Lime softening

Lime, sodium carbonate (soda ash), and/or sodium hydroxide (caustic soda) are added to water to convert soluble calcium and magnesium hardness to insoluble calcium carbonate and magnesium hydroxide in a contact vessel for 60–90 min [10,12,14]. Lime is not a true coagulant but reacts with bicarbonate alkalinity to precipitate calcium carbonate. Magnesium hydroxide precipitates at high pH levels. The solids are collected as sludge from the bottom of the settling basin. The sludge-free water is clarified by filtration to remove any turbidity and remaining solids thereby ensuring there is no carryover.

Calcium and magnesium occur as bicarbonates primarily. Lime and caustic soda break down the bicarbonate ions (HCO₃)⁻ into water molecules and insoluble carbonate ions (CO₃)²⁻ as follows:

$\mathrm{Ca}{\left(\mathrm{OH}\right)}_2+\mathrm{Ca}{\left({\mathrm{HCO}}_3\right)}_2\to 2{\mathrm{CaCO}}_3\downarrow +2{\mathrm{H}}_2\mathrm{O}$

$2\mathrm{NaOH}+\mathrm{Ca}{\left({\mathrm{HCO}}_3\right)}_2+{\mathrm{Ca}}^{2+}\to 2{\mathrm{CaCO}}_3\downarrow +2{\mathrm{H}}_2\mathrm{O}+2{\mathrm{Na}}^{+}$

Similarly, lime or caustic soda are used to provide hydroxide for precipitation of magnesium hardness as Mg(OH)₂. Hardness may also be present in non-carbonate form as sulphate, chloride, or nitrate. In the case of non-carbonate calcium compounds, carbonate for precipitation is provided by adding sodium carbonate (soda ash):

${\mathrm{CaCl}}_2+{\mathrm{Na}}_2{\mathrm{CO}}_3\to 2\mathrm{NaCl}+{\mathrm{CaCO}}_3\downarrow$

Cold lime softening is never complete because calcium carbonate and magnesium hydroxide are slightly soluble in water. Hot lime softening (HLS) with water heated to 100°C, on the other hand, significantly reduces silica in addition to hardness and alkalinity. Heating aids in the completion of the softening reaction, which in turn increases the efficacy of silica removal by providing more solids, particularly magnesium hydroxide that absorbs silica. HLS is not, however, used for potable water applications. Lime softening is followed by filtration to remove any turbidity and solids. Inorganic coagulants are typically used to minimise carryover in cold process softening, whereas anionic polymers are used with hot process softening. Effluent hardness from a cold process softener ranges from 35 to 80 mg/l; effluent hardness from a hot process softener is as low as 10–40 mg/l. In order to reduce the hardness further (1 mg/l), ion-exchange softening is required.

Coagulants are often added in conjunction with lime to increase the settling rate of calcium carbonate and magnesium hydroxide. Most of these coagulants are acidic in nature and react with the alkalinity of the water. Commonly used coagulants include aluminium sulphate (alum), sodium aluminate, ferric sulphate and ferrous sulphate (Table 6.8). Alum reacts with natural alkalinity in water to form aluminium hydroxide floc (Equations 2.5–2.8) [14]. About 1 ppm of alum decreases water alkalinity by 0.5 ppm and produces 0.44 ppm of CO₂:

${\mathrm{Al}}_2{\left({\mathrm{SO}}_4\right)}_3\cdot 18{\mathrm{H}}_2\mathrm{O}+3\mathrm{Ca}{\left({\mathrm{HCO}}_3\right)}_2\iff 2\mathrm{Al}{\left(\mathrm{OH}\right)}_3+3{\mathrm{CaSO}}_4+18{\mathrm{H}}_2\mathrm{O}+6\mathrm{C}{\mathrm{O}}_2$

${\mathrm{Al}}_2{\left({\mathrm{SO}}_4\right)}_3\cdot 18{\mathrm{H}}_2\mathrm{O}+3\mathrm{Mg}{\left({\mathrm{HCO}}_3\right)}_2\iff 2\mathrm{Al}{\left(\mathrm{OH}\right)}_3+3{\mathrm{MgSO}}_4+18{\mathrm{H}}_2\mathrm{O}+6\mathrm{C}{\mathrm{O}}_2$

${\mathrm{Al}}_2{\left({\mathrm{SO}}_4\right)}_3\cdot 18{\mathrm{H}}_2\mathrm{O}+6\mathrm{Na}{\left({\mathrm{HCO}}_3\right)}_2\iff 2\mathrm{Al}{\left(\mathrm{OH}\right)}_3+3\mathrm{N}{\mathrm{a}}_2\mathrm{S}{\mathrm{O}}_4+18{\mathrm{H}}_2\mathrm{O}+6\mathrm{C}{\mathrm{O}}_2$

$\mathrm{A}{\mathrm{l}}_2{\left(\mathrm{S}{\mathrm{O}}_4\right)}_3\cdot 18{\mathrm{H}}_2\mathrm{O}+3\mathrm{N}{\mathrm{a}}_2\mathrm{C}{\mathrm{O}}_3\iff 2\mathrm{Al}{\left(\mathrm{OH}\right)}_3+3\mathrm{N}{\mathrm{a}}_2\mathrm{S}{\mathrm{O}}_4+15{\mathrm{H}}_2\mathrm{O}+3\mathrm{C}{\mathrm{O}}_2$

