Pressurized cell-side feeding versus submerged PVDF ultrafiltration reactor for in-situ ozonation: A comparative...

The configuration of a coupled ultrafiltration (UF) with polyvinylidene fluoride (PVDF) membrane and ozonation reactor for in-situ ozonation was critical in drinking water treatment. Thus, different pilot-scale systems combining in-situ ozonation with either pressurized cell-side feeding PVDF UF or submerged PVDF UF followed by biological activated carbon filters (O₃/PUF/BAC or O₃/SUF/BAC) were comparatively analyzed. Experimental results indicated that the membrane fouling resistance of O₃/PUF decreased by 41 %, compared with that of O₃/SUF, owing to cross-flow flushing and enhanced ozonation driven by positive transmembrane pressure. These conditions also induced microbial lysis on the membrane surface, leading to the release of intracellular organic matters (IOMs) in the O₃/PUF. Ozonation further degraded these released IOMs into smaller molecules that passed through the membrane, increasing the effluent protein content by 48 %. Conversely, the O₃/SUF showed minimal IOMs leakage and presented stabler effluent quality due to foulant layer rejection and biological metabolism. For example, the relative abundance of Burkholderia-Caballeronia-Paraburkholderia reached 0.40 in the O₃/PUF and 0.84 in the O₃/SUF, suggesting a potential role in organics degradation and UF effluent quality improvement. While the O₃/PUF effluent showed elevated haloacetonitriles (HANs) formation potential and disinfection toxicity, subsequent BAC treatment effectively mitigated these effects. Therefore, while PUF demonstrates superior membrane fouling control, it carries the risk of protein leakage and increased generation of HANs during later operational stages. However, these issues were resolved when integrated with a post BAC filter. In comparison, the O₃/SUF/BAC system achieved stable water purification without the deterioration of the effluent quality during the whole operation, but suffered from severe membrane fouling.