Les procédés membranaires se sont progressivement démocratisés dans les usines de traitement des eaux. Cependant, leur développement est limité par le colmatage. En présence de mélange de colloïdes organiques et de particules minérales, les mécanismes de colmatage deviennent complexes et impliquent de nombreuses interactions. Ainsi, l'objectif de ces travaux a été de mieux comprendre les mécanismes mis en jeu lors de la filtration de suspensions composées de particules minérales et de matières organiques dissoutes et colloïdales. Des extractions et caractérisation de matières organiques ont été réalisées sur une eau de rivière et un effluent secondaire. Les fractions colloïdales (> 3,5 kDa) et dissoutes ont été filtrées sur des membranes d'ultrafiltration en absence et en présence de particules minérales (i.e. argiles). En raison de leur masse moléculaire élevée, les colloïdes organiques ont engendré le plus fort pouvoir colmatant quelle que soit leur origine et leur composition chimique. Par la suite, la filtration de suspensions mixtes constituées de particules minérales et de matières organiques a montré des comportements très différents selon les mélanges étudiés. L'élucidation des mécanismes mis en jeu a permis de développer un modèle phénoménologique permettant de décrire les différentes chutes de flux observées en incluant des mécanismes de blocage de pores et de tamisage du dépôt de particules. / Membrane processes have been gradually implemented in water treatment plants. However, their development is still limited by fouling. The mechanisms of fouling generated by mixtures of organic colloids and mineral particles are more complex and involve interactions between mineral and organic constituents. Thus, the objective of this work was to better understand the mechanisms involved in the filtration of suspensions composed of mineral particles and dissolved and colloidal organic matter. The organic matters from river water and a secondary effluent were first extracted and characterized. The dissolved and colloidal fractions (> 3.5 kDa) were filtered on ultrafiltration membranes in the absence and presence of mineral particles (i.e. clay). Due to their high molecular weight, the organic colloids were the main contributor to membrane fouling whatever their origin and chemical composition. Then, the filtration of suspensions constituted of both mineral particles and organic matter showed very different behaviors depending on the nature of each constituent. A phenomenological model has been developed to describe the observed flux decrease. The model includes pore blocking mechanisms associated to a screening effect of mineral particles cake layer.

[pt] A queda do fluxo permeado através da membrana com o tempo de operação causada pelo fenômeno de colmatação é o principal fator limitante do uso de processos de separação por membranas, como a nanofiltração (NF). O sulfato de cálcio é um dos principais agentes do tipo mais severo de colmatação, a incrustação, em sistemas de dessulfatação de águas marinhas para injeção em poços produtores de petróleos e abrandamento de águas subterrâneas salobras e duras para uso industrial por NF. Apesar do vasto número de estudos que têm sido publicados sobre a incrustação por sulfato de cálcio em NF, ainda não foram esclarecidos os mecanismos de formação e inibição química do fenômeno. O objetivo do presente trabalho foi estudar experimentalmente e conceitualmente os mecanismos envolvidos na geração e no tratamento por inibição e remoção química da incrustação de sulfato de cálcio em membrana de NF. Para tal, foram utilizados modelos matemáticos de interpretação de mecanismos e técnicas de inibição e caracterização dos depósitos formados, correlacionando-os com indicadores de desempenho do processo. As análises de comportamento de queda de fluxo mostraram uma alternância na predominância entre os mecanismos de bloqueio de poros e por formação de torta. O estudo de inibição mostrou que os mecanismos de ação do antiincrustante SHMP o tornaram mais eficiente que o EDTA na manutenção do desempenho em permeabilidade e