· · · Institutional Repository

This study showed that BF is an effective multiple objective barrier for removal of different contaminants present in surface water sources including bulk organic matter and organic micropollutants (OMPs) like pharmaceutically active compounds and endocrine disrupting compounds. It was found that biodegradation and adsorption play primary and secondary roles, respectively, in the removal of OMPs during soil passage. Furthermore, using field data from BF sites and chemical properties of OMPs, models were developed to estimate the removal of OMPs during soil passage.

Natural organic matter (NOM) generally influences water treatment processes such as coagulation, oxidation, adsorption, and membrane filtration. NOM contributes colour, taste and odour in drinking water, fouls membranes, serves as a precursor for disinfection by-products, increases the exhaustion and usage rate of activated carbon and may promote microbial growth in water distribution networks. High performance size exclusion chromatography and fluorescence excitation-emission matrices were used to characterize NOM relatively quickly and with minimal sample preparation. These and other tools were used to improve our understanding of NOM character and behaviour during drinking water treatment. The study demonstrates the potential of multiple NOM characterization tools for the selection, operation and monitoring of water treatment processes.

Seawater desalination is a rapidly growing coastal-based industry. The combined production capacity of all seawater desalination plants worldwide has increased by 30% over the last two years: from 28 million cubic meters per day in 2007—which is the equivalent of the average discharge of the River Seine at Paris—to more than 36 million cubic meters per day in 2009. Seawater desalination is an energy-intensive process. It furthermore consumes considerable amounts of natural resources in the form of chemicals and materials, and may have negative effects on the marine environment due to the discharges of concentrate waste waters and residual chemicals into the sea. The growing number of desalination plants worldwide and the increasing size of single facilities emphasizes the need for greener desalination technologies and more sustainable desalination projects. Two complementing approaches are the development and implementation of best available technology (BAT) standards and best practice guidelines for environmental impact assessment (EIA) studies. While BAT is a technology-based approach, which favors state of the art technologies that reduce resource consumption and waste emissions, EIA aims at minimizing impacts at a site- and project-specific level through environmental monitoring, evaluation of impacts, and mitigation where necessary. The dissertation contains a comprehensive evaluation and synthesis of the potential environmental impacts of desalination plants, with emphasis on the marine environment and aspects of energy use, followed by the development of strategies for impact mitigation. A concept for BAT for seawater desalination technologies is proposed, in combination with a methodological approach for the EIA of desalination projects. The scope of the EIA studies are outlined, including environmental monitoring, toxicity and hydrodynamic modeling studies, and the usefulness of multi-criteria analysis as a decision support tool for EIAs is explored and used to compare different intake and pretreatment options for seawater reverse osmosis plants.

Particulate/colloidal and organic fouling in seawater reverse osmosis (SWRO) systems results in flux decline, higher energy costs, increased salt passage, increased cleaning frequency, and use of chemicals. In practice, indices like SDI and MFI are used to assess particulate fouling, but they are performed at very high initial flux (> 1500 L/m2-h) and do not take into account the deposition of particles/colloids in RO systems. In this study, the Modified Fouling Index with ultrafiltration membranes (MFI-UF) at constant flux was further developed by incorporating the effects of particle/ colloidal deposition and flux. The percentage of particle deposition in real RO plants was between 10-30 % for Mediterranean seawater and between 80-90 % for North seawater. The biopolymer fraction (~60 %) and humic substances (~10-40 %) were found more likely to deposit on RO membranes in full scale desalination plants. A new portable set-up was developed capable of working with membranes of various pore sizes (10-100 kDa) and flux ranges between 10-350 L/m2-h. A model equation to predict particulate fouling was further developed to incorporate the effects of particle/ colloid deposition and flux. Employing the new improved model, the rate of particulate/ colloidal fouling potential of pre-treated seawater was found to be close to that of full scale desalination plants (between 0.2-1 bar/month), using a 10 kDa membrane at similar flux rate to a real RO system. The new developments presented in this study will enable engineers, plant operators and scientists not only to design better plants, but also to improve operation and monitoring of organic and particulate/colloidal fouling in SWRO systems.

Arsenic contamination of shallow tube well drinking water is an urgent health problem in Bangladesh. Current arsenic mitigation solutions, including (household) arsenic removal options, do not always provide a sustainable alternative for safe drinking water. A novel technology, Subsurface Arsenic Removal, relies on the existing technology of Subsurface Iron Removal. The principle of this technology is that aerated water is periodically injected into an anoxic or anaerobic aquifer through a tube well. The injection water partially displaces the original iron and arsenic containing groundwater. The oxygen-rich injection water oxidized adsorbed iron on the soil grains around the tube well. Once the flow direction is reversed, the oxidized iron (precipitated as iron (oxy)hydroxides) provides adsorption sites for soluble iron and arsenic. Subsequently groundwater with reduced iron and arsenic concentrations can be abstracted. This technology has the potential to be an affordable, robust and chemical-free arsenic removal solution for decentralized application. In this PhD study a combination field and laboratory research, in Bangladesh and the Netherlands, has resulted in better understanding of the subsurface processes determining the sustainable operation in diverse geochemical settings.

