This thesis investigates the fouling of a microfiltration membrane by biological solutions. Membrane fouling is recognized as a major drawback for the application of microfiltration in the purification of biotechnology products. Membrane fouling was analyzed and compared for filtrations performed with a hollow fiber microfiltration module operated at constant permeate flux using bovine serum albumin (BSA) solutions or Chinese hamster ovary (CHO) cell culture broths as feed solutions. A mechanistic model was developed to represent the fouling of a cross-flow microfiltration membrane operated at constant permeate flux. Fouling was observed as an increase in the transmembrane pressure (TMP) and assumed to occur first by pore blockage followed by cake formation over the blocked pores. The effect of the cross-flow action was described by the removal of deposits from the membrane surface thereby reducing the pore blockage and the mass of the cake. The model was fitted to the TMP profiles obtained during the filtration of BSA solutions and Chinese hamster ovary (CHO) cell culture broths with a 0.45 µm polysulfone hollow fiber membrane. According to the fitted model, pores blocked faster and more cake was formed with increasing BSA concentration. In the case of CHO cell culture broth, increasing the wall shear rate (proportional to cross-flow velocity) seemed to lead to the formation of a less pronounced cake but more pore blockage. The cross-flow mechanistic model was applied to the microfiltration of CHO cell culture supernatants harvested at different days of the fermentation process. The filtrations were performed at two different shear rates and with two different membrane pore sizes. The cell culture supernatant caused membrane fouling observed as an increase in both the TMP and the membrane hydraulic resistance estimated from water flux measurements at the end of the filtrations. The highest TMP increase was observed for the filtrations with the smaller membrane pore size (0.2 µm) and the higher shear rate (8000 s-1). The hydraulic resistance estimates of the fouled membrane also revealed a higher irreversible fouling for the smaller (0.2 µm) membrane pore size while the model analysis indicates that more fouling occurred at the entrance of the membrane pores. The shear rate was found to strongly influence the contribution of the reversible fouling to the total hydraulic resistance of the membrane. The cross-flow mechanistic model indicates a higher pore blockage for the most severe membrane fouling observed experimentally for the smaller membrane pore size and the higher shear rate. At the same time, a smaller cake deposit was predicted for the higher shear rate. The different cell culture harvest time investigated in this study did not reveal any differences in membrane fouling. BSA solutions were used to evaluate the contribution of BSA aggregates, fresh cell culture medium and the non-ionic surfactant Pluronic F-68 to membrane fouling. A simple empirical model was developed to represent the TMP increase and to derive the initial fouling rate. The initial fouling rate, the normalized TMP and the irreversible membrane resistance at the end of the filtration were analyzed to determine the effect of BSA solution characteristics on membrane fouling. The initial fouling rate increased with increasing BSA aggregate content consistent with a two-step fouling mechanism that was proposed for membrane fouling by BSA. Increased BSA concentration and the use of fresh cell culture medium compared to potassium phosphate buffer resulted in an increase in initial fouling rate, TMP and irreversible membrane fouling. The addition of the non-ionic surfactant Pluronic-F68 to the BSA solutions decreased the long-term fouling and the irreversible fouling but did not affect the initial fouling rate.

