1.1 BACKGROUND
The demand for high-quality drinking water is growing dramatically throughout the world, particularly with a rise in urbanisation and population growth. However, contamination of surface water resources through the discharge of municipal and industrial wastewaters necessitates intensive industrial treatment (Ray et al., 2011). Schwarzenbach et al. (2006) reported that chemical contamination of surface water is a serious environmental problem facing humanity. In the last few decades, a growing variety of environmental contaminants have been detected at elevated concentrations in freshwater resources including organic compounds such as humic acids, compounds used in personal care products, pesticides, insecticides, pharmaceuticals, and synthetic chemicals; and inorganic chemicals such as nitrogen, phosphorus, and metals. This presents numerous challenges for drinking water treatment systems, such as odour, colour, and taste, as well as raising the required dosage of chemicals for coagulation and disinfection processes (Matilainen et al., 2010). Most developing countries employ conventional water-treatment technologies that are no longer considered viable for effectively disinfecting polluted water and eliminating or reducing contaminants to the required levels (Maeng, 2010).
Arid and semi-arid societies face even more severe water management challenges due to the scarcity of water resources (Sophocleous, 1997, 2002), Indeed, many countries with an arid climate struggle to supply good quality drinking water at a low economic cost. The hydrological conditions in arid climates can be extreme and highly variable (Sophocleous, 2000). High temperatures primarily influence the effectiveness of conventional treatment processes, such as adsorption, coagulation and disinfection and thereby, the quality of drinking water provided (Sugiyama et al., 2013). Moreover, this increases the required chemical dosages for disinfection processes (LeChevallier, 2004). Geriesh et al. (2008) suggested that pre-treatment involving filtering surface water and reducing organic content would increase the quality of drinking water supplied in arid climates. Therefore, natural treatment systems, such as bank filtration (BF), offer potentially viable options for water supply schemes in arid and semi-arid areas. These systems involve treatment and/or pre-treatment steps to remove pathogens, algal toxins and organic matter (OM), and reduce turbidity and chemical pollutants in the drinking water (Hiscock et al., 2002).
1.2 BANK FILTRATION
BF is regarded as a simple and sustainable technique that can provide good-quality drinking water. After the Second World War, European surface water resources became
heavily contaminated with industrial and municipal waste water, and BF was the only means considered able to secure drinking water of acceptable quality (Shamrukh et al., 2011). BF is a process in which surface water undergoes a subsurface flow caused by the lowering of the hydraulic head prior to abstraction from vertical or horizontal wells (Grischek et al., 2003). The raw water extracted from the production well consists of a mixture of infiltrated surface water and ambient groundwater. It has been shown that under suitable hydrogeological conditions, well-operated BF facilities may provide relatively low-cost, high-quality drinking water that requires little further treatment (Tufenkji et al., 2002). Alluvial aquifers are the most suitable sites given their high production capacity, high connectivity to surface water sources, and accessibility to regions of demand (Doussan et al., 1997).
BF can improve water quality effectively by reducing turbidity, microbial contaminants, microcystins, pathogens, heavy metals (HMs), OM, and inorganic pollutants (Gandy et al., 2007; Hiscock et al., 2002; Sontheimer, 1980). BF has a high capability to eliminate such soluble contaminants that are difficult to remove in surface water treatment plants. For example, BF has been shown to reduce dissolved organic matter (DOM) and disinfected by-product (DBP) precursors by 50% (Ray et al., 2002). Attenuation of pollutants relies on mechanisms such as biodegradation, adsorption, precipitation, and filtration. The effectiveness of this approach depends on a variety of factors, including aquifer geology, aquifer structure, surface water flow, surface and groundwater OM type, river bed composition and clogging processes as well as land use in the local catchment region (Hiscock et al., 2002).
BF has long been used as a multi-objective natural treatment technology that eliminates much of the surface water contamination. BF also equilibrates temperature and dampens accidental chemical load peaks. It can be used to replace or support existing water treatment techniques by providing a robust barrier and reducing the cost of treatment. BF also helps reduce the use of chemical disinfectants to produce biologically-stable water (Sharma et al., 2009). Another advantage of BF is that it may be used in regions with seasonally variable precipitation and run-off regimes (e.g., monsoon-, flood-, and drought-prone regions) as a means of increasing water-storage capacity (HĂŒlshoff et al., 2009). Moreover, mixing bank filtrate with local groundwater increases the groundwater supply and dilutes contaminants (Grischek et al., 2003).
1.3 EXPERIENCE OF BANK FILTRATION
BF has been utilised by several water supply companies in Europe and North America for the production of drinking water. In Germany, BF is primarily used around the Rhine River at the Lower Rhine in the region between the Sieg and Ruhr tributaries, the Elbe River between Dresden and Torgau, and the Berlin district. Its application also covers the Rhine and Meuse Rivers in the Netherlands, as well as in many other European countries including Austria, Switzerland, and France. BF provides 50% of Franceâs drinking water supply, 15% in Germany, and 12% in the Netherlands (Deyi, 2012; Hiscock et al., 2002). On the Rhine River at Dusseldorf, Germany, BF has been used widely since 1870 to provide high-quality drinking water. In 1960, the quality of the river Rhine started to deteriorate and anaerobic conditions developed in the infiltration zone. Iron and manganese reduction rates also increased. As a result, there was a need for post-treatment of river bank-filtrate. In the last few decades, the water quality of the raw water has improved; however, periodic changes in river water quality and hydraulics due to climatic conditions are on-going issues (Eckert et al., 2006). Currently, granular activated carbon is used in conjunction with ozonation and filtration to further treat bank filtrate and eliminate chemical contaminants.