Ultrafiltration (UF) is a proven technology in water treatment nowadays. However, fouling remains a major challenge in the operation of UF, especially in regard to colloidal NOM fouling. In general, a number of colloidal NOM fouling mechanisms may occur, such as adsorption, gel formation. Colloidal NOM fouling is influenced by multivalent cations, ionic strength and pH. In order to control membrane fouling, different pretreatments such as powder activated carbon adsorption, lime softening, ion exchange, conventional media filtration and coagulation with inorganic and polymeric coagulant have been investigated. In-line coagulation is the most commonly used pretreatment for UF of surface water. However, the problem with in-line coagulation is that a large amount of backwash-derived waste sludge containing dosed coagulants is produced. Since the backwash waste sludge with coagulant has to be treated before its discharge (in especially Western Europe), this option creates additional cost for the membrane treatment plant (up to 20% of the total cost of the whole plant). This dissertation investigates the technical possibility of controlling the UF fouling by backwashing with low ionic strength water (demineralized water in the Netherlands), in order to reduce the ionic strength and the amount of multivalent cations and thus reduce NOM fouling. Chapter 1 briefly introduces this dissertation. The effectiveness of deminerlaized water backwashing is generally investigated on a pilot scale (with a 2.4m2 membrane) in Chapter 2. Results show that regarding the removal of NOM foulants via hydraulic backwashes, demineralized water is better than UF and NF permeate. That is probably due to the absence of cations, reducing the charge screening effect and/or Ca-bridging effect between the negatively charged membrane and NOM, leading to a restoration of repulsion force and consequently an easy removal of fouling layer. However, it is not clear which components in backwash water lead to the low foulant removal in this chapter. Therefore, Chapter 3 investigates the influence of backwash water composition on fouling control. Different amount of CaCl2 and NaCl was dosed in demineralized water to test their effect on fouling control. It became clear that the presence of monovalent and divalent cations in backwash water reduces the fouling control efficiency. Moreover, by isolating the organic matter in UF permeate for backwashing, it is found that the organic matter in UF permeate itself does not cause fouling problems when they are in backwash water. In terms of the influence of monovalent and divalent cations, both the elimination of the charge screening effect and the breakdown of the calcium bridging effect are possible mechanisms to explain this improvement. Therefore, these two effects are presumed to be the mechanisms of demineralized water backwashing. The investigation of the hypotheses of demineralized water backwashing is reported in Chapter 4, including the charge screening and the calcium bridging effects. By determining the zeta potential of the membranes and the colloidal NOM compounds at different conditions, the impact of pH and electrolyte valence and concentration on their charge was assessed. Furthermore, the adsorption of calcium on the membranes and the NOM compounds was also illustrated. Results showed that a membrane became less negatively charged when the pH decreased and the concentration of electrolyte increased, proving the presence of the charge screening effect. Furthermore, divalent cation has a much stronger effect on the increase of membrane zeta potential than monovalent cations which is generally in consistent with the DLVO theory. Calcium ions indeed adsorbed on either new or fouled membranes, and bridged NOM and membranes afterwards. However, the interaction of calcium with fouled membranes is more substantial than with new membranes. However, the charge screening effect played a dominant role in the membrane fouling and fouling control by demineralized water backwashing. Most of the fouling caused by calcium bridging is difficult to remove even with a demineralized water backwash. Chapter 5 illustrates the effectiveness of demineralized water backwashing on ultrafiltration fouling of different fractions of NOM. Results of natural waters show the same fouling removal via demineralized water backwashing as the previous chapters. Furthermore, LC-OCD analysis of Schie Canal water showed that biopolymers can be flushed away by hydraulic backwashes of either demineralized water or UF permeate. Compared to almost zero removal of humics and LMW substances by UF permeate backwashes, demineralized water backwashing was able to remove a substantial amount of humics, and a small amount of LMW substances. Fouling of sodium alginate model compound showed a high reversibility no matter what kind of backwash water was used. This is also consistent with the LC-OCD analysis of Schie Canal water. However, not all biopolymers were removed by hydraulic backwashes. A low fouling reversibility was observed for BSA fouling, but BSA may be in the part of unremoved biopolymers with demineralized water. No improvement in fouling control for fouling of Suwannee River humic acid (SRHA) was observed as well when demineralized water was used for the backwash. This is probably because the calcium bridging via carboxyl functional groups is the main mechanism for SRHA fouling, which is difficult to break down. Since the charge screening effect is the main mechanism of demineralized water backwashing, theoretically speaking, its application on seawater treatment is also possible. Chapter 6 demonstrates that demineralized water backwashing can substantially improve seawater UF fouling control, similar to the previous findings in surface canal water. However, the duration of a successful demineralized water backwash should be extended from one to two minutes. This is due to the high salinity of seawater and thus more demineralized water was required to dilute the seawater and limit a higher dispersion effect of seawater than surface water. Monovalent cations in backwash water showed their impact on the fouling control efficiency, indicating the existence of a charge screening effect. Furthermore, the different UF membrane fouling behaviors in winter and spring indicated the impact of a seasonal influence on UF membrane fouling. In spring, the membrane showed more fouling probably due to the algae bloom which is widely considered an important fouling factor. The results of the long-term experiment reconfirmed the effectiveness of backwashing with SWRO permeate (similar quality as demineralized water) on the fouling control of seawater UF. Since it is very easy to access SWRO permeate in a UF-RO desalination plant, this approach can be implemented easily. In order to apply this technique in industry, optimization work was conducted on a pilot scale with a standard membrane element (40 m2) and reported in Chapter 7. Results show that SWRO permeate (having similar qualities as demineralized water) backwashing substantially improved the seawater UF fouling control, consistent with the previous studies with small-scale membrane modules. The effectiveness of SWRO permeate backwashing on UF fouling control was observed at a recovery rate up to 95.8% during a low fouling period. Furthermore, the results of the DEMIFLUSH pilot were similar to the Evides desalination plant having the same operational settings, suggesting that the results obtained from the DEMIFLUSH pilot are applicable to full-scale plants. However, the results of high fouling period are missing. If the optimization results can be repeated in high fouling period as well, the application of SWRO permeate is also economically feasible due to the low consumption of SWRO permeate. Optimization work should be continued to reduce the consumption of SWRO permeate or demineralized water, since the usage of these water is expensive.

