Affecting Membrane Fouling Research Articles

A deeper investigation of membrane fouling in anaerobic membrane bioreactors for wastewater treatment: influencing factors and fouling layer characteristics

Identifying the core parameters affecting membrane fouling and analyzing fouling layer characteristics are crucial for membrane fouling mitigation of anaerobic membrane bioreactors (AnMBRs). This study investigated the influence of various operating parameters on membrane fouling and the characteristics of different fouling layers. The ratio of flux to specific gas demand per unit of membrane area (SGD) was proposed as a key parameter for membrane fouling control and was applicable under various flux, SGD, and sludge concentration conditions. The membrane resistance and specific filtration resistance of foulants at high flux and sludge concentration reached 1.56×1012 m-1 and 3.56×1015 m-1/kg, respectively. Solid foulants accumulated on the membrane surface during rapid fouling stage. Protein-like pollutants accounted for a higher proportion (85%) of soluble foulants on the membrane surface. Humic acids were enriched on the cake/gel layer and the longitudinal enrichment process from the cake layer to the gel layer was uneven. Proteocatella (>45%) at the phylum level, Desulfovibrio (>3.1%), Syntrophobacter (>2.8%), and Treponema (>0.25%) at the genus level colonized in the gel layer and were the pioneers of membrane biofouling. Their enrichment on the membrane surface was primarily based on their own characteristics and was less sensitive to the operating conditions of AnMBRs. Therefore, this study provides a deeper understanding of membrane fouling formation process, which contributes to the long-term stable operation of AnMBR and scales up its engineering application.

Microfiltration of concentrated albumin solutions and the role of processing conditions on membrane fouling

Membrane processing is a subject of growing scientific and industrial interest as a valuable technology to replace conventional energy-intensive operations, but is hindered by the progressive decrease of permeate flux caused by membrane fouling. Although physico-chemical factors affecting membrane fouling have been widely investigated, less is known concerning the effects of processing conditions. Here, we focus on microfiltration of aqueous bovine serum albumin (BSA) solutions by using a novel miniaturized modular unit allowing to study both dead-end (DE) and crossflow (CF) filtration. Pure water flux measurements and Computational Fluid Dynamics (CFD) simulations were firstly performed to characterize the flow field inside the module. In presence of BSA, at the same transmembrane pressure (TMP) of 20 mbar and average pore size of 0.45 μm, a remarkable decrease in the fouling progression with time was found for CF with a peristaltic pump as compared to pressure-driven DE filtration. Furthermore, BSA concentration in the permeate was the same as in the feed for peristaltic CF, while negligible for DE. By either increasing transmembrane pressure or lowering membrane pore size, a stronger decay of permeate flux with time was observed. Scanning electron microscopy images of the membrane showed the presence of a thick surface cake layer under peristaltic CF, which became thinner by increasing TMP or reducing membrane pore size. Overall, the better performance of peristaltic CF can be interpreted in terms of reduced BSA residence time in the membrane, membrane deformability and formation of a permeable cake layer at lower TMP and larger membrane pore size. The innovative design of our filtration module and the combination of permeate flux and concentration measurements with CFD results and microscopy observations constitute the most significant aspects of this research work, which provides a multiscale picture of membrane fouling. The insight emerging from results aims to provide qualitative guidelines towards an optimal setting of processing conditions in membrane operations, which is highly needed to improve the membrane innovation pipeline.

Using artificial intelligence-based algorithms to identify critical fouling factors and predict fouling behavior in anaerobic membrane bioreactors

Membrane fouling is one of the major obstacles that hinder the widespread applications of anerobic membrane bioreactors (AnMBRs) for wastewater treatment. Due to the intrinsic complexity of membrane fouling, identifying critical fouling factors and predicting fouling behavior are of great significance in membrane fouling control. Herein, artificial intelligence (AI) algorithms and their modeling framework were developed to predict membrane fouling behaviors in AnMBRs. Operating parameters, biomass properties, and membrane characteristics were considered as input variables for membrane fouling prediction. The results of hyper-parameter optimization showed that the optimal architecture of artificial neural network (ANN) model for membrane fouling prediction was “14-9-6-1” while the best hyper-parameters of random forest (RF) model were n_trees = 1200 and n_features = 14. After hyper-parameter adjusting, RF had more robustness of predictive capabilities (R2=0.906, mean squared error (MSE) = 0.061) for membrane fouling than the ANN model (R2=0.800, MSE = 0.118). More importantly, the feature importance and Shapley additive explanations analysis indicated SMPp/SMPc (0.281) > EPSp/EPSc (0.110) > organic loading rate (0.106), which were the most critical factors affecting membrane fouling. Partial dependence plot analysis further verified the marginal and interaction effects of various input features on membrane fouling. The revealed partial dependence relationships of critical variables can provide theoretical reference for optimization of practical operation. Overall, this study established a novel AI-based approach for predicting membrane fouling and comprehensively understanding the complicated effects of various influencing factors on membrane fouling while overcoming the “black-box” nature of conventional models.