seletividade. A autópsia da membrana com MEV/EDS e DRX revelou majoritariamente cristais de gipsita. A metodologia de cálculo e análise integrada dos parâmetros de desempenho mostraram-se ferramentas de grande importância para se buscar as medidas mais eficientes e convenientes de controle e redução da incrustação. / [en] Membrane fouling is a major limitation in efficient operation of nanofiltration (NF) plants. Calcium sulfate is a major player in the most severe type of fouling – scaling - in sulfate removal systems for oil fields seawater injection operations and brackish water softener for industrial use by NF. Despite the vast number of studies that have been published on the scaling by calcium sulfate NF, have not yet been elucidated the mechanisms of formation and chemical inhibition of the phenomenon. The aim of this work was to study experimentally and conceptually the mechanisms involved in the generation and treatment by inhibition and chemical removal of fouling of calcium sulfate in NF membrane. Mathematical models of interpretation techniques and mechanisms of inhibition and characterization of deposits formed were used, trying to correlate them with performance indicators of process. The supersaturated solutions in a laboratory scale system confirmed the severity of fouling by calcium sulfate. The analysis of behavior showed a decrease of flow in alternating dominance between the mechanisms of pore blocking and cake formation. The study showed that inhibition of the mechanisms of action of the anti-fouling SHMP become more efficient than EDTA in maintaining performance in permeability and selectivity. The autopsy of the membrane with SEM / EDS and XRD revealed mostly gypsum crystals. The calculation methodology and integrated analysis of performance parameters proved to be an important tool to seek the most efficient and convenient measures to control and reduce fouling.

This diploma thesis discusses fouling processes and mechanisms of industrial heat exchangers, with focus on shell side fouling of heat transfer area by flue gases, and eventually taking fouling into account, or more precisely its modeling and predicting by calculation. Thermal-hydraulic design of heat exchanger is focused mainly on shell side, as flue gases are dominant process medium in terms of fouling and heat transfer. Based on available operational data of evaluated industrial heat exchanger, specific process has been developed to determine by calculation, or more precisely predict change of fouling factor throughout operational period of heat exchanger. Usage and various options of this specific calculation technique are being discussed in conclusion, whether there is a potential to use this technique regarding more complex operational inputs or prediction of fouling process for different operational regime of heat exchanger.

In diesem Beitrag wird exemplarisch anhand des Verschmutzungsmechanismus von Wärmetauschern (Fouling) gezeigt wie datengetriebene Methoden zur Vorhersage des Verschmutzungsgrades und somit zu einem effizienteren Anlagenbetrieb genutzt werden können. Hierzu werden zu Beginn neben der wirtschaftlichen Bedeutung des Foulings die strömungsphysikalischen und thermodynamischen Hintergründe vorgestellt. Danach wird der konkrete Anwendungsfall mit den zur Verfügung stehenden Daten aufgezeigt. Aufgrund der stark limitierten Datenlage wird mit dem Verfahren der segmentierten Regression ein relativ einfacher jedoch robuster Ansatz für ein Vorhersagemodell erarbeitet und diskutiert. / This contribution uses the fouling mechanism of heat exchangers (fouling) as an example to show how data-driven methods can be used to predict the degree of fouling and thus achieve more efficient plant operation. In addition to the economic significance of fouling, the flow-physical and thermodynamic background is presented at the beginning. Then the specific application case with the available data is shown. Due to the very limited data available, a relatively simple but robust approach for a prediction model is developed and discussed using the segmented regression method.