Ion-exchange softening

Water softening by ion exchange (IX) uses strong acid cation (SAC) resins in the sodium form (− SO₃Na) to remove scale-forming cations from water [10,12,14,15]. Ion-exchange softening involves the exchange of hardness ions such as calcium, magnesium, strontium and barium for sodium ions to yield low hardness, or “soft” water (the softened water has a higher total dissolved solids (TDS)), as shown in Equation (2.9), where a cation resin (R) selectively removes calcium ions by the following reaction:

$\mathrm{R}-2\mathrm{Na}+\mathrm{C}{\mathrm{a}}^{2+}\to \mathrm{R}-\mathrm{Ca}+2\mathrm{N}{\mathrm{a}}^{+}$

IX softening is typically used when the hardness is in the range of 50–500 mg/l [16], and for feed water flow rates of up to 60 m³/h (see also Chapter 6).

The exhausted resin is regenerated with a dilute NaCl (brine) solution. This removes calcium and magnesium in the form of their soluble chlorides and at the same time restores the resin to its original sodium form. The bed is rinsed free of undesirable salts and returned to service. The regeneration reaction may be written as:

$\mathrm{Ca}/\mathrm{Mg}-\mathrm{R}+2\mathrm{NaCl}\to {\mathrm{Na}}_2\mathrm{R}+{\mathrm{CaCl}}_2/{\mathrm{MgCl}}_2$

The IX softeners normally operate at linear velocities of 14–20 m/h. About 8.5 lb (3.9 kg) of salt (NaCl) is required to regenerate 1 ft³ (0.3 m³) of resin, and remove approximately 4 lb (1.8 kg) of hardness. Hardness is given in grains/gallon or in ppm as CaCO₃. The reduction is directly related to the amount of cations present in raw water and the amount of salt used to regenerate the resin bed. Typically, 6 lb of NaCl/ft³ of resin is used for regenerating SAC resins.

Ion-exchange dealkalisation usually employs weak acid cation (WAC) carboxylic acid resins operating on the hydrogen cycle. Dealkalisation and alkaline hardness removal are synonymous terms that mean the water is only partially softened after such a treatment. Complete softening may be achieved by treating it further in the conventional sodium cycle to remove the remaining permanent hardness. The high selectivity for divalent ions such as calcium and magnesium over monovalent sodium ions results in the preferential exchange of calcium and magnesium ions (denoted M^2 +) [15]:

$2\mathrm{R}-\mathrm{COOH}+{\mathrm{M}}^{2+}\to {\left(\mathrm{RCOO}\right)}_2\mathrm{M}+2{\mathrm{H}}^{+}$

In the presence of hydrogen ions, the above exchange does not occur. However, if alkaline hardness (bicarbonate) is present, the exchanged hydrogen ions are immediately neutralised by the basic bicarbonate and carbonate anions to give carbon dioxide, which dissolves in water as weak carbonic acid:

${\mathrm{H}}^{+}+{\mathrm{HCO}}_3^{-}\to {\mathrm{H}}_2\mathrm{O}+{\mathrm{CO}}_2$

$2{\mathrm{H}}^{+}+{\mathrm{CO}}_3^{2-}\to {\mathrm{H}}_2\mathrm{O}+{\mathrm{CO}}_2$

The exchange of calcium and magnesium ions continues until all basic anions are neutralised after which time no further exchange can occur. Hence, the extent of hardness removed is equivalent to the alkaline hardness of the water. The resins are regenerated with either dilute hydrochloric acid or dilute sulphuric acid over a period of 30 min. The weak acid resin is regenerated at virtually 100% efficiency. Weak base anion (WBA) resins are also effective in removing strong mineral acid anions such as sulphates, chlorides, and nitrates. Hence, as in the case of IX softeners, WBA are sometimes used to treat RO feed water, thereby, reducing the potential of mineral scaling of RO membrane surface. The exhausted bed is regenerated with sodium hydroxide (caustic soda).

NF softening

Membrane softening by NF is a relatively new application as discussed in Chapter 1. NF membranes (“loose RO”) operate at a lower feed pressure than RO membranes, and have a high rejection (99%) of divalent hardness ions. It is a more attractive alternate to lime softening and IX softening because not only is it a reliable process, no regeneration is required and, thus, there is no chemical wastewater. NF separation like RO is a continuous process and is independent of the plant capacity (flow rate) and feed water hardness. It reduces both the hardness and the TDS to a much greater degree than IX and lime softening [5].

