Dynamic membranes are considered to be an attractive anti-fouling remedy for membrane filtration, because once fouled, they can be removed and reformed in-situ, thereby prolonging the support membrane???s lifetime. However, large-scale application of dynamic membranes has been limited due to the numerous formation parameters that influence their properties. This thesis provides better understanding of the mechanisms of the dynamic membrane formation process through fundamental formation and characterisation studies of dynamically formed titanium dioxide membranes in laboratory scale dead-end and crossflow systems. The dynamic membranes exhibited water fluxes ranging from 30-1147 L/m2h and dextran (500 kDa) rejections as high as 99.9%. Of the six formation parameters studied, the pH and constant flux conditions had the greatest influence on dynamic membrane properties. The pH affects dynamic membrane properties by changing particle aggregation prior to dynamic membrane formation, while constant flux conditions affect the drag force on particles during deposition thereby altering cake compressibility. The advantage of using the novel concept of constant flux formation over traditional constant pressure formation is that it enables greater control of particle deposition during dynamic membrane formation. Dextran rejection data also suggested the existence of a critical mass loading, above which dynamic membrane flux and rejection properties are reduced. This thesis also demonstrated the utility of a factorial design experiment for preliminary identification and evaluation of the critical factors affecting dynamic membrane formation, a method which could be invaluable for tailoring dynamic membranes for use in specific applications. In addition, cake removal data suggested that more than 80% of the dried cake could be removed providing a high potential for membrane regeneration. For the formation conditions studied, it was concluded that convection was the dominant mechanism governing particle transport during dynamic membrane formation. The fluxes and cake properties of the dynamic membranes were best described by the resistance-in-series model for simple dead-end microfiltration. Furthermore, the higher cake void fraction required to fit the experimental data (at low formation pressure or constant flux conditions) with model predictions suggested that the ratio of shear to convection was an important mechanistic parameter determining dynamic membrane properties.

Dynamic membranes are considered to be an attractive anti-fouling remedy for membrane filtration, because once fouled, they can be removed and reformed in-situ, thereby prolonging the support membrane???s lifetime. However, large-scale application of dynamic membranes has been limited due to the numerous formation parameters that influence their properties. This thesis provides better understanding of the mechanisms of the dynamic membrane formation process through fundamental formation and characterisation studies of dynamically formed titanium dioxide membranes in laboratory scale dead-end and crossflow systems. The dynamic membranes exhibited water fluxes ranging from 30-1147 L/m2h and dextran (500 kDa) rejections as high as 99.9%. Of the six formation parameters studied, the pH and constant flux conditions had the greatest influence on dynamic membrane properties. The pH affects dynamic membrane properties by changing particle aggregation prior to dynamic membrane formation, while constant flux conditions affect the drag force on particles during deposition thereby altering cake compressibility. The advantage of using the novel concept of constant flux formation over traditional constant pressure formation is that it enables greater control of particle deposition during dynamic membrane formation. Dextran rejection data also suggested the existence of a critical mass loading, above which dynamic membrane flux and rejection properties are reduced. This thesis also demonstrated the utility of a factorial design experiment for preliminary identification and evaluation of the critical factors affecting dynamic membrane formation, a method which could be invaluable for tailoring dynamic membranes for use in specific applications. In addition, cake removal data suggested that more than 80% of the dried cake could be removed providing a high potential for membrane regeneration. For the formation conditions studied, it was concluded that convection was the dominant mechanism governing particle transport during dynamic membrane formation. The fluxes and cake properties of the dynamic membranes were best described by the resistance-in-series model for simple dead-end microfiltration. Furthermore, the higher cake void fraction required to fit the experimental data (at low formation pressure or constant flux conditions) with model predictions suggested that the ratio of shear to convection was an important mechanistic parameter determining dynamic membrane properties.