A Review on Membrane Fouling Prediction Using Artificial Neural Networks (ANNs).

Membrane fouling is a major hurdle to effective pressure-driven membrane processes, such as microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and reverse osmosis (RO). Fouling refers to the accumulation of particles, organic and inorganic matter, and microbial cells on the membrane's external and internal surface, which reduces the permeate flux and increases the needed transmembrane pressure. Various factors affect membrane fouling, including feed water quality, membrane characteristics, operating conditions, and cleaning protocols. Several models have been developed to predict membrane fouling in pressure-driven processes. These models can be divided into traditional empirical, mechanistic, and artificial intelligence (AI)-based models. Artificial neural networks (ANNs) are powerful tools for nonlinear mapping and prediction, and they can capture complex relationships between input and output variables. In membrane fouling prediction, ANNs can be trained using historical data to predict the fouling rate or other fouling-related parameters based on the process parameters. This review addresses the pertinent literature about using ANNs for membrane fouling prediction. Specifically, complementing other existing reviews that focus on mathematical models or broad AI-based simulations, the present review focuses on the use of AI-based fouling prediction models, namely, artificial neural networks (ANNs) and their derivatives, to provide deeper insights into the strengths, weaknesses, potential, and areas of improvement associated with such models for membrane fouling prediction.

Anoxic Treatment of Agricultural Drainage Water in a Venturi-Integrated Membrane Bioreactor.

Due to low sludge production and being a clean source without residuals, hydrogen-based autotrophic denitrification appears to be a promising choice for nitrate removal from agricultural drainage waters or water/wastewater with a similar composition. Although the incorporation of hydrogen-based autotrophic denitrification with membrane bioreactors (MBRs) enabled almost 100% utilization of hydrogen, the technology still needs to be improved to better utilize its advantages. This study investigated the anoxic treatment of both synthetic and real drainage waters using hydrogen gas in a recently developed membrane bioreactor configuration, a venturi-integrated submerged membrane bioreactor, for the first time. The study examined the effects of the inflow nitrate concentration, and the use of a venturi device on the removal efficiency, as well as the effects of the presence of headspace gas circulation and circulation rate on membrane fouling. The study found that using the headspace gas circulation through a venturi device did not significantly affect the treatment efficiency, and in both cases, a removal efficiency of over 90% was achieved. When the inlet NO3--N concentration was increased from 50 mg/L to 100 mg/L, the maximum removal efficiency decreased from 98% to 92%. It was observed that the most significant effect of the headspace gas circulation was on the membrane fouling. When the headspace gas was not circulated, the average membrane chemical washing period was 5 days. However, with headspace gas circulation, the membrane washing period increased to an average of 12 days. The study found that the headspace gas circulation method significantly affected membrane fouling. When the upper phase was circulated with a peristaltic pump instead of a venturi device, the membrane washing period decreased to one day. The study calculated the maximum hydrogen utilization efficiency to be approximately 96%.

The impact of sunlight on fouling behaviors and microbial communities in membrane bioreactors

Membrane fouling is a significant obstacle to applying membrane bioreactors (MBRs) for wastewater treatment. Here we report the impact of sunlight irradiation on membrane fouling, biopolymers, signal molecules, and microbial communities in MBRs. The degradation of signal molecules, which induce membrane biofouling, occurred through solar photolysis in batch tests. However, MBR sludge exposed to sunlight exhibited different biological behaviors creating more soluble microbial products (SMP) and signal molecules (particularly autoinducer-2). Cell lysis and deflocculation occurred when the MBR mixed liquor was exposed to sunlight. MBR fouling rates coincided with the temporal concentration profiles of SMP and signal molecules. Sunlight caused drastic changes in the MBR microbial community, stimulating the preferential growth of specific bacteria (e.g., Deinococcus Runella, Flavitalea, Glaiimonas, and Rurimicrobium). The nonmetric multidimensional scaling analysis of the MBR community structures with and without sunlight irradiation showed two distinct microbial community clusters and their reversibility. Network analysis based on Spearman's rank correlations revealed that with sunlight irradiation, fouling rates had significant positive connections with SMP proteins and the Flavitalea genus. The findings of this study demonstrate that sunlight is a considerable factor affecting membrane fouling and microbial ecology in MBRs, needing shade for fouling mitigation and sustainable operation.