Continuous cross-flow filtration

Cross-flow UF and MF systems are mainly used for applications in biotechnology, dairy, colour removal from groundwater and industrial wastewater. Industrial wastewater treatment includes electroplating rinse water processing for paint recovery, treatment of oil/water emulsions, processing wastewater containing heavy metals, oil and grease prior to effluent discharge, textile wastewater, and pulp and paper wastewater. Cross-flow UF and MF systems – mostly tubular and hollow-fibre membranes – operate in continuous mode. The feed flow is inside out, i.e. the feed flows in the tube side and the permeate flows radially out of the tube wall. Ninety percent of the reject is typically recycled back to the feed tank at a high velocity of ~ 4–5 m/s. Because of the large i.d. as is the case with tubular membranes, the membrane elements can handle feeds with solid levels of up to 5%. The process is facilitated by operating in a turbulent regime to reduce the build-up of solute cake on the membrane surface. High cross-flow velocity (turbulent flow and surface shear) and large i.d. often eliminates the need for pre-filtration. The permeate flow rate is 10–15% of the cross-flow rate at a flux of ~ 300–500 lmh. Thus, a UF/MF system, designed for 20 m³/h permeate flow rate, may require a feed/recirculation pump rated for 250 m³/h at 3–4 bar g feed pressure. Backpulsing with filtered water or permeate reduces the frequency of cleaning cycles. Overall wastewater recoveries of up to 95% are achieved. Cross-flow systems are best suited for relatively small flow systems (up to 100 m³/h) and special applications. They are not economical for treating large systems such as municipal and seawater treatment because of several drawbacks:

• High cross-flow velocity (high shear rate) means high feed/recycle flow rate resulting in high energy cost of pumping.

• Low surface area and packing density (especially tubular membranes) result in a large footprint.

• Recovery per pass is only 10%.

• Frequent chemical cleaning is required to remove or dissolve the strongly held accumulated particles that are not dislodged with backwashing.

• High Capex and Opex

Semicontinuous dead-end filtration

The most commonly used membranes for municipal water treatment are hollow fibre modules that are supplied in three operating formats: (i) pressure driven inside feed – PDI, (ii) pressure driven outside feed – PDO and (iii) submerged vacuum driven – SUBO [6,9,17]. Classification and specifications of membrane modules are detailed in Table 6.14. These UF/MF systems operate in semicontinuous dead-end mode with intermittent backwash often combined with air scour either during filtration and/or backwash cycles [9]. The flow regime typically is from outside of the fibre (shell side) through the pores in the membrane wall to the lumen side. As the water gets filtered, the solids rejected by the membrane form a layer on the membrane surface. The foulant cake layer is removed by backwashing the membrane elements by air scouring and air-assisted backwash every 30–60 min to maintain the flux (constant flux operation). Over time, chemical cleaning is required to remove traces of foulants that are difficult to dislodge by backwash. Backwashing reduces the frequency of chemical cleaning, thereby, enabling the membrane system to run on feed water with turbidity as high as 500 NTU. The filtrate or product water turbidity is typically ≤ 0.1 NTU. Since the solids content is low (< 0.5%) in water treatment applications, submerged membranes are now increasingly used instead of pressurised membranes.

Membrane bioreactors

Membrane bioreactors (MBRs) are a special application of membrane filtration and discussed in detail in Chapter 3. It is a novel process that combines a biological stage with a membrane element. In this process, biological degradation of organic pollutants is carried with microorganisms in the bioreactor followed by membrane filtration to separate microorganisms [20,21]. The use of membranes to remove solids from treated wastewater is the main difference between MBRs and conventional biological treatment plants; higher removal efficiency than conventional treatment plants, e.g. the MBR, allows a higher biomass concentration, higher COD removal (> 90%) and higher separation of solid suspensions (complete retention of the biomass). The MBR process has shown to be more efficient in removing total BOD, turbidity and coliforms. By eliminating the problem of poorly settling flocs when using a MBR system, there is more biological degradation resulting in higher treatment efficiency.

The types of membranes used in MBRs are (i) flat sheet (FS), (ii) hollow fibre (HF) and multitube (MT). There are two types of MBR technologies: (a) external MBR (eMBR) in which the membrane modules are placed outside the bioreactor (MT only), and (b) submerged MBR (sMBR) where the membrane module is placed inside the bioreactor (FS and HF) [20]. The sMBR is more economical and energy efficient: (a) there is no recycle pump since aeration generates a cross-flow across the membrane surface, and (b) the operating conditions are milder because of lower values of TMP and tangential velocity. The TMP values are 1–4 bar for eMBR and 0.5 bar for sMBR systems. The permeate flux for sMBR, however, is lower: 15–50 lmh vs. 50–120 lmh for eMBR. Polymeric MF membranes with a pore size of 0.1–0.4 μm are the main membranes used in sMBR systems while tubular inorganic membranes are generally used in eMBR units. The eMBR units are preferred to sMBR units when treating concentrated effluents or concentrated biomass to avoid membrane fouling; higher shear rate is achieved because of higher recycle flow rate.

For municipal wastewater treatment, MBR systems are economically attractive where space is limited especially in urban areas or when high effluent quality is required for water reuse. Most MBR units use submerged membrane plates or hollow fibres. Fouling by microorganisms as a result of microbial products, concentration and size of particles is one key area of process deficiency. Different strategies have been considered for controlling fouling including backwashing. In addition, there is concern whether oxygen can become the limiting factor during aerobic biological activity [20]. Since the particulate solids content in the wastewaters treated by submerged MBR systems is relatively high (e.g. TSS = 10–15 mg/l) with additional colloids and macrosolutes, bubbled cross-flow is used to minimise concentration polarisation and subsequent fouling [9].