This thesis investigates the fouling of a microfiltration membrane by biological solutions. Membrane fouling is recognized as a major drawback for the application of microfiltration in the purification of biotechnology products. Membrane fouling was analyzed and compared for filtrations performed with a hollow fiber microfiltration module operated at constant permeate flux using bovine serum albumin (BSA) solutions or Chinese hamster ovary (CHO) cell culture broths as feed solutions. A mechanistic model was developed to represent the fouling of a cross-flow microfiltration membrane operated at constant permeate flux. Fouling was observed as an increase in the transmembrane pressure (TMP) and assumed to occur first by pore blockage followed by cake formation over the blocked pores. The effect of the cross-flow action was described by the removal of deposits from the membrane surface thereby reducing the pore blockage and the mass of the cake. The model was fitted to the TMP profiles obtained during the filtration of BSA solutions and Chinese hamster ovary (CHO) cell culture broths with a 0.45 µm polysulfone hollow fiber membrane. According to the fitted model, pores blocked faster and more cake was formed with increasing BSA concentration. In the case of CHO cell culture broth, increasing the wall shear rate (proportional to cross-flow velocity) seemed to lead to the formation of a less pronounced cake but more pore blockage. The cross-flow mechanistic model was applied to the microfiltration of CHO cell culture supernatants harvested at different days of the fermentation process. The filtrations were performed at two different shear rates and with two different membrane pore sizes. The cell culture supernatant caused membrane fouling observed as an increase in both the TMP and the membrane hydraulic resistance estimated from water flux measurements at the end of the filtrations. The highest TMP increase was observed for the filtrations with the smaller membrane pore size (0.2 µm) and the higher shear rate (8000 s-1). The hydraulic resistance estimates of the fouled membrane also revealed a higher irreversible fouling for the smaller (0.2 µm) membrane pore size while the model analysis indicates that more fouling occurred at the entrance of the membrane pores. The shear rate was found to strongly influence the contribution of the reversible fouling to the total hydraulic resistance of the membrane. The cross-flow mechanistic model indicates a higher pore blockage for the most severe membrane fouling observed experimentally for the smaller membrane pore size and the higher shear rate. At the same time, a smaller cake deposit was predicted for the higher shear rate. The different cell culture harvest time investigated in this study did not reveal any differences in membrane fouling. BSA solutions were used to evaluate the contribution of BSA aggregates, fresh cell culture medium and the non-ionic surfactant Pluronic F-68 to membrane fouling. A simple empirical model was developed to represent the TMP increase and to derive the initial fouling rate. The initial fouling rate, the normalized TMP and the irreversible membrane resistance at the end of the filtration were analyzed to determine the effect of BSA solution characteristics on membrane fouling. The initial fouling rate increased with increasing BSA aggregate content consistent with a two-step fouling mechanism that was proposed for membrane fouling by BSA. Increased BSA concentration and the use of fresh cell culture medium compared to potassium phosphate buffer resulted in an increase in initial fouling rate, TMP and irreversible membrane fouling. The addition of the non-ionic surfactant Pluronic-F68 to the BSA solutions decreased the long-term fouling and the irreversible fouling but did not affect the initial fouling rate.

No description available.

The specific research work described in this thesis forms part of a much larger research project that was funded by the Technology Programme of the UK Government. This larger project considered improving the design and efficiency of membrane bioreactor (MBR) plant by using modelling, simulation and laboratory methods. This research work uses phenomenological mechanistic models based on MBR filtration and biochemical processes to measure the effectiveness of alternative behavioural models based upon input-output system identification methods. Both model types are calibrated and validated using similar plant layouts and data sets derived for this purpose. Results prove that although both approaches have their advantages, they also have specific disadvantages as well. In conclusion, the MBR plant designer and/or operator who wishes to use good quality, calibrated models to gain a better understanding of their process, should carefully consider which model type is selected based upon on what their initial modelling objectives are (e.g. using either a physically mechanistic model or an input-output behaviourial model). Each situation usually proves unique. In this regard, this research work creates a "Model Conceptualisation Procedure" for a typical MBR which can be used by future researchers as a theoretical framework which underpins any newly created model type. There has been insufficient work completed to date on using a times series input-output approach in the model development of a wastewater treatment plant, so only general conclusions can be made from this research work. However, it can be stated that this novel approach seems to be applicable for a membrane filtration model if care it taken to select appropriate input-output model structures, such as those suggested in the "Model Conceptualisation Procedure". In the case of the development of a MBR biological model, it is thought that a conventional Activated Sludge model produced by the IWA could be coupled to a input-output model structure as suggested by this report to give a hybrid model structure that may have the advantages of both model types. Further research work is needed in this area. Future work that should follow on from this research study should focus on whether these input-output models could be used for predictive control purposes, whether an integrated model could be created, and whether a benchmark could be created for the three main MBR configurations.