Horizontal collector wells have been used in the United States for more than 80 years to pump bank filtrate. Due to the high production capacity, horizontal wells were mainly used to supply water for industrial uses, although approximately one-third of the horizontal wells that were initially produced were used for drinking purposes. In recent years, this technique has been implemented to produce drinking water in vast amounts and to ensure water quality standards are met (Hunt, 2003). BF has proved effective in the removal of chemical and biological contaminants including organic micropollutants (OMPs), Giardia and Cryptosporidium parasites, and microcystins, which are not adequately removed by conventional treatment techniques (Ray et al., 2002). In North America, BF is widely used as a pre-treatment system to enhance the quality of raw water and reduce the cost of treatment (Wang, 2003).
The use of BF has expanded in developing countries in recent years, including Kenya, Malawi, Bosnia, Russia, Egypt, India, Korea, and China (Bartak et al., 2014; Bosuben, 2007; Chaweza, 2006; Dash et al., 2008; Ray et al., 2011). Most of the BF wells established in these countries are vertical wells, which are mostly recharged from the local surface water system and are not designed as BF wells (Shamrukh et al., 2011). Shamrukh et al. (2008) illustrated that mixing bank filtrate with highly-polluted groundwater, and the dissolution of iron and manganese along the infiltration path, are the main problems currently restricting the wider use of this technique in developing countries.
1.4 BANK FILTRATE QUALITY
The effectiveness of BF in the production of high-quality drinking water is dependent on a multiple of variables, including raw water quality, hydrological characteristics, and geological setting. Hydrological characteristics have substantial effects on the travel time and redox conditions of the infiltration zone, which have direct influences on BF efficiency and pumped water quality (Ray et al., 2002). This section outlines the main factors impacting the quality of bank filtrate.
1.4.1 Raw water quality
The concentration of pollutants in the raw water system is one of the main parameters affecting the performance of BF and the need for post-treatment. Schijven et al. (2003) reported that if the concentration of Cryptosporidium is greater than 0.075 oocysts/L in raw water, further treatment would be needed to provide safe drinking water. Kedziorek et al. (2009) stated that if the electron trapping capacity (ETC) (calculated from the summation of the dissolved O2 and NO3- concentrations) of the infiltrate water is greater than 0.2 mmol/L, the concentration of manganese in the abstracted bank filtrate would be very low (< 0.1 ”m) unless the ambient groundwater has a higher concentration of this contaminant. The quality characteristics of raw water are influenced by hydrological and climatic conditions. Surface water systems with low flow velocities and high nutrient concentrations have a higher potential for the formation of algal blooms. In the same regard, climate has a significant impact on redox processes taking place in surface water bodies. Furthermore, dissolution of metals from the bank and bottom sediments is high under arid conditions. For example, Zwolsman et al. (2008) observed that the concentrations of HMs (cadmium, chromium, mercury, lead, copper, nickel, and zinc) in the Rhine River were higher during the 2003 drought.
1.4.2 Travel time
Travel time has a significant effect on the efficiency of BF and should, therefore, be taken into consideration throughout the design phase (Sprenger et al., 2011). Long travel times provide more opportunity and time for sorption and biodegradation, which are essential for the elimination of chemical and biological pollutants. However, it may also have an adverse effect by enabling the development of anaerobic conditions, which can increase the dissolved metal load in the pumped water. The travel times required in the BF systems is mainly determined by the persistence of pathogens. It was suggested, for example, that travel times of 60 days in Germany and 70 days in the Netherlands are adequate to ensure biologically stable drinking water (Azadpour-Keeley, 2003; Hiscock et al., 2002). The proposed travel time to remove 90% of OMPs is > 6 months (Drewes et al., 2003). In North America, BF is often used/considered as a pre-treatment technique for conventional drinking water treatment plants, for which the travel times range from hours to a few weeks at most (GrĂŒnheid et al., 2005).
Various methods have been proposed to calculate travel times. Wang et al. (2008) used bromide and tritium to determine travel times and the influence of soil type and land cover on recharge rates along Hebei Plain (China). Atrazine has also been used to estimate the travel time and river discharge at the banks of the Platte River (Nebraska, United States) (Duncan et al., 1991; Verstraeten et al., 1999). Fluctuations in electrical conductivity were used to determine the travel time of infiltrated water in aquifer adjacent to the Thur River in Switzerland (Vogt et al., 2010). Dyes and temperature variations have also used as tracers to estimate travel times (Anderson, 2005; Hoehn et al., 2006; Verstraeten et al., 1999). Recently, a wide range of chemical isotopes has been used to assess infiltration processes and retention times in BF systems (KĂĄrmĂĄn et al., 2013; Vogt et al., 2010). Modelling is also applied to estimate the travel times and flow path of BF systems. For example, advection models, such as PMPATH (Bosuben, 2007), can be used to calculate pore-water velocities for estimating travel times. These studies indicate that travel time is mainly affected by the distance to the riverbank, pumping rate, drawdown, and well spacing.
1.4.3 Redox process
The nature of the redox environment is very important in BF processes as this influences the occurrence of HMs in the bank filtrate such as copper, zinc, lead, iron, and manganese (Massmann et al., 2008). Furthermore, redox conditions determine the fate of OMPs in the zone of infiltration (Maeng, 2010) and influence pH of the bank filtr...