Manganese greensand oxidation-filtration

Some well waters, usually brackish waters, contain divalent ferrous iron, manganese and sometimes sulphide in the absence of oxygen. When water is exposed to air or is chlorinated, these compounds are oxidised resulting in the formation of insoluble colloidal hydroxides and elemental sulphur as [12,14]:

$4\mathrm{Fe}{\left(\mathrm{HC}{\mathrm{O}}_3\right)}_2+{\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}\to 4\mathrm{Fe}{\left(\mathrm{OH}\right)}_3+8\mathrm{C}{\mathrm{O}}_2$

$4\mathrm{Mn}{\left(\mathrm{HC}{\mathrm{O}}_3\right)}_2+{\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}\to 4\mathrm{Mn}{\left(\mathrm{OH}\right)}_3+8\mathrm{C}{\mathrm{O}}_2$

$2{\mathrm{H}}_2\mathrm{S}+{\mathrm{O}}_2\to 2\mathrm{S}+2{\mathrm{H}}_2\mathrm{O}$

The manganese greensand filtration (MGF) process employs manganese greensand as the filter medium to remove soluble iron, manganese and hydrogen sulphide by one-step oxidation filtration. The filter bed consists of anthracite filter media over manganese zeolite (greensand). Greensand is processed from glauconite, which is an iron potassium alumino-silicate material of marine origin. The filter-bed material acts as a catalyst for the reaction of iron with oxygen. It can supply the necessary oxygen, which must be replenished with an oxidising regenerant solution such as potassium permanganate (KMnO₄).

The most common method of oxidative MGF is known as continuous regeneration. Permanganate solution (1–2%) is injected into the raw water feed water line. Precipitation of iron and manganese occurs in the anthracite bed as per the reaction in Equation (2.19). Heavier precipitates are filtered out in the anthracite bed, and the remaining fines and residual dissolved metal are removed by greensand. The water pH should be about 7.2.

$\mathrm{Fe}{\left(\mathrm{HC}{\mathrm{O}}_3\right)}_2+\mathrm{KMN}{\mathrm{O}}_4\to \mathrm{Fe}{\left(\mathrm{OH}\right)}_3\downarrow +\mathrm{Mn}{\mathrm{O}}_2\downarrow$

Iron concentrations up to 15 ppm can be effectively removed, although concentrations greater than about 3 ppm require low service flow rates, which results in short runs between the backwash cycles. At low feed water iron concentrations, the flow velocity is 14–20 m/h for up to 36 h between the backwash cycles. Lower flow velocity 4 m/h and shorter service run lengths are advised when the feed water iron concentration is 3–15 ppm.

Chlorination

The disinfectant action of chlorine results from its strong oxidising action on bacterial cell’s chemical structure that destroys the enzymatic processes required for life [12]. The effectiveness of chlorine disinfection is a function of the product of contact time and chlorine residual. Chlorine gas is soluble in water (7160 mg/l at 20°C and 1 bar), and hydrolyses rapidly to form hypochlorous acid or as the hypochlorites of sodium and calcium:

${\mathrm{Cl}}_2+{\mathrm{H}}_2\mathrm{O}\to \mathrm{HOCl}+{\mathrm{H}}^{+}+{\mathrm{Cl}}^{-}$

$\mathrm{NaOCl}+{\mathrm{H}}_2\mathrm{O}\to \mathrm{HOCl}+{\mathrm{NaOH}}^{-}$

$\mathrm{Ca}{\left(\mathrm{OCl}\right)}_2+{\mathrm{H}}_2\mathrm{O}\to 2\mathrm{HOCl}+\mathrm{Ca}{\left(\mathrm{OH}\right)}_2$

Hypochlorous acid (HOCl) dissociates in water to hydrogen ions and hypochlorite ions (OCl⁻). The sum of Cl₂, NaOCl, Ca(OCl)₂, HOCl and OCl⁻ is referred to as “free available” chlorine. The power of free chlorine residual decreases with increasing pH. Hypochlorous acid concentration at 20°C is 90% at pH 7, 50% at pH 7.6 and 10% at pH 8.6. Almost the reverse is true for hypochlorite ions. Hence, automatic monitoring of residual chlorine and automatic feedback of injection rate is necessary to prevent over-dosage or inadequate disinfection. Minimum chlorine residuals for bactericidal disinfection after 60 min of contact vary between 1 ppm at pH 6.0 and 1.8 ppm at pH 8.0. A hypochlorous residual of 0.5–1.0 ppm is effective within 30 min.

In gaseous form chlorine is extremely hazardous. Liquid chlorine is shipped in pressurised steel cylinders. One volume of liquid chlorine yields about 450 times vapour volume. Because of safety concerns liquid chlorine compounds such as sodium hypochlorite are used instead. Sodium hypochlorite (NaOCl) is handled in liquid form at concentrations between 5 and 15% available chlorine. Calcium hypochlorite [Ca(OCl)₂] contains about 70% available chlorine.

