Membrane fouling is a complicated issue in microfiltration and ultrafiltration. Clearly identifying the dominant fouling mechanisms during the filtration process is of great significance for the phased and targeted control of fouling. To this end, we propose a semi-empirical multiple linear regression model to describe flux decline, incorporating the five fouling mechanisms (the first and second kinds of standard blocking, complete blocking, intermediate blocking, and cake filtration) based on the additivity of the permeate volume contributed by different coexisting mechanisms. A piecewise fitting protocol was established to distinguish the fouling stages and find the significant mechanisms in each stage. This approach was applied to a case study of a microfiltration membrane filtering a model foulant solution composed of polysaccharide, protein, and humic substances, and the model fitting unequivocally revealed that the dominant fouling mechanism evolved in the sequence of initial adaptation, fast adsorption followed by slow adsorption inside the membrane pores, and the gradual growth of a cake/gel layer on the membrane surface. The results were in good agreement with the permeate properties (total organic carbon, ultraviolet absorbance, and fluorescence) during the filtration process. This modeling approach proves to be simple and reliable for identifying the main fouling mechanisms during membrane filtration with statistical confidence.
Membrane fouling is still the bottleneck affecting the technical and economic performance of the ultrafiltration (UF) process for the surface water treatment. It is very important to accurately understand fouling mechanisms to effectively prevent and control UF fouling. The rejection performance and fouling mechanisms of the UF membrane for raw and coagulated surface water treatment were investigated under the cycle operation of constant-pressure dead-end filtration and backwash. There was no significant difference in the UF permeate quality of raw and coagulated surface water. Coagulation mainly removed substances causing turbidity in raw surface water (including most suspended particles and a few organic colloids) and thus mitigated UF fouling effectively. Backwash showed limited fouling removal. For the UF process of both raw and coagulated surface water, the fittings using single models showed good linearity for multiple models mainly due to statistical illusions, while the fittings using combined models showed that only the combined complete blocking and cake layer model fitted well. The quantitative calculations showed that complete blocking was the main reason causing flux decline. Membrane fouling mechanism analysis based on combined models could provide theoretical supports to prevent and control UF fouling for surface water treatment.
It is of great importance to quantitatively characterize feed fouling potential for the effective and efficient prevention and control of reverse osmosis membrane fouling. A gradient filtration method with microfiltration (MF 0.45 μm) → ultrafiltration (UF 100 kDa) → nanofiltration (NF 300 Da) was proposed to extract the cake layer fouling index, I, of different feed foulants in this study. MF, UF, and NF showed high rejection of model suspended solids (kaolin), colloids (sodium alginate and bovine serum albumin), and dissolved organic matters (humic acid) during constant-pressure individual filtration tests, where the cake layer was the dominant fouling mechanism, with I showing a good linear positive correlation with the foulant concentration. MF → UF → NF gradient filtration tests of synthetic wastewater (i.e., model mixture) showed that combined models were more effective than single models to analyze membrane fouling mechanisms. For each membrane of gradient filtration, I showed a positive correlation with the targeted foulant concentration. Therefore, a quantitative assessment method based on MF → UF → NF gradient filtration, the correlation of combined fouling models, and the calculation of I would be useful for characterizing the fouling potentials of different foulants. This method was further successfully applied for characterizing the fouling potential of real wastewater (i.e., sludge supernatant from a membrane bioreactor treating dyeing and finishing wastewater).