Optimizing PVDF-Tannic acid membrane for treating actual river water using response surface methodology

• Optimized PVDF-tannic acid membrane achieved 91.5 % NOM rejection with RSM. • Developed membrane shows high RWP of 546.2 L/m²·h·bar under optimal conditions. • Coating duration and Fe³⁺ significantly influence water permeability outcomes. • Membrane hydrophilicity improved, reducing fouling with TA-modified surface. • Experimental results closely match predictions, validating RSM modeling accuracy. In this study, we refined the development process of polyvinylidene difluoride (PVDF) membranes using Response Surface Methodology (RSM). to address a key research gap in multi-step tannic acid (TA)–based membrane modification for real river water treatment. This multi-step approach integrates phase inversion with surface coatings to minimize morphological alterations commonly encountered in single-step methods. Our goals were to maximize both pure and river water permeability (PWP and RWP, respectively) and achieve high rejection of natural organic matter (NOM). The Central Composite Design matrix was utilized to effectively assess how various factors influence these goals and develop predictive regression models. Three key parameters were selected: tannic acid [TA] in the range of 0–1 g/L, [Fe 3+ ] in the range of 0–1 g/L, and coating duration spanning from 0.5 to 5 min. Performance assessment of the membranes was carried out using a cross-flow microfiltration system enhanced practical applicability by treating actual river water. The results indicate that the developed membranes exhibit enhanced hydrophilicity and anti-fouling properties compared to the pristine PVDF membranes. According to the analysis of variance, all parameters hold statistical significance in relation to PWP and NOM rejection, with [Fe 3+ ] and coating duration being particularly significant for the RWP model. The optimal experimental conditions for PWP, RWP, and NOM rejection were [TA] at 0.542 g/L, [Fe 3+ ] at zero, and a coating duration of 0.5 min. Remarkable performances were obtained, including a maximum NOM rejection of 91.546 %, RWP of 546.203 L/m 2 h bar, and PWP of 739.997 L/m 2 h bar. The error percentages for PWP, RWP, and NOM rejection between the predicted and experimental values were found to be 7.7 %, 2.9 %, and 1.9 %, respectively, confirming the validity and accuracy of the models.