An alternate technique is on-site generation of NaOCl, which is achieved through electrolysis by applying an electrical current to a solution of salt (preferably food-grade) and water. The electrolytic cell is the heart of the oxidant producing unit. The cell consists of two electrodes placed so that both make contact with the water and brine solution. A by-product of the electrolysis reaction is hydrogen gas, which is safely removed from the cell and the oxidant storage system. On-site generators are used to provide disinfection for swimming pools, cooling towers and sanitation for clean-in-place operations. The largest application of on-site NaOCl generators is for municipal drinking water disinfection.

Ozonation

Ozone is a strong oxidant and disinfectant. It has similar bactericidal properties to chlorine and is at least equal to chlorine in its ability to perform virus inactivation [12]. Ozone is also an option for odour control in reclaimed wastewater treatment. Ozone oxidises organic contaminants via a reaction with molecular ozone (O₃) or through a reaction with the hydroxyl radical (OH^•), which is formed often in the presence of a catalyst when ozone is added to water. Ozone is generated on-site from air or oxygen zone when a high voltage is imposed on a discharge gap. It is generated on-site because it is a relatively unstable gas with a half-life of about 10 min. When ozone is added to water, it rapidly reverts to oxygen so that no chemical residuals remain in the ozonated water. Chlorine is added for post-disinfection, if required. There are, however, some adverse health effects due to the formation of bromates and aldehydes as by-products. The maximum allowable limit of bromate, a carcinogen, in drinking water is 25 ppb (μg/l).

UV irradiation

Ultraviolet (UV) light represents a band of electromagnetic light in the 100–400 nm range. It is a non-chemical disinfectant process that uses a very short contact time (< 5 s). UV light is used to break specific chemical bonds, sometimes by direct photolysis, but usually by the creation of highly reactive hydroxyl (OH-) radicals. Photolysis applications include dechlorination, de-ozonation, removal of TOC and more recently to counter the threats caused by endocrine disruptors and pharmaceutical compounds (both metabolised and un-metabolised).

UV disinfection avoids a major disadvantage of chlorination, i.e. the generation of harmful by-products such as trihalomethanes. UV disinfection ensures that the drinking water is free from pathogens such as e-Coli, Legionella and Cryptosporidium. UV is also extensively used in manufacturing processes such as bottled water, beer, carbonated soft drinks and ultrapure water. Disinfection by UV radiation involves damaging the genetic material of organisms by energy in the form of light of wavelength 254 nm emitted from a low-pressure mercury vapour lamp [12]. This particular wavelength alters the genetic material (DNA) of bacteria, viruses and other microorganisms. With their DNA altered, they are unable to reproduce and die within minutes. The dead bacteria produce TOC.

For microbial destruction, 254 nm UV energy is used, whereas shorter (and more powerful wavelength) electromagnetic radiation, 185 nm UV energy, is used to reduce organic compounds and chlorine destruction. The energy of a light beam is inversely proportional to the wavelength. Thus, 185 nm UV irradiation carries more energy and is more powerful than the 254 nm light; 185-nm energy oxidises total oxidisable carbon (TOC) to form carbon dioxide and water. Most microorganisms are damaged at a UV irradiation dosage level of 10,000–30,000 μW-s/cm², whereas the dosage level required for reducing ozone, chlorine and TOC reduction is 90,000 μW-s/cm².

UV systems generally consist of a reactor with a number of lamps that emit UV radiation. Each lamp is encased in a quartz tube. As the water flows thorough the lamps, it gets radiated. Polytetrafluoroethylene (PTFE) is also used since PTFE is transparent to UV radiation. UV radiation dosage is determined by residence time in the reactor and the intensity of radiation. The effectiveness of a UV system depends on the hydraulic design of the reactor, absorbance of liquid, presence of particles and transmittance of the lamps and the quartz tubes. Intimate contact between the liquid and the lamps needs to be maintained at all points in the reactor since radiation diminishes with distance. Contact time for radiation exposure is maximised by maintaining plug flow in the reactor and by creating turbulence to cause transverse dispersion. The UV lamp's intensity typically drops to 60% in a year or after 8000 h of use.

Operating parameters

The quality of treated water depends on the amount of ions produced (mg) or charge loading (A h), i.e. current and time. The supply of current determines the amount of Al^3 + (or Fe^2 +) released from the respective electrodes. A large current means a small EC unit. However, a very high a current usually results in wasting electrical energy in heating water. The optimum current density for EC is 20–25 A/m². However, the current density value should take into account other operating parameters such as pH, temperature and flow rate to ensure high current efficiency.

The optimal temperature for operating the system is 60°C resulting in lower energy consumption as a result of higher conductivity than at ambient temperature. Higher water conductivity, for example, with the addition of table salt (NaCl), is beneficial as it results in reduced power consumption. Besides its ionic contribution in carrying the electric charge, chloride ions can significantly reduce the adverse effects of bicarbonate and sulphate ions; HCO₃⁻ and SO₄²⁻ ions can result in the precipitation of Ca^2 + and Mg^2 + ions forming an insulating layer on the surface of the electrodes resulting in reduced current efficiency [23]. Further, electrochemically generated chlorine is effective in water disinfection.