Particles and macromolecular components, including biopolymers (protein and carbohydrate), are viewed as the main foulants in the complex feed submerged membrane filtration systems such as membrane bioreactor (MBR). This work focused on two aspects of fouling in complex fluids: 1- Assessing fouling propensity and mechanisms for various model solutions. 2- Using of two specific solutions modelling biomass found in MBR for a better understanding of the fouling mechanisms in submerged MBR processes. Filtrations were carried out with 0.22 ??m PVDF hollow fibre membrane. Alginate was used as a model for polysaccharide, bovine serum albumin (BSA) as a model for protein, (un)washed yeast and bentonite were representing suspended solid contents. According to the data obtained during this study the fouling propensity of each model solution was classified as follow in a decreasing order: Alginate &gt unwashed yeast &gt washed yeast &gt BSA &gt bentonite for one-component solutions; and Alginate-washed yeast &gt Alginate-BSA &gt Alginate-bentonite &gt Alginate-unwashed yeast for two-component solutions. Introducing the alginate increased the reversible fouling (except BSA). Passive adsorption had a significant effect on fouling of alginate even before the beginning of the filtration. Washed yeast and a mixture of washed yeast + BSA were then used as model solutions to simulate the activated sludge found in MBR. The concentration of washed yeast and BSA used in this study were calculated in order for the characterisations of the two model solution to match (in terms of biopolymer contents) those of MBR biomasses reported in the literature. By rinsing, backwashing and chemical cleaning of the membrane, three fouling layers of upper, intermediate and lower were defined respectively. Results obtained from the analysis of the biopolymers found in the cleaning solutions allow a better understanding of the fouling mechanisms occurring for the two model solutions used in this study: for washed yeast, the lower layer and for washed yeast + BSA , the upper and intermediate layers were found to have relatively high biopolymeric composition. This was explained by higher concentration of solids on the membrane surface and by higher biopolymer interactions when washed yeast was mixed with BSA.

The market for polyamide desalination membranes is expected to continue to grow during the coming decades. Purification of alternative water sources will also be necessary to meet growing water demands. Purification of produced water, a byproduct of oil and gas production, is of interest due to its dual potential to provide water for beneficial use as well as to reduce wastewater disposal costs. However, current polyamide membranes are prone to fouling, which decreases water flux and shortens membrane lifetime. This research explored surface modification using poly(ethylene glycol) diglycidyl ether (PEGDE) to improve the fouling resistance of commercial polyamide membranes. Characterization of commercial polyamide membrane performance was a necessary first step before undertaking surface modification studies. Membrane performance was found to be sensitive to crossflow testing conditions. Concentration polarization and feed pH strongly influenced NaCl rejection, and the use of continuous feed filtration led to higher water flux and lower NaCl rejection than was observed for similar tests performed using unfiltered feed. Two commercial polyamide membranes, including one reverse osmosis and one nanofiltration membrane, were modified by grafting PEGDE to their surfaces. Two different PEG molecular weights (200 and 1000) and treatment concentrations (1% (w/w) and 15% (w/w)) were studied. Water flux decreased and NaCl rejection increased with PEGDE graft density ([microgram]/cm2), although the largest changes were observed for low PEGDE graft densities. Surface properties including hydrophilicity, roughness and charge were minimally affected by surface modification. The fouling resistance of modified and unmodified membranes was compared in crossflow filtration studies using model foulant solutions consisting of either a charged surfactant or an oil in water emulsion containing n-decane and a charged surfactant. Several PEGDE-modified membranes demonstrated improved fouling resistance compared to unmodified membranes of similar initial water flux, possibly due to steric hindrance imparted by the PEG chains. Fouling resistance was higher for membranes modified with higher molecular weight PEG. Fouling was more extensive for feeds containing the cationic surfactant, potentially due to electrostatic attraction with the negatively charged membranes. However, fouling was also observed in the presence of the anionic surfactant, indicating hydrodynamic forces are also responsible for fouling. / text