High propensity of membrane fouling and the underlying mechanisms in a membrane bioreactor during occurrence of sludge bulking

While sludge bulking often occurring in activated sludge processes generally leads to serious membrane fouling in membrane bioreactors (MBR), the underlying causes are still unclear. In this study, fouling behaviors of a MBR operated at stages of normal and sludge bulking were compared, and the fouling mechanisms of the different behaviors were explored. It was found that, the MBR could be stably operated in normal stage without membrane cleaning for about 60 days, whereas, daily membrane cleaning had to be carried out when operated in sludge bulking stage. The bulking sludge possessed a rather high specific filtration resistance (SFR) of about 1.36×1014 m·kg−1, which is over 5.33 times than that of the normal sludge. A series of characterizations demonstrated that the bulking sludge had rather lower dewaterability, smaller particle size, higher fractal dimension, higher viscosity, abundant filamentous bacteria and different functional groups of extracellular polymer sustains (EPS). It was suggested that microbial community transition was responsible for the occurrence of sludge bulking, further affecting membrane fouling. Based on these characterizations, it was reported that adhesion propensity (indicated by the thermodynamic interaction) of the bulking sludge to the membrane surface is about 3.6 times than that of the normal sludge. It was proposed that, extra force should be provided to offset a chemical potential gap caused by foulant layer structure transition during sludge bulking in order to sustain filtration of the bulking sludge, resulting in extremely high SFR. This study offered deep thermodynamic mechanisms of MBR fouling during occurrence of sludge bulking.

The effect of ion environment changes on retention protein behavior during whey ultrafiltration process.

The factors affecting membrane fouling are very complex. In this study, the membrane fouling process was revealed from the perspective of ion environment changes, which affected the whey protein structure during ultrafiltration. It was found that the concentrations of Ca2+ and Na+ were overall increased and the concentrations of K+, Mg2+ and Zn2+ were decreased at an ultrafiltration time of 11min, which made more hydrophilic groups buried inside and increased the content of α-helix, leading to more protein aggregation. The relatively higher K+ ratio in retention could lead to an antiparallel β-sheet configuration, aspartic acid, glutamic acid and tryptophan increased, which resulted in more protein aggregation and deposition on the membrane surface at 17min. When the ion concentration and ratio restored the balance and were close to the initial state in retention, the protein surface tension decreased, and the hydrophilic ability increased at 21-24min.

Two strategies of stubborn biofouling strains surviving from NaClO membrane cleaning: EPS shielding and/or quorum sensing

The declined performance of repeated chemically-enhanced-backwashing (CEB) seriously hampered the sustainable operation of membrane bioreactor (MBR) in long-term, and could be partially attributed to the strengthened anti-cleaning properties of residual stubborn microbes. Although plenty of research has been done towards either the model strains or the whole post-CEB microbial community, little was known about the resisting behavior of practical stubborn strains when confronting oxidative stresses induced by NaClO. Hence, this study isolated 21 strains from samples in a large-scale MBR plant with routine CEB treatment. To unravel how they survive and affect membrane fouling, their anti-oxidation ability, fouling potential and quorum sensing (QS) effect before and after NaClO stimuli were evaluated. The composition and molecular weight distribution of extracellular polymeric substance (EPS) were also investigated to understand their roles during the anti-CEB process. It was found that typical stubborn strains tended to secrete more EPS as protective shields, where polysaccharides (especially the ones >1 kDa) made major contribution. However, sometimes EPS could not well resist the stimuli, with consequent low survival rate and high intracellular ROS level. Under such circumstances, stubborn strains would rather choose to be sensitive with surged QS level and quick population regrowth to maintain vitality under the oxidative stresses. Both strategies aggravated biofouling and eventually enhanced the anti-cleaning properties of biofilm.

