Understanding key factors determining the effect of particle scouring efficiency on membrane fouling mitigation in AnFMBRs: Correlation analysis via machine learning

Abstract

Particle scouring methods offer several advantages, including low energy consumption, ease of operation, and effective fouling control. The anaerobic fluidized bed membrane bioreactor is considered to be the most efficient wastewater treatment method with zero energy consumption, as it produces bioenergy and has lower energy requirements compared to conventional processes. However, membrane fouling in AnFMBRs can be affected by various factors such as fluidized particle hydrodynamics, particle properties, and operating conditions. To identify the primary factors affecting membrane fouling control in AnFMBRs, machine learning methods were utilized. Analysis of datasets from previous studies revealed several key findings. The location of the membrane in the reactor, particle momentum, and particle size were identified as the major parameters that govern membrane fouling reduction. Additionally, the optimal condition for AnFMBRs was determined to involve a membrane height-to-reactor height ratio of ≤0.5 and the use of particles with diameters of 1.5–3.0 mm. Furthermore, the use of smaller fluidized particles was shown to decrease particle scouring efficacy, making it less cost-effective. Machine learning models were successfully established to predict fouling, thus providing a comprehensive understanding of the dominant factors influencing membrane fouling control in AnFMBRs and offering an efficient method for optimizing fouling mitigation.