Membrane filtrations are widely used in process industries but are almost always limited by fouling, a highly studied and significant problem. This is defined as unwanted material deposited on a membrane surface or within its pores, which can significantly impair performance and/or reduce operating life. The strategies to mitigate fouling include cleaning in place, modified membranes, and optimisation of operating conditions. In order to correctly select or target improvements to any such measures a detailed mechanistic understanding of the fouling process is important, which requires more than just performance data from unit operations. One key mechanism is that of cake fouling, which describes the build-up of particle layers on the surface of the membrane. Its growth and physical properties are difficult to assess. In this project the technique of fluid dynamic gauging (FDG) has been explored as a means to study cake fouling. This simple, yet robust method allows for estimation of thickness and strength of cake fouling at high concentrations and opacity, without any prerequisite knowledge of feed properties. Studies described herein focused on microfiltrations through cellulosic membranes. FDG was used to track cake growth during filtrations of polydisperse yeast suspensions (which contained large agglomerates), demonstrating its capability to work with non-ideal, food-like substances. Later studies used more predictable suspensions of hollow glass spheres, which were used to assess various filtration models. The most effective was found to be an interpretation of the critical flux laws, which were used to successfully identify pore fouling during filtrations of Kraft lignin, an observation supported by FDG measurements. Another novel achievement of this project was the development of an automated apparatus for performing FDG in cross-flow membrane filtration. This allowed for much faster acquisition of results, and demonstrated the potential for its development into an autonomous system capable of making thickness measurements on the fly during filtrations. The most reliable protocol for determining cake growth rates was by repeated filtrations in which destructive thickness testing was performed at selected time points. This was because continuous or even repeated thickness measurements during a single filtration were found to cause too much disturbance to the fouling layer. Computational fluid dynamics was used to simulate shear stress profiles on the fouling layer, while also providing a more accurate means to calibrate the automated apparatus. Erosion caused by FDG readings, when viewed under a microscope, was found to conform to the shear stress profiles predicted by simulations.

The membrane bioreactor (MBR) is becoming increasingly popular for wastewater treatment, mainly due to its capability of producing high quality effluent with a relatively small footprint. However, high plant maintenance and operating costs due to membrane fouling limit the wide spread application of MBRs. Membrane fouling generally depends on the interactions between the membrane and, the activated sludge mixed liquor, which in turn, are affected by the chosen operating conditions. The present research study aimed to explore the process performance and membrane fouling in the membrane enhanced biological phosphorus removal (MEBPR) process under different operating conditions by, (1) comparing two MEBPRs operated in parallel, one with constant inflow and another with a variable inflow, and by, (2) operating the MEBPRs with different solids retention times (SRT). On-line filtration experiments were conducted simultaneously in both MEBPR systems by using test membrane modules. From the transmembrane pressure (TMP) data of the test membrane modules, it was revealed that fouling propensities of the MEBPR mixed liquors were similar in both parallel reactors under the operating conditions applied, although the fouling propensity of the aerobic mixed liquors of both reactors increased when the SRT of the reactors was reduced. Routinely monitored reactor performance data suggest that an MEBPR process with a varying inflow (dynamic operating condition) performs similarly to an MEBPR process with steady operating conditions at SRTs of 10 days and 20 days. Mixed liquor characterization tests were conducted, including critical flux, capillary suction time (CST), time to filter (TTF) and, bound and soluble extracellular polymeric substances (EPS) were quantified, to evaluate their role on membrane fouling. The tests results suggest that the inflow variation in an MEBPR process did not make a significant difference in any of the measured parameters. With decreased SRT, an increase in the concentrations of EPS was observed, especially the bound protein, and the bound and soluble humic-like substances. This suggests that these components of activated sludge mixed liquors may be related to membrane fouling. No clear relationship was observed between membrane fouling and other measured parameters, including critical flux, normalized CST and normalized TTF.