The objective of this research was to develop fluorescence-based tools that are suitable for performing rapid, accurate and direct characterization of natural organic matter (NOM) and colloidal/particulate substances present in natural water. Most available characterization methods are neither suitable for characterizing all the major NOM fractions such as protein-, humic acid-, fulvic acid- and polysaccharide-like substances as well as colloidal/particulate matter present in natural water nor are they suitable for rapid analyses. The individual and combined contributions of these NOM fractions and colloidal/particulate matter present in natural water contribute to membrane fouling, disinfection by-products formation and undesirable biological growth in drinking water treatment processes and distribution systems. The novel techniques developed in this research therefore, provide an avenue for improved understanding of these negative effects and proactive implementation of control and/or optimization strategies. The fluorescence excitation-emission matrix (EEM) method was used for characterization of NOM and colloidal/particulate matter present in water. Unlike most NOM and colloidal/particulate matter characterization techniques, this method can provide fast and consistent analyses with high instrumental sensitivity. The feasibility of using this method for monitoring NOM at very low concentration levels was also demonstrated with an emphasis on optimizing the instrument parameters necessary to obtain reproducible fluorescence signals. Partial least squares regression (PLS) was used to develop calibration models by correlating the fluorescence EEM intensities of water samples that contained surrogate NOM fractions with their corresponding dissolved organic carbon (DOC) concentrations. These fluorescence-based calibration models were found to be suitable for identifying/monitoring the extent of the relative changes that occur in different NOM fractions and the interactions between polysaccharide- and protein-like NOM in water treatment processes and distribution systems. Principal component analysis (PCA) of fluorescence EEMs was identified as a viable tool for monitoring the performance of biological filtration as a pre-treatment step, as well as ultrafiltration (UF) and nanofiltration (NF) membrane systems. The principal components (PCs) extracted in this approach were related to the major membrane foulant groups such as humic substances (HS), protein-like and colloidal/particulate matter in natural water. The PC score plots generated using the fluorescence EEMs obtained after just one hour of UF or NF operation could be related to high fouling events likely caused by elevated levels of colloidal/particulate-like material in the biofilter effluents. This fluorescence EEM-based PCA approach was sensitive enough to be used at low organic carbon levels present in NF permeate and has potential as an early detection method to identify high fouling events, allowing appropriate operational countermeasures to be taken. This fluorescence EEM-based PCA approach was also used to extract information relevant to reversible and irreversible membrane fouling behaviour in a bench-scale flat sheet cross flow UF process consisting of cycles of permeation and back-washing. PC score-based analysis revealed that colloidal/particulate matter mostly contributed to reversible fouling, while HS and protein-like matter were largely responsible for irreversible fouling. This method therefore has potential for monitoring modes of membrane fouling in drinking water treatment applications. The above approach was further improved by utilizing the evolution of the PC scores over the filtration time and relating these to membrane fouling by the use of PC scores??? balanced-based differential equations. Using these equations the proposed fluorescence-based modeling approach was capable of forecasting UF fouling behaviours with good accuracy based solely on fluorescence data obtained at time = 15 min from the initiation of the filtration process. In addition, this approach was tested experimentally as a basis for optimization by modifying the UF back-washing times with the objective of minimizing energy consumption and maximizing water production. Preliminary optimization results demonstrated the potential of this approach to reduce power consumption by significant percentages. This approach was also useful for identifying the fouling components of the NOM that were contributing to reversible and irreversible membrane fouling. Grand River water (Southwestern Ontario, Canada) was used as the natural water source for developing the techniques presented in this thesis. Future research focusing on testing these methods for monitoring of membrane fouling and treatment processes in large-scale drinking water treatment facilities that experience different sources of raw water would be useful for identifying the limitation of these techniques and areas for improvements.