The topic of this thesis is a concept of zero liquid discharge for nanofiltration technology in drinking water treatment. Nanofiltration, defined as a process between Ultrafiltration (UF) and Reverse Osmosis (RO), is a rapidly emerging technology. The origin of NF membranes can be traced back to the late 1950s when it was developed to treat sea water. Now it is applied in drinking water treatment, wastewater treatment and oil separation and also in the food industry. It is even called the future process of 21st century. However, nanofiltration is also a controversial process because of the concentrate problem, especially for inland installations. In the Netherlands, this problem is more considerable due to the need of high quality of drinking water and the decreasing quality of the surface water. Concentrate contains high concentrations of dissolved organic and inorganic compounds in high concentration. The conventional approach of concentrate discharge to surface water becomes a growing problem due to the environmental guidelines from the authorities. The discharge of the concentrate is not the only problem. It is also a cost problem that 20% of the feed water is wasted. In the Netherlands, taxes are paid for extracted groundwater. Also the pretreatment is costly. So with every m3 discharged concentrate money is wasted. Therefore, there is a need for a technology by which the discharge of concentrate is not necessary. That is so-called Zero Liquid Discharge Technology. The primary problem to be solved with zero liquid discharge is the recovery. We expect through an innovative technology the nanofiltration membrane installation can be operated at very high recovery (99%) without increasing the treatment cost of drinking water. The cost for residuals treatment and disposal can be minimized because the amount of the concentrate is decreased about 20 times. A series of pretreatment processes is used for removing the scaling components from the feed water. The scaling components mainly include bivalent ions, silica and etc. Preventing fouling and scaling can guarantee a constant flux, reduce membrane area, lower chemical cleaning frequency, extend the lifespan of membrane and decrease energy consumption. In this research, a pilot experiment was performed with sludge softening, sedimentation, weak acid cation exchange and nanofiltration at Kiwa Water Research. By using this treatment process, the recovery can be handled successfully at 99% for at least 11 days. The pretreatment concept can remove the bivalent ions completely. Sludge softening is used to remove most of the bivalent ions, like calcium, magnesium and barium. The remaining bivalent ions can be effectively removed by weak acid cation exchange. In this way, the waste stream from the ion exchange is reduced. In theory, silica can be removed by sludge softening at high pH as the co-precipitation of of Mg(OH)2 and CaCO3. But in this experiment the removal efficiency of silica is low probably due to the shortage of magnesium in the feed water. After this treatment process, the remaining concentrate (1%) 3 is evaporated, only remaining salt which can be sold or discharged to a waste facility. In order to improve and guarantee the good performance of silica removal by sludge softening, a jar test is performed to define the influencing factor for silica removal. pH and the magnesium concentration can influence the silica removal efficiency. Higher magnesium concentration is necessary for silica removal. Also, a ternary ion exchange model is needed to predict the breakthrough of cation concentration in order to guarantee the good quality of the feed water for nanofiltration. A ternary system is more complex than binary system. This short report can not include all the aspects of this model. This is the first step to build a ternary model. In the first phase, the basic equation has been already found. The basic model concept has been built up, but needs to be checked and improved. And some batch experiments have been done to obtain model parameters like equilibrium constants and kinetic constants. Also two groups of column experiments have been done in order to measure breakthrough curves for Ca2+, Na+ and H+(pH). The experimental result is expected to be compared with the model result in the future. The pretreatment concept consisting of sludge softening at high pH (around 10), weak acid cation exchange in series can remove calcium, magnesium and barium completely. The calcium removal by the ion exchange is quite good. After pH breakthrough is reached, calcium still can be removed by the resin because it is exchanged with the sodium on the resin. With high magnesium concentration, silica concentration also can be reduced by the sludge softening to some extent. This combination of the treatment processes is possible to make the recovery of NF membrane reach 99% without scaling at least for 11 days. To put this innovative concept into practice needs lots of efforts on validation and testing. Stable operation of pilot experiment at 99% for a longer time is needed to check the feasibility of this concept. It is interesting and significant work. With increasing demand of drinking water, more and more people in this world needs this kind of technology to improve their living condition and environment. We expect the occurrence of this new big step of drinking water treatment technology.