Discrepant effects of monovalent cations on membrane fouling induced by colloidal polymer: Evaluation and mechanism investigation

Understanding how ionic conditions affect membrane fouling induced by anionic polyacrylamide (APAM) is important for achieving long-term and stable operation of a polymer flooding produced wastewater (PFPW) membrane separation process. However, there is lack of studies on the effects of monovalent cations (Na+ and K+) on APAM-based membrane fouling. In this work, the effects of Na+ and K+ on filtration efficiency, flux decline behavior, fouling resistance, and cleaning efficiency were studied through a series of microfiltration tests. Moreover, the influencing mechanism of membrane fouling was further comprehensively revealed from the aspects of the extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) theory, the hydration force, and the microstructure characterizations. The XDLVO theory agreed well with membrane fouling behavior at various ionic strengths. The increase in ionic strength (0–10,000 mg/L) of Na+ and K+ exacerbated the reduction of relative flux (J/J0) and the accumulation of fouling resistance, as well as made the porous APAM-induced fouling layer denser and more compact, boosting removal efficiency. Furthermore, K+ had a stronger aggravating effect on membrane fouling than Na+. Specifically, the final value of J/J0 for APAM+K+ (0.08) was lower than that for APAM + Na+ (0.12), and the fouling resistance for APAM+K+ (12.25 × 1011 m−1) was higher than that for APAM + Na+ (12.01 × 1011 m−1) at an ionic strength of 10,000 mg/L, which was owing to the larger hydration force caused by Na+ with a smaller ionic radius. This research offers practical guidance for the PFPW membrane filtering process.

Unravelling temperature-dependent fouling mechanism in a pilot-scale anaerobic membrane bioreactor via statistical modelling

Operational temperature is one of the most vital factors that profoundly affect membrane fouling during anaerobic membrane bioreactor operation. This study uncovered the temperature-dependent fouling mechanism in a pilot-scale anaerobic membrane bioreactor treating real municipal wastewater at ambient temperature seasonally fluctuating from 35 to 5 °C. It showed that compared to time accumulation effect of fouling layer, temperature played a more profound role on influencing membrane fouling. Low temperature caused severe membrane fouling by affecting the release of proteinous soluble microbial products (SMP) and extracellular polymeric substances (EPS) and the abundance of related microbial community structure, like Proteobacteria, Firmicutes, Bacteroidetes and Chloroflexi. The path analysis model established in this study, which was statistically significant at 5% level, revealed that temperature could substantially affect SMP and EPS, especially proteinous SMP (β = −0.94, p < 0.01), and thus indirectly influence total fouling rate, since SMP and EPS had significant influence on total fouling rate (p < 0.05). A temperature-based model (R2 > 0.95) was successfully developed in this study for predicting membrane fouling development over the seasons. The output of these two models varied with the type of membrane module. As calculated from the path coefficients, the influence of temperature on the total fouling rate of polymeric membrane (0.70) was greater than that on the total fouling rate of ceramic membrane (0.56). The regression coefficients of the formula obtained from the temperature-based model using the TMP results of polymeric membrane were around twice as large as those obtained from the model using the TMP results of ceramic membrane. These results suggested that ceramic membrane had better anti-fouling property, probably due to its chemical and thermal stability, hydrophilic surface, and well-organized pore structure, as confirmed by the resistance-in-series analysis and excitation-emission matrix (EEM) analysis of the foulant layer.

Distinct membrane fouling characteristics of anammox MBR with low NO2−-N/NH4+-N ratio

The bacterial growth and death, and extracellular polymeric substances (EPS) and soluble microbial products (SMP) in aerobic membrane bioreactor (MBR) cause severe membrane fouling. Anammox bacteria grow slowly but produce much EPS and SMP. Therefore, the membrane fouling characteristic of anammox MBR is still indistinct. A NO2−-N/NH4+-N < 1.0 into in the influent of an anammox MBR applies to investigate: 1) the slowest growing anammox bacteria (Candidatus Jettenia) could be enriched or not; 2) its membrane fouling characteristic. Results showed that Candidatus Jettenia successfully accumulated from 0.01% to 26.19%. The fouling characteristic of anammox MBR was entirely different from other MBRs. Firstly, obvious low transmembrane pressure (<4 KPa, 125 days) and low amount of foulants (0.22 gVSS/m2) might result from N2 production and the slow-growing Candidatus Jettenia. Secondly, the analysis of the components of membrane foulants indicated that polysaccharides of SMP in the gel layer and pore foulants were the key factors affecting membrane fouling. Finally, the large particle size of foulants (200 μm) might be caused by anammox bacteria living inside the foulants under anaerobic conditions. This study provides systematic insights into membrane characteristics of anammox MBR and a basis for the enrichment of anammox bacteria by MBR.