Particles and macromolecular components, including biopolymers (protein and carbohydrate), are viewed as the main foulants in the complex feed submerged membrane filtration systems such as membrane bioreactor (MBR). This work focused on two aspects of fouling in complex fluids: 1- Assessing fouling propensity and mechanisms for various model solutions. 2- Using of two specific solutions modelling biomass found in MBR for a better understanding of the fouling mechanisms in submerged MBR processes. Filtrations were carried out with 0.22 ??m PVDF hollow fibre membrane. Alginate was used as a model for polysaccharide, bovine serum albumin (BSA) as a model for protein, (un)washed yeast and bentonite were representing suspended solid contents. According to the data obtained during this study the fouling propensity of each model solution was classified as follow in a decreasing order: Alginate &gt unwashed yeast &gt washed yeast &gt BSA &gt bentonite for one-component solutions; and Alginate-washed yeast &gt Alginate-BSA &gt Alginate-bentonite &gt Alginate-unwashed yeast for two-component solutions. Introducing the alginate increased the reversible fouling (except BSA). Passive adsorption had a significant effect on fouling of alginate even before the beginning of the filtration. Washed yeast and a mixture of washed yeast + BSA were then used as model solutions to simulate the activated sludge found in MBR. The concentration of washed yeast and BSA used in this study were calculated in order for the characterisations of the two model solution to match (in terms of biopolymer contents) those of MBR biomasses reported in the literature. By rinsing, backwashing and chemical cleaning of the membrane, three fouling layers of upper, intermediate and lower were defined respectively. Results obtained from the analysis of the biopolymers found in the cleaning solutions allow a better understanding of the fouling mechanisms occurring for the two model solutions used in this study: for washed yeast, the lower layer and for washed yeast + BSA , the upper and intermediate layers were found to have relatively high biopolymeric composition. This was explained by higher concentration of solids on the membrane surface and by higher biopolymer interactions when washed yeast was mixed with BSA.

L’amélioration de la récupération d’énergie dans les procédés industriels passe par une meilleure compréhension des phénomènes d’encrassement dans les échangeurs de chaleur. L’encrassement se caractérise par la formation de dépôts non désirés sur les surfaces d’échange de l’échangeur. Cette étude porte sur l’amélioration de l’efficacité énergétique des raffineries de pétrole par réduction de l’encrassement dans les échangeurs de chaleur du train de préchauffe.Pour cela, une boucle d’essai reproduisant les niveaux d’échange et les écoulements rencontrés en raffinerie a été employée pour tester un échangeur de type tubes et calandre. Les fluides traités sont du pétrole brut et de résidu de distillation atmosphérique (coupe lourde du pétrole) comme dans le procédé industriel. On s’intéresse à l’influence de la géométrie de l’échangeur. Des faisceaux de tubes avec des corrugations internes hélicoïdales et des ailettes externes sont successivement testés. Pour chacune des technologies, les paramètres opératoires optimums permettant de limiter la formation d’un dépôt encrassant sont identifiés sur une gamme de température de film et de cisaillement s’échelonnant respectivement de 230 à 300°C et de 1,3 à 8,8 N/m2. L’efficacité des tubes optimisés pour lutter contre l’encrassement est évaluée par comparaison avec l’encrassement obtenu sur le faisceau de tubes lisses pris comme référence. Les deux technologies se sont révélées efficaces pour réduire l’encrassement. Par rapport aux tubes lisses, l’encrassement a été réduit d’un ordre de grandeur à cisaillement et température de film équivalentes aussi bien avec les tubes structurés qu’ailetés.En parallèle, une simulation d’encrassement numérique (CFD) a été élaborée pour mieux comprendre le développement de l’encrassement dans l’échangeur équipé de tubes lisses. Les phénomènes dominants à l’origine de l’encrassement de l’échangeur ont été déterminés : la vitesse de formation du dépôt, peu affectée par les variations spatiales de la thermo-hydraulique, pilote l’encrassement. Les hétérogénéités d’encrassement sont causées par les disparités locales de la vitesse d’arrachement du dépôt qui varie en grande proportion mais dont l’influence sur la vitesse d’encrassement est faible. Elles sont dues à une inhomogénéité spatiale du cisaillement qui est la conséquence de l’établissement de l’écoulement dans les tubes. Une tentative d’amélioration de la précision de la prédiction de l’encrassement a été également entreprise en optimisant les paramètres d’un modèle d’encrassement existants et en tenant compte du vieillissement du dépôt. / Improving energy recovery in industrial processes requires a better understanding of fouling phenomena in heat exchangers. Fouling is the grow up of