Membrane fouling mechanism of submerged membrane bioreactor during erythromycin removal

Background: Based on the previous studies, antibiotics can have affected biological properties of biomass and fouling properties of mixed liquor in aeration tank. The present study was conducted to explore the fouling mechanisms of membrane bioreactor (MBR) system during the treatment of wastewater containing erythromycin (ERY) antibiotic under several mixed liquor suspended solids (MLSS) concentrations. Methods: A lab-scale two-chamber MBR system equipped with a polypropylene hollow fiber submerged membrane was fed with synthetic wastewater containing different initial concentrations of ERY. MBR system was operated under the constant flux mode and different MLSS concentrations (5.0-13.0 g/L) and the obtained results were evaluated using different individual and combined fouling models. Results: The variation of MLSS concentrations had not significantly affected the kind of best-fitted model. From the individual models, the standard model indicated the best performance for permeate prediction under different MLSS concentrations (R2adj > 0.997). For all studied MLSS concentrations, the R2adj values of combined fouling models were higher than 0.986 and demonstrated good fitness performance of combined models compared to individual models. Overall, the cake-intermediate model showed the lowest fitness, and cake-complete and complete-standard models were the most successful models in filtrated volume prediction in comparison with other combined fouling models. Conclusion: This study indicated that mechanistic models are suitable for fouling prediction of MBR systems in ERY removal and under a wide range of MLSS concentrations and provide valuable information on fouling mechanisms of full-scale MBR systems.

Automated summary

This study investigated the fouling mechanisms of a membrane bioreactor (MBR) system during the treatment of wastewater containing erythromycin (ERY) antibiotic under different mixed liquor suspended solids (MLSS) concentrations. The results showed that the variation of MLSS concentrations did not significantly affect the best-fitted model, and the combined fouling models performed better than individual models. The study provides valuable information on fouling mechanisms in MBR systems, which is important for full-scale MBR systems.