Treated waters

EC is effective in water treatment for drinking water supply, membrane pre-treatment, marine operation, and boiler water supply for small systems. It is very effective in treating colloids found in natural water by reducing turbidity and colour as well as removing suspended solids, oil and grease, iron, silicates, humus and microorganisms. A comprehensive summary of pollutants removed by EC is given in [24].

IX resins

Most IX resins are based on a styrene-divinylbenzene copolymer, which is appropriately treated to graft on ionic or functional groups. For example, sulphonation produces a cation resin, while amination produces an anion resin [15].

Strongly acidic cation resins

These resins have a sulphonic acid (–SO₃) functional group as shown in Figure 2.8. The ones in the H-form (–SO₃H) are capable of removing all cations from water. For most deionisation applications, SAC resin with an 8% divinylbenzene (DVB) content are used. Generally, they show maximum selectivity for trivalent ions, followed by divalent and monovalent ions. They are regenerated with strong acids, usually HCl or H₂SO₄, and require 200–300% of the theoretical stoichiometric quantity. When used in the hydrogen cycle, the effluent is acidic. The maximum operating temperature of the resin is 135°C.

Weak acid cation resins

This class of cation-exchange resin is typified by the carboxylic acid (–COOH) functional group, which in its acid form is only very weakly ionised, e.g. methacrylic-divinylbenzene, –R′C(CH₃)COOH. The degree to which dissociation occurs is highly pH dependent, increasing with increasing external pH. The resins are used for removing alkaline hardness as discussed under “softening.”

Strongly basic anion resins

There are two types of strongly basic anion (SBA) resins: Type I and Type II. Type I resin has the highest basicity, and, therefore gives the optimum effluent quality. Type II has a lower basicity and, therefore, requires less caustic for regeneration. In general, Type II resin is used where silica effluent quality is not critical. Both resins are usually based on styrene DVB; however, recent resins now include acrylic materials. The functional group in Type I is quaternary benzyltrimethyl ammonium chloride, RCH₂N ⋅ (CH₃)₃⁺⋅ Cl⁻, and the functional group in Type II class is RCH₂N⋅(CH₃)₂(C₂H₄OH)⁺⋅ Cl⁻. These resins are capable of removing all anions from water. Selectivity is maximum for divalent ions. SBA resins are regenerated with a strong alkali such as caustic soda. The maximum operating temperature of SBA resin is 60°C (OH⁻ form) and 77°C (Cl⁻ form).

Weak base anion resins

The WBA resins come in two forms: free base form, CH₂N(CH₃)₂, and chloride form –CH₂N(CH₃)₂ ⁺Cl⁻. WBA resins operate essentially as acid absorbers. They are more efficient than strong base anion resins for removing free mineral acids (FMAs) such as HCl, H₂SO₄ and HNO₃, but are not effective at removing carbon dioxide or silica.

Resins selection

Other than the selection of appropriate resin type (cation or anion) and its strength (strong or weak) there are no firm rules, and experience outweighs all other considerations. Guidelines concerning the options related to matrix, structure and particle size grading are provided by resin manufacturers. Cost is also a major factor. Matrix-modified resins (acrylic) and structure modified materials (macroporous) can increase resin cost between 20 and 150% compared with standard gel styrenic products when comparing anion or cation resins, respectively.

Physical characteristics

The size and range (0.3–2 mm) of resin beads is an important consideration vis-à-vis performance. Uniform beads have been developed to produce stronger resins that resist attrition and eliminate cross-contamination of the resins during backwashing in mixed-bed systems. Uniform size is important since it imparts [10,15,16]:

• Higher operating capacity

• Lower ionic leakage

• Cleaner separation of resins during mixed-bed regeneration

• Faster rinsing

Diffusion of ions into and out of a large bead limits its performance. Hence, the diffusional path length should be as short as possible to achieve:

• Increased utilisation of the functional sites located within the bead

• Higher efficiency with which regenerating solution chemicals get into and out of the bead

Limitations

The basic limitation of an IX resin to remove and exchange ions and concentrate them within its structure is determined by the “selectivity” and the “capacity” of the resin. The ions being exchanged compete for the exchange sites with the resins having a different affinity for each type of ion. Note that since IX resins are primarily designed for removing soluble ionic species, they cannot be expected to absorb macro- or colloidal ionic species except at the surface of the particle.

Demineralisation

There are two types of resins involved in deionisation water treatment applications. These are strongly acidic cation-exchange resins and strongly basic anion-exchange resins. If the regenerating solution was a strong acid, such as hydrochloric or sulphuric acid, instead of a salt, the metallic ions would be replaced with hydrogen, and the resin would then remove cations, replacing them with hydrogen ions. This is called hydrogen-cycle cation-exchange. Thus,

$\mathrm{R}-2\mathrm{H}+{\mathrm{Ca}}^{2+}\to \mathrm{R}-\mathrm{Ca}+2{\mathrm{H}}^{+}$

Anions are exchanged in a similar fashion. Anion resins usually operate in either the hydroxide cycle (regenerated with a strong alkali) or the chloride cycle (regenerated with sodium chloride):

$\mathrm{R}-2\mathrm{OH}+{{\mathrm{SO}}_4}^{2-}\to \mathrm{R}-{\mathrm{SO}}_4+2{\mathrm{OH}}^{-}$

By using hydrogen-cation and hydroxide-anion resins in series, all cations are replaced with H⁺ ions and all anions with OH⁻ ions; the result is pure water. This process is called deionisation or demineralisation.

