and Mg concentrations to the actual equilibrium solutions. Struvite mass was also accurately predicted. Elemental mass balances in P and Mg in each batch experiments also demonstrated very good consistency in the experimental results, leading to enhanced confidence in the results. The dynamic model was validated against two separate continuous experiments, each operating more than 4 days. They were performed at the same hydraulic residence time (feed flow rate = 0.6 L/h), but varied the Mg/P feed molar ratio (0.34, 0.64 and 1.29). Model predictions of P and Mg concentration in the liquid phase well matched the laboratory measurements. Predicted struvite mass was also within measurement uncertainty. The precipitated struvite was characterised by x-ray diffraction, elemental analysis of Mg and P, microscopy and particle size using electric sensing zone.