Initial performance and scalability assessment of a pilot-scale reverse osmosis system for nutrient recovery from source-separated urine

Human urine is a nutrient-rich resource containing nitrogen, phosphorus, and potassium, yet it continues to be treated as waste. Reverse osmosis (RO) offers a scalable, energy-efficient method for concentrating these nutrients into a liquid fertilizer. However, its practical application is often hindered by membrane scaling, particularly calcium carbonate (CaCO 3 ) formation when treating urine stabilized with calcium hydroxide (Ca(OH) 2 ). While laboratory-scale studies have demonstrated various calcium removal pretreatments, their integration with RO has yet to be evaluated at the pilot scale. This study addresses this gap by evaluating the performance of a 300 L·h -1 pilot-scale RO system, focusing on the effectiveness of calcium removal pretreatment methods and membrane performance, specifically water removal, permeate flux, and nutrient rejection. The study was conducted in two distinct phases. First, three pretreatment methods (air bubbling, pure carbon dioxide (CO 2 ) sparging, and potassium bicarbonate dosing) were evaluated and compared in terms of calcium removal, process kinetics, energy requirements, and total treatment cost. Secondly, the RO system’s performance was evaluated using air-bubbled human urine. This pretreatment was selected for the pilot-scale RO trials to establish a conservative baseline for system performance, as its highest residual calcium (post pretreatment) and therefore scale-up constraints presented a worst-case flux and scaling scenario. Additionally, concentration-matched synthetic solutions were used to isolate the specific effects of osmotic pressure, scaling, and organic fouling on permeate flux and nutrient rejection. Among the pretreatment methods investigated, potassium bicarbonate dosing had the fastest reaction kinetics (99% calcium removal in 0.17 h) but resulted in the highest total treatment cost ($28.2 m -3 ). In contrast, air bubbling was kinetically limited by scale-up constraints (0.39 L air ·min -1 ·L urine −1 ) and the low ambient CO 2 partial pressure (0.04%), requiring 33 h to achieve a 78% calcium removal (824–182 mg·L -1 ). This prolonged aeration added 66 kWh·m -3 to the baseline RO energy requirement of 8.64 kWh·m -3 , rendering it energetically prohibitive. Consequently, carbon dioxide sparging emerged the most cost-effective integrated process ($6.96 m -3 ), decoupling the process from high chemical costs and the energy penalty of prolonged aeration, thereby enhancing process resilience to electricity price fluctuations. During RO operation, linear flux declines across all synthetic and human urine streams (R 2 = 0.926–0.952) confirmed that performance was governed by osmotic pressure, as opposed to membrane fouling or scaling. Urea rejection decreased from 87% to 76% as feed concentration increased, driven by enhanced diffusive flux at the membrane interface, exposing a critical operational trade-off. The pilot-scale concentration using RO (67%–69% water removal) successfully produced a liquid fertilizer with urea and potassium concentrations of up to 11.5 g-N·L -1 and 5.2 g-K·L -1 , respectively. These results demonstrate the RO system can decouple the final product quality from feed composition variability by adjusting water removal rates, thereby ensuring a standardized fertilizer composition regardless of the initial urine concentration.