Dual-bed demineralisation

In a two-stage or dual-bed IX demineralisation process, raw water is first passed through a SAC exchange resin bed in the hydrogen form. The effluent from the cation column is then passed through a strongly basic anion-exchange resin bed in the hydroxide form. Across the cation exchanger, all cations exchange for hydrogen ions to give a dilute acidic effluent (“free mineral acids” or FMAs) made up of acid sulphates, nitrates, and chlorides together with dissolved carbon dioxide. Upon passing through the anion exchanger, neutralisation of the FMAs occurs through the exchange of all anions for hydroxide to give deionised water. A typical IX vessel is shown in Figure 2.9 (left). Dual-bed sample design calculations are given in Chapter 6.

Mixed-bed ion exchange

A mixed-bed (MB) contains a uniform mixture of strong acid and strong base (Type I) resins in the H⁺ and OH⁻ forms, respectively. Demineralisation occurs through simultaneous exchange of cation and anions in a single vessel. In effect, this is equivalent to having an infinite series of two-bed demineralisers in series. Every anion bead reacts instantly with the acid produced by a neighbouring cation bead, removing the acid as it forms and driving the reaction to complete demineralisation. A pictorial drawing of a MB vessel is shown in Figure 2.9 (right).

The development of the MB-IX system is considered to be the most important contribution to the application of IX technology since the development of IX resins. The key to the successful operation of the mixed-bed is the ability to separate the two resins after the service cycle so that they can be regenerated separately. Uniform beads have been developed to produce stronger resins that resist attrition and eliminate cross-contamination in separation of the resins from one another during backwashing. Not only is cross-contamination of the resins undesirable from the obvious loss of capacity, but it can also reduce the quality of water. One way to reduce cross-contamination occurring at the interface distributor is by the use of an inert resin. The inert resin has a specific gravity intermediate between those of the anion and cation resins that, hydraulically, classifies between the anion and cation resins when backwashed. The quantity of the inert resin (10–15% of total resin volume) should be enough to cover the interface distributor, thereby, effectively reducing the cross-contamination problem.

Normal operation

An IX vessel is designed to distribute influent water evenly over the resin bed so that it passes through the bed uniformly. This ensures that the bed remains in a packed condition. Disruption of the bed during service would increase the amount of impurity allowed through (called slippage), and reduce the resin bed’s operating capacity. In down-flow service, those ions for which the resins have the strongest affinity or selectivity are held at the top of the bed, i.e. those with the lowest selectivity are displaced from the exchange sites by other ions and move down the bed. When the exchange front eventually reaches the bottom of the bed as shown in Figure 2.10, the vessel is taken off line for regenerating the resins. The first sign of column breakthrough or ionic leakage is indicated by the most loosely held ions, silica (HSiO₂⁻) for anion resins and sodium (Na⁺) for cation resins.

The operation of a MB is similar to a single-bed IX operation. The regeneration cycle shown in Figure 2.11 is, however, complex. Backwashing, in addition to cleaning as discussed above, helps to classify the resin: the lighter anion beads rise to the top of the bed, the heavier cation beads dropping to the bottom. An inert resin bed with intermediate density forms a layer between the two resins helping to separate them during regeneration and reduces likely cross-contamination of resins. Acid solution flows through the bottom and upward through the cation resin bed and out through the interface collector. Caustic soda solution is introduced above the bed level and flows down through the anion resin bed and out through the interface collector. The resins are then rinsed as described above and mixed with air or nitrogen before returning to service. The regeneration process is described in detail in Chapter 4.

Regeneration

Step I – Backwashing. This is accomplished by passing water upward through the bed for 10–20 min at a velocity of 7–15 m/h sufficient to expand the bed by 50–100%. Backwashing removes particulate matter, relieves any bed compression, and allows trapped gases to escape.

Step II – Chemical regeneration. Regeneration displaces the ions that were exchanged during the service run, returning the resin to the original ionic form. This is accomplished by using solutions of sufficient strength (e.g. 8–10% NaCl for softener cation resin) to drive the equilibrium established during operation in the reverse direction. Cation-exchanger resins are regenerated with strong acids such as H₂SO₄ and HCl. Regeneration with H₂SO₄ is usually done at two different concentrations; initially at 2–3% acid followed by 5–6% dilute acid. Anion-exchanger resins are regenerated with NaOH at 4–5% dilute caustic soda solution.

The regenerating solution flow rate is low enough (2–4 m³/h m³ of resin) to allow the chemical to diffuse into the resin and to allow the largest impurity ions to migrate out of the resin. Regeneration efficiency can sometimes be improved, significantly, by heating the regenerating solution. Heating caustic soda to 49°C aids in removing polymerised silica from SBA resins.