unwanted materials on heat transfer surfaces. This study focuses on improving the energy efficiency of petroleum refineries by reducing fouling in heat exchangers of the pre-heat train.For this purpose, a test rig reproducing thermal and flow characteristics encountered in the last heat exchanger of the pre-heat train was used to test a pilot-scale shell-and-tube heat exchanger. Fluids are the same as the ones processed in refineries: crude oil and atmospheric tower bottom, a heavy residue of oil. This study focuses on the influence of the heat exchanger geometry. Internally helically-finned tubes and externally low-finned tube bundles are successively tested. For each one, optimum operating parameters which reduce fouling are identified over a 230-300°C film temperature range and a 1.3-8.8 N/m2 wall shear stress range. The fouling mitigation efficiency of the enhanced tubes is compared with smooth tubes taken as a reference. Compared to smooth tubes, fouling on helically-finned and low-finned tubes is reduced by an order of magnitude when they are operated at equivalent wall shear stress and film temperature.In addition, a numerical (CFD) fouling simulation has been developed to provide a better understanding of the fouling in the heat exchanger with smooth tubes. Dominant phenomena driving fouling in the heat exchanger were determined. The deposition rate is weakly impacted by the spatial variations of the thermo-hydraulic and controls the overall fouling rate. Heterogeneities in fouling rate are caused by the local scattering in the removal rate, which varies in a broad range but whose impact on the overall fouling rate is low. The broad range of the shear stress is the consequence of the fluid flow entrance effects. An attempt to improve accuracy of the fouling model has been undertaken by optimizing the model parameters and by considering aging of the deposit.

Komati Power Station has installed a membrane plant consisting of ultrafiltration, double pass reverse osmosis and continuous electro-deionisation to treat cooling tower blowdowns in order to produce demineralised water and to conduct sidestream chemistry control of the cooling water circuit. This plant has replaced the existing ion-exchange plant that was used for the production of demineralised water and thus serves to reduce the loading of mobile salts in the ash dam (90% reduction) by eliminating regeneration effluent from the ion-exchange plant. Due to oil contamination in the cooling water circuit (when oil from oil coolers leaks into the cooling water), the membrane plant was also designed to operate on raw water from either the Nooigdedacht or the Vygeboom Dam or a blend of both dams. This is considered to be an emergency intervention under abnormal conditions to prevent possible irreversible fouling of the membranes due to oil in the cooling water. The Nooigtedach Dam water contains high concentrations of organic matter and is also enriched with nutrients due to raw sewage influent into the Dam water. This poses a challenge with regard to treatment of the high fouling feed water on the membrane plant. Natural organic matter in water has the ability to foul reverse osmosis membranes. This adversely affects the operation of the reverse osmosis process. However, very little information is available regarding the fouling characteristics of natural organic material in the raw and cooling water at Komati Power Station for the reverse osmosis membranes. Therefore, a pilot study was undertaken to determine the influence of natural organic matter on membrane fouling, to optimise the process for the removal of natural organic matter and to assess the ability of two different reverse osmosis membranes to effectively treat the high fouling feed water at Komati Power Station. The ability of a polyethersulphone hollow-fibre ultrafiltration membrane system was first evaluated to remove natural organic matter in the feedwater, by conducting pilot tests, initially without coagulation of the raw water and thereafter with in-line coagulation for organics removal. Jar tests were conducted in the laboratory to determine the most suitable coagulant and dosage for turbidity and natural organic matter removal. Various coagulants were tested and, based on the results of the jar tests, a coagulant (U3000) was identified based on optimal removal of both total organic carbon and turbidity at a dosing level of 20 mg/L. During the operation of the ultrafiltration pilot plant, permeate flow; feed pressure and feed temperature were monitored. Performance of the ultrafiltration membrane was monitored in terms of flux versus time for operation with and without a coagulation process. The results indicated that there was very little total organic carbon removal (maximum removal of 4%) without coagulation and a slight decrease in flux. The flux declined as a result of fouling but could be recovered by performing hydraulic backwashes and CEB procedures. Permeate flux, however, could be maintained at about 90 Lmh (from 642 hours of operation). Since most