System dynamic modelling of nanofiltration membrane in processing river water into drinking water

This research aimed to determine factors influencing the performance of nanofiltration (NF) membrane as a drinking water treatment unit using system dynamic (SD) modelling. STELLA was used to model the system and to predict the behavioural pattern and performance of the system in removing pollutants (chemical oxygen demand (COD), Atrazine (ATZ), Acetaminophen (ACT), and phosphate) and producing permeate flux. SD modelling stages consisted of system identification, model conceptualization, structure building, verification, and validation. The main model of NF membrane has four sub-models: size exclusion, electrostatic repulsion, rejection, and flux. The SD model was run using experimental data obtained under laminar crossflow conditions at 50 psi and ambient temperature (25 ± 2°C). Results showed that the concentration of pollutants in the feed water, membrane characteristics, feed characteristics, and operational parameters played significant roles in the NF process. Simulation results by SD modelling matched with the real system with an error value of less than 0.05. NF membrane removed 76.55–86.04% of COD, 60.01–70.24% of ATZ, 33.77–49.22% of ACT, 54.77–83.08% of phosphate, and produced a permeate flux of 15.57 L/m².h. ATZ and ACT concentration in permeate water still exceeded the international standard, in which increasing pressure, pH, and pore size were simulated afterward. All three scenarios had a positive effect on the removal of those parameters, resulting in an average permeate concentration lower than the permissible limit (0.003 mg/L and 0.2 mg/L, respectively, for ATZ and ACT).