Fouling Behavior in a High-Rate Anaerobic Submerged Membrane Bioreactor (AnMBR) for Palm Oil Mill Effluent (POME) Treatment.

The characteristics of foulant in the cake layer and bulk suspended solids of a 10 L submerged anaerobic membrane bioreactor (AnMBR) used for treatment of palm oil mill effluent (POME) were investigated in this study. Three different organic loading rates (OLRs) were applied with prolonged sludge retention time throughout a long operation time (270 days). The organic foulant was characterized by biomass concentration and concentration of extracellular polymeric substances (EPS). The thicknesses of the cake layer and foulant were analyzed by confocal laser scanning microscopy and Fourier transform infrared spectroscopy. The membrane morphology and inorganic elements were analyzed by field emission scanning electron microscope coupled with energy dispersive X-ray spectrometer. Roughness of membrane was analyzed by atomic force microscopy. The results showed that the formation and accumulation of protein EPS in the cake layer was the key contributor to most of the fouling. The transmembrane pressure evolution showed that attachment, adsorption, and entrapment of protein EPS occurred in the membrane pores. In addition, the hydrophilic charge of proteins and polysaccharides influenced the adsorption mechanism. The composition of the feed (including hydroxyl group and fatty acid compounds) and microbial metabolic products (protein) significantly affected membrane fouling in the high-rate operation.

Application of machine learning algorithms in MBR simulation under big data platform

Abstract Membrane bioreactors (MBRs) are a sewage treatment process that combines membrane separation with bioreactor technology. It has great advantages in sewage treatment. Membrane fouling hinders MBR process development, however. Studies have shown that the degree of membrane fouling can be judged using the membrane flux rate. In this study, principal component analysis was used to extract the main factors affecting membrane fouling, then the random forest algorithm on the Hadoop big data platform was used to establish an MBR membrane flux prediction model, which was tested. In order to verify the model's effectiveness, BP neural network and SVM support vector machine models were established using the same experimental data. The experimental results from the different models were compared, and the results showed that the random forest algorithm gave the best MBR membrane flux predictions.

Recovery of cooling tower blowdown water through reverse osmosis (RO): review of water parameters affecting membrane fouling and pretreatment schemes

Recovery of cooling tower blowdown water through reverse osmosis (RO): review of water parameters affecting membrane fouling and pretreatment schemes

Box-Behnken response surface approach to identify factors affecting membrane fouling in a hybrid membrane bioreactor treating domestic sewage.

The effect of hydraulic retention time (HRT) and sludge retention time (SRT) on extracellular polymer substrate (EPS) content and resistance of a hybrid membrane bioreactor (HMBR) treating domestic sewage was analyzed by Box-Behnken response surface methodology. The quadratic response surface model demonstrated significant effects of both HRT and SRT on EPS content (both P value < 0.05), SRT on membrane resistance (P value = 0.0119), and their interaction was significant (P value = 0.0273) for EPS but not membrane resistance (P value = 0.0609). Model optimization indicates that the optimal conditions for the HMBR to control membrane fouling were an HRT of 10 h and SRT of 30 days. Under these optimal conditions, both the EPS content and the predicted membrane resistance closely matched the actual average value with the error about 8%. Thus, the feasibility of applying response surface methodology to an HMBR for treating domestic sewage was demonstrated. According to the detection result of the three-dimensional fluorescence (excitation-emission matrix), humic acid-like and fulvic acid-like substances gain much higher levels in the suspended carriers than those in the membrane and sludge, suggesting that these are key components of the membrane pollutants. Graphical abstract .