of the organics passed through the ultrafiltration membrane, it was concluded that the loss in flux was due to colloidal fouling of the membrane. This was observed when the operation was carried out using raw water as feed as well as when cooling water was used. The total organic carbon removal increased to 30% when the plant was operated with inline coagulation. The flux remained relatively stable during the first 600 hours of operation and only decreased significantly during the last 200 hours of operation as a result of fouling. The reduction in flux prior to cleaning was less than the 15% (maximum flux decline of 9.9% during the test period) which is acceptable according to the industry norm of 15%. It appeared that flux could be maintained at around 90 Lmh which was about the same as when no coagulant was applied. The 30% total organic carbon reduction that was obtained was not sufficient to reduce the organics to the level of 6mg/L dissolved organic carbon that was specified by the membrane manufacturer for the standard brackish water reverse osmosis membrane. Two reverse osmosis membranes – the standard brackish water reverse osmosis membrane (BW30-2540) and the extra-low-fouling membrane (BW30XFR-2540) – were assessed in terms of their ability to remove dissolved organic carbon, ease of cleaning of the membrane and the ability to recover flux after cleaning. This was done to establish which membrane is more suited to Komati’s high-fouling feedwater. The evaluation of the performance of the two reverse osmosis membranes was conducted using pre-treated water (filtered water after in-line coagulation, anti-scalant and biocide dosing) as well as using water that was not pre-treated. During operation (under both conditions), the normalised permeate flux, conductivity, dissolved organic carbon and organics absorbing at UV254 were monitored. It was established that in terms of flux decline that the extra low-fouling membrane gave slightly superior performance to that of the standard membrane, achieving longer production runs (up to 5 days compared with 3 days achieved by the standard brackish water membrane) without requiring chemical cleaning. The low fouling membrane achieved better CWF recovery after the cleaning cycles (81.26% Lmh of the virgin membrane on the occasions when there was flux loss) compared to the standard membrane (restored to 77.35% of CWF of the virgin membrane) when using untreated feed water. This performance improved when pre-treated feed water was used and the low fouling membrane’s CWF regained after the CIP was 95.89% which was within the industry norm of a flux recovery of 95%, indicating that the CIP had been effective. It was determined that the TOC rejection of the low-fouling membrane was higher (average TOC rejection of 97%, maximum TOC rejection of 99%) than that of the standard membrane (average TOC rejection of 95.3%, maximum TOC rejection of 97%). Preliminary efforts to optimize the pre-treatment for organics removal in order to reduce organic loading for the RO membranes confirmed that the use of granular activated carbon and use of an organic scavenger resin might not be economically feasible due to the relatively quick TOC breakthrough (8910BV, approximately 18000BV and less than 18000BV for the Filtrasorb 300, Filtrasorb 400 and organic scavenger resin, respectively). Although further investigations should still be conducted, the preliminary results indicate that it would be beneficial to also identify other options that can be further investigated for optimization of organics removal at Komati Power Station. Decline in the normalised flux as well as the evidence of biofouling were witnessed during the pilot operation suggesting that the membranes were fouled. Autopsies were performed on both membranes to identify foulants responsible for the decline in flux that was observed during the pilot study. The results did not indicate an organic foulant on the membrane surface. Biofouling should however, be monitored in the main plant as this was suspected to have resulted in the flux decline during the pilot study. The low fouling membrane demonstrated a better capability to treat the Komati raw and cooling water and would be expected to achieve lower operating costs for the plant (CIP costs and membrane replacement costs) while achieving better organics removal and it is therefore recommended that the low-fouling membranes be used at Komati Power Station as they are superior to the standard membrane and the cost of the low-fouling membranes is comparable to that of the standard membrane. While this would provide somewhat better performance than that obtained with the standard brackish water membranes, it is proposed that further investigation into pre-treatment optimization for organics removal as well as more efficient cleaning solutions be investigated to improve the performance and economics of the main water treatment plant at Komati power Station. / Dissertation (MSc)--University of Pretoria, 2013. / gm2014 / Chemical Engineering / unrestricted