Ion exchange resins have found an increasing application in the drinking water treatment sector over the last few decades. The ion exchange resins have a positive ability to remove the charged natural organic matter (NOM). To apply this process in full-scale treatment, the most suitable resin has to be selected and the hydraulic behaviour of the treatment process must be known. This study has the purpose to study the NOM fractions removal with different resins and select the most suitable resin for NOM removal of the Weesperkarspel water. Another purpose is to study the hydraulic behaviours of fluidized ion exchange and testing of the fluidization models of Ergun and Richardson-Zaki. The thesis study is characterizing NOM into the specific fraction and later observing the removal of each fraction with the different ion exchange resins. The kinetic and the equilibrium of NOM removal of each resin are also studied. Four ion exchange resins are tested (Lewatit VP OC 1071, Purolite Macronet 200, Duolite A7 and MIEX DOC). The ultraviolet absorbance method (UV), dissolved organic carbon detection method (DOC), specific ultraviolet absorbance (SUVA), fluorescence excitation emission matrix (fluorescence EEM) and liquid chromatography-organic carbon detector method (LC-OCD) were applied. Weesperkarspel water contains a high degree of aromatic NOM fractions, mostly in the form of humic substances. The Lewatit VP OC 1071 is the most suitable resin for removal of NOM in general view, especially aromatic NOM and humic substances fractions. It appears to be removed as high as 65% and 94% respectively. The MIEX DOC removed 57% aromatic NOM fraction and 70% of humic substances. The pH was found as the dominant parameter for the NOM removal by the weak base Duolite A7. In normal raw water (pH = 7.8), this resin is almost ineffective while it can remove the aromatic NOM up to 35% and humic substances fraction of 45% in raw water pH adjusted to 5. The sorbent Purolite Macronet 200 can remove only biopolymers and neutral NOM fractions. Adsorption is an important mechanism for the removal process of high molecular weight NOM fractions and the neutrals fractions. By using the linear driving force model (LDF) to describe the ion exchange rate of the resin, it was found that MIEX DOC resin can remove NOM faster than other resins. The LDF- k constant of MIEX DOC is also higher than other resins. MIEX DOC has the smallest resin bead size. This can be the reason for the fast removal of NOM. The resin exchange capacity is related with the Freundlich constant K and n . Increase of K and n values lead to increase of exchange capacity. The Lewatit VP OC 1071 has the highest K and n values and thus the highest exchange capacity. The exchange rate and exchange capacity is a specific property of each resin. With help from the LDF model and Freundlich isotherm the breakthrough curve can be determined. The Lewatit VP OC 1071 has longest running time compared with the MIEX DOC and Duolite A7 due to the highest exchange capacity. The hydraulics behaviour of fluidized bed ion exchange has been investigated with the strong base gel resin Lewatit VP OC 1071. The temperature and the feed velocity influence the expanded behaviour. The wet density and wet porosity are important parameters for the model prediction. Combination of mathematical modelling of ion exchange and the treatment conditions of Weesperkarspel drinking water treatment plant, the fluidized ion exchange process can be designed. For Weesperkarspel drinking water treatment plant, the aim is to decrease the DOC concentration of 7.2 mg C/l to 3.0 mg C/l with the fluidized ion exchange process. The 20 fluidized ion exchange reactors with a height of 9 m, 10 m2 of resin bed surface area, a bed height of 2 m and a feed velocity of 20 m/h are designed. The reactors consist of 4 groups, each group is started with delays of 15 days. With this operational step, the running time of each reactor is 60 days. The cost is estimated 11-euro cent per m3 treated water.