Optimization and organic fouling behavior of zwitterion-modified thin-film composite polyamide membrane for water reclamation: A comprehensive study

Membrane fouling can hinder the widespread application of thin film composite (TFC) reverse osmosis (RO) for water treatment. This study evaluated a novel zwitterion-grafted TFC RO as a mean to address organic fouling for water reclamation. The membrane exhibited the best permeability at the grafting condition of 45 °C in 1 h. This modified membrane consistently possessed improved antifouling ability irrespective of organic foulants. Among individual foulants, surfactant Dodecyl Trimethyl Ammonium Chloride (DTAC) posed the worst fouling potential due to its low molecular weight and positive charge, whereas fouling induced by other substances were relatively analogous and minor. In the mixture of DTAC and proteins, the former played a key role in governing the membrane fouling. While, their interplay affected membrane fouling, the fouling extent varied upon the membrane materials. The extended Derijaguin, Landau, Verwey and Overbeek (xDLVO) theory was unable to fully describe the interactions between surfactant foulants and the membrane materials. The complementary use of quartz crystal microbalance with dissipation (QCM-D), otherwise, concurred the fouling potential and gave the plausible interpretation for fouling mechanisms by providing insightful information of foulant layer on the polyamide-coated sensor. This study provided critical insights of organic foulants’ behavior on TFC RO membrane and offered the promising industrial implication of the novel membrane.

Quantitative analysis of cake characteristics based on SEM imaging during coagulation-ultrafiltration process.

Cake formed by flocs is a crucial factor to affect membrane fouling during coagulation-ultrafiltration process. To investigate the role of floc properties on cake, cake characteristics under various coagulant dosage conditions were calculated by scanning electron microscope (SEM) imaging. Results found that one SEM image with × 5000 magnification could accurately estimate cake porosity with relative error lower than 5.00% for all conditions, whereas more SEM images with × 10,000 magnification or × 20,000 magnification should be applied to calculate cake porosity precisely. This could be explained by different pore information of SEM images with various magnifications. Compared to single SEM image with × 10,000 magnification and × 20,000 magnification, single SEM image with × 5000 magnification contained the most comprehensive pore information and slightly overestimated pore area for pore smaller than 0.4 μm2 due to lower resolution. To verify feasibility by SEM image evaluating cake characteristics, cake porosity calculated by SEM image and Carman-Kozeny equation were analyzed. The results showed that cake porosity estimated by these two methods were nearly the same, proving the feasibility of this method. Moreover, with the increase of coagulant dosage, cake porosity presented similar variation with floc average size, indicating that floc average size was likely to dominate cake porosity in this study. For pore characteristics, pore average characteristic length and pore average area were in accordance with floc fractal dimension, whereas pore fractal dimension and pore amount were consistent with floc average size. This gives specific information about the relation between floc properties and cake characteristics.

Control Of Surface Fouling Of Membrane Modules In Membrane Bioreactors

The article considers the approaches in the field of membrane fouling control in membrane bioreactors used for wastewater treatment. It is highlighted that a significant number of research world-wide is devoted to the study of membrane fouling, which makes it possible to point out and identify the main factors affecting membrane fouling. Generally, such factors are the parameters of operation of the biological treatment reactor with a submerged membrane module, the properties of the biomass of activated sludge, as well as the characteristics of the membranes themselves, while the process of fouling is a three-stage mechanism. This article considers two approaches to control the process of the surface fouling of membrane modules: by improving the functioning of the aeration system and using floating biomass carriers. For aeration systems the concept of the minimum allowable aeration intensity is defined as a value which positively affects the reduction of fouling. The use of biomass carriers in addition to the function of biomass accumulation makes it possible to create additional shear forces on the surface of the membrane module, which contribute to more intensive removal of contaminants from the surface. It is established that the combination of several ways to control fouling is particularly effective, which has a positive effect on the operation of the treatment systems.

Mechanism of biofouling enhancement in a membrane bioreactor under constant trans-membrane pressure operation

In a membrane bioreactor, flux and organic loading rate conditions essentially affect membrane fouling. In constant TMP mode, flux and organic loading rate dynamically change, while they are rather static in constant flux mode. This study investigated fouling mechanism in MBR under these different filtration modes. Fouling development was more significant in constant TMP mode than in constant flux mode when MBR was operated under an equal flux condition in average. In constant TMP mode, microbial community in cake sludge was distinct from bulk sludge and had higher abundance of biofilm-forming bacterial group. The cake sludge contained more SMP and LB-EPS with high polysaccharide contents in constant TMP mode. These results suggested that, in constant TMP mode, (i) initially high flux in a filtration cycle enhanced initial deposition of bacterial cells and rapid development of fouling layer, which resulted in distinct microbial community in cake sludge; (ii) the high abundance of the biofilm-forming bacteria in cake sludge and fluctuation of organic loading rate stimulated polysaccharide production, (iii) higher polysaccharide contents in cake layer caused faster development of fouling layer and increase of hydraulic resistance by compression.