In spite of abundant water resources, world is suffering from the scarcity of usable water. Seawater Reverse Osmosis (SWRO) desalination technology using polymeric membranes has been recognized as a key solution to water scarcity problem. However, economic sustainability of this advanced technology is adversely impacted by the membrane fouling problem. Fouling of RO membranes is a highly studied phenomenon. However, literature is found to be lacking a detailed study on kinetic and dynamic aspects of SWRO membrane fouling. The factors that impact the fouling dynamics, i.e., pretreatment and water quality were also not adequately studied at full–scale of operation. Our experimental protocol was designed to systematically explore these fouling aspects with the objective to improve the understanding of SWRO membrane fouling mechanisms. An approach with multiple analytical techniques was developed for fouling characterization. In addition to the fouling layer characterization, feed water quality was also analysed to assess its fouling potential. Study of SWRO membrane fouling dynamics and kinetics revealed variations in relative abundance of chemical and microbial constituents of the fouling layer, over operating time. Aromatic substances, most likely humic–like substances, were observed at relatively high abundance in the initial fouling layer, followed by progressive increase in relative abundances of proteins and polysaccharides. Microbial population grown on all membranes was dominated by specific groups/species belonging to different classes of Proteobacteria phylum; however, similar to abiotic foulant, their relative abundance also changed with the biofilm age and with the position of membrane element in RO vessel. Our results demonstrated that source water quality can significantly impact the RO membrane fouling scenarios. Moreover, the major role of chlorination in the SWRO membrane fouling was highlighted. It was found that intermittent mode of chlorination is better than continuous mode of chlorination of seawater, as anti–biofouling strategy. It was also confirmed that significant biofilm development was inevitable even with the use of chlorine to disinfect SWRO membranes. Our findings on the dynamic patterns of SWRO membrane fouling should help in further elaborating research projects focusing on the development of better strategies to minimize this troublesome phenomenon.

The membrane bioreactor (MBR) is becoming increasingly popular for wastewater treatment, mainly due to its capability of producing high quality effluent with a relatively small footprint. However, high plant maintenance and operating costs due to membrane fouling limit the wide spread application of MBRs. Membrane fouling generally depends on the interactions between the membrane and, the activated sludge mixed liquor, which in turn, are affected by the chosen operating conditions. The present research study aimed to explore the process performance and membrane fouling in the membrane enhanced biological phosphorus removal (MEBPR) process under different operating conditions by, (1) comparing two MEBPRs operated in parallel, one with constant inflow and another with a variable inflow, and by, (2) operating the MEBPRs with different solids retention times (SRT). On-line filtration experiments were conducted simultaneously in both MEBPR systems by using test membrane modules. From the transmembrane pressure (TMP) data of the test membrane modules, it was revealed that fouling propensities of the MEBPR mixed liquors were similar in both parallel reactors under the operating conditions applied, although the fouling propensity of the aerobic mixed liquors of both reactors increased when the SRT of the reactors was reduced. Routinely monitored reactor performance data suggest that an MEBPR process with a varying inflow (dynamic operating condition) performs similarly to an MEBPR process with steady operating conditions at SRTs of 10 days and 20 days. Mixed liquor characterization tests were conducted, including critical flux, capillary suction time (CST), time to filter (TTF) and, bound and soluble extracellular polymeric substances (EPS) were quantified, to evaluate their role on membrane fouling. The tests results suggest that the inflow variation in an MEBPR process did not make a significant difference in any of the measured parameters. With decreased SRT, an increase in the concentrations of EPS was observed, especially the bound protein, and the bound and soluble humic-like substances. This suggests that these components of activated sludge mixed liquors may be related to membrane fouling. No clear relationship was observed between membrane fouling and other measured parameters, including critical flux, normalized CST and normalized TTF.

The membrane bioreactor (MBR) is becoming increasingly popular for wastewater treatment, mainly due to its capability of producing high quality effluent with a relatively small footprint. However, high plant maintenance and operating costs due to membrane fouling limit the wide spread application of MBRs. Membrane fouling generally depends on the interactions between the membrane and, the activated sludge mixed liquor, which in turn, are affected by the chosen operating conditions. The present research study aimed to explore the process performance and membrane fouling in the membrane enhanced biological phosphorus removal (MEBPR) process under different operating conditions by, (1) comparing two MEBPRs operated in parallel, one with constant inflow and another with a variable inflow, and by, (2) operating the MEBPRs with different solids retention times (SRT). On-line filtration experiments were conducted simultaneously in both MEBPR systems by using test membrane modules. From the transmembrane pressure (TMP) data of the test membrane modules, it was revealed that fouling propensities of the MEBPR mixed liquors were similar in both parallel reactors under the operating conditions applied, although the fouling propensity of the aerobic mixed liquors of both reactors increased when the SRT of the reactors was reduced. Routinely monitored reactor performance data suggest that an MEBPR process with a varying inflow (dynamic operating condition) performs similarly to an MEBPR process with steady operating conditions at SRTs of 10 days and 20 days. Mixed liquor characterization tests were conducted, including critical flux, capillary suction time (CST), time to filter (TTF) and, bound and soluble extracellular polymeric substances (EPS) were quantified, to evaluate their role on membrane fouling. The tests results suggest that the inflow variation in an MEBPR process did not make a significant difference in any of the measured parameters. With decreased SRT, an increase in the concentrations of EPS was observed, especially the bound protein, and the bound and soluble humic-like substances. This suggests that these components of activated sludge mixed liquors may be related to membrane fouling. No clear relationship was observed between membrane fouling and other measured parameters, including critical flux, normalized CST and normalized TTF. / Applied Science, Faculty of / Civil Engineering, Department of / Graduate

To minimize membrane fouling many water treatment plants pre-treat water prior to microfiltration (MF) or ultrafiltration (UF). Coagulation/flocculation/sedimentation is a common form of pre-treatment, but little research has been conducted on floatation as a part of the pre-treatment. The objective of this thesis is to compare pre-treatment with floatation and with sedimentation for Ottawa River water, a typical Northern Canadian water with a high natural organic matter (NOM) content and a large hydrophobic (HPO) NOM fraction. Fouling tests consisted of multiple filtration/backwashing cycles performed by an automated bench-scale UF hollow fiber membrane system. Test were conducted with Ottawa River water (ORW) and ORW subjected to three different types of pre-treatment conducted at closely-located full-scale water treatment plants, including one using floatation. Both Alum pre-treatments resulted in decreases in NOM (63% and 68% TOC) and HPO NOM (56% and 68%TOC) which helped to reduce fouling. However, the remaining NOM and HPO NOM still caused significant hydraulically and chemical irreversible fouling. The water pre-treated with floatation produced the least severe hydraulically irreversible fouling for all experiments while Raw ORW produced the highest. During the early stages of membrane filtration (~10 hours), the TMP sharply increases which may imply that adsorption is dominant. Statistical analysis during the initial stages of filtration showed that the HPO fraction of NOM was linked to hydraulically irreversible fouling, which may be attributed to adsorption. Raw ORW also had the highest hydraulically reversible fouling while all pre-treatments were able to reduce this type of fouling. Statistical analysis suggested that the transphilic (TPI) fraction of NOM and particulate organic carbon (POC) were responsible for hydraulically reversible fouling during subcritical flux experiments, which may be attributed to cake formation on the membrane surface. It was found that for all waters and experiments, hydraulically irreversible fouling was greater than hydraulically reversible fouling. This may be because of the high HPO concentrations in the ORW. Hydraulically reversible fouling and backwash efficiencies were found to fluctuate with time. It is hypothesised that the cake formation adheres to the membrane surface and is not fully removed until enough backwash pressure has developed. Further investigation into alternative cleaning procedures is required as the NaOH cleaning was not very effective for some of the pre-treated waters.