Step III – Rinsing. Excess regenerating chemicals from the resin bed are flushed by a two-step rinsing process. The first step is a displacement (slow) rinse, in which the flow rate is the same as that used for adding the regenerant solutions (typically replacement of two bed volumes). This ensures that the chemicals are fully utilised. The second step is a fast rinse, typically 10 min. The high flow rate of water during rinsing flushes out any traces of the regenerant solution.

Organics removal

Organic matter in surface water is supposed to behave as a colloid, and is often complexed with silica and heavy metal ions such as iron, aluminium and manganese. In order to minimise the risk of organic fouling, SBA resins operating on a co-flow chloride cycle are often employed. It is not essential that the resin has a high ion-exchange capacity. Rather, it is more desirable that they possess greater porosity even at the expense of capacity. Typically, 50–70% of organics are removable.

Nitrates removal

Biological denitrification and IX are the processes used for removing nitrates from contaminated groundwaters when present in carcinogneic concentrations (> 50 mg/l). Ion exchange is the easiest and most economic. The process uses SBA resin bed in either co-flow or counter-flow arrangement. To ensure reliable operation nitrate-selective resins have been developed. Regeneration is carried out using NaCl solutions.

Industrial wastewater treatment

SAC and SBA resins are typically used to treat wastewater effluents from metal finishing processes such as plating and anodising, pulp and paper manufacture, chemical leaching, zinc smelting, metal pickling and photographic processing plants. For example, in metal pickling where hydrochloric acid is used in the steel galvanising process, cation is removed as a complex anion. Strong base anion resins in the chloride form readily take up the chloro-complex ions (${\mathrm{Fe}}^{3+}{{\mathrm{Cl}}_4}^{-}$ and ${\mathrm{Zn}}^{2+}{{\mathrm{Cl}}_4}^{-}$), thereby rejuvenating hydrochloric acid. Similarly, SBA resins and sometimes chelating resins are used for recovering heavy metals from various process streams.

Carbohydrate refining

IX processes in carbohydrate treatment are used for the purification of juices and syrups from cane sugar, beet sugar and corn starch hydrolysates. The main operations are decalcification (softening), deashing (demineralising) and decolourising (removal of organic colour bodies). These processes improve the yield and quality of the final recrystallised sugar or concentrated syrup. One interesting application is the inversion of sucrose to invert sugar (fructose + glucose) using SAC resins in the hydrogen form. Since raw sugar processes yields a fairly viscous syrup, macroporous resins are often used in sugar refining. Decolourising is usually done with macroporous WBA resins.

Pharmaceutical processing

IX is widely used to concentrate and recover antibiotics and vitamins from fermentation broths. For example, streptomycin molecule, which is moderately basic, is very favourably exchanged by the sodium form of macroporous WAC resins. General categories of biological compounds that are recovered and purified by IX include antibiotics, vitamins, nucleotides, amino acids, proteins, enzymes and viruses.

Organic fouling

Humic acid and fulvic acid represent up to 80% of the TOC – dissolved organics – of natural waters. Humic substances are weak acidic electrolytes with carboxylic- and phenolic-OH groups with a micelle-like structure and with a molecular weight between 500 (fulvic) and 100,000 (humic). The chemical structure of humic acid is shown in Figure 6.1 [15]. They become hydrophobic as pH decreases, and thus foul hydrophobic membranes more [18]. Humic fouling in the case of natural waters is aggravated by the presence of calcium (Ca⁺) by forming a bridge between the membrane and the negatively charged membrane surface, and/or between negatively charged carboxyl groups of the humic acid. Other organic components that foul membranes include microbial slime, polyhydroxy aromatics and polysaccharides. Feed water organic levels measured as TOC should be low to prevent fouling with organic molecules as well as to minimise the potential for microbial fouling since organics provide nutrients that support microbial growth.

Colloidal fouling

Most common colloids found in natural waters are clays, silica, iron and aluminium hydroxides, and organic debris. In the case of industrial process streams, colloids may be paint pigments, proteins, bacterial and yeast cells, and high molecular weight alcohols. Colloidal fouling in a membrane system is caused by the convective deposition of colloids on the membrane surface, and the higher the permeate flux, the higher the rate of colloidal fouling. Membrane systems and operating conditions are designed to reduce the risk of colloidal fouling since it is the most severe [31,32]. These control measures include feed water pre-filtration, reduced recovery, higher cross-flow velocity and frequent chemical cleanings.

Silica fouling

Silica is one of the most common elements, and is present in many natural waters. It is found in waters in the form of (a) reactive silica, (b) colloidal silica, and (c) suspended particles (sand). Reactive silica is called silicon dioxide, and in this form is generally not ionised at normal pH levels of water [33,37]. Colloidal silica is either polymerised silicon with multiple units of silicon dioxide, or silicon that has formed loose bonds with organic compounds or other complex inorganic compounds such as calcium oxide and aluminium silicate. Well waters contain the most silica, from 50 to 100 mg/l. It is mostly reactive silica, a result of dissolving from rock, whereas surface waters contain more colloidal silica even though surface waters contain more reactive silica than colloidal silica. In the case of surface waters, silica chemistry is complex due to biological activity.

Silica has a low solubility of 120 mg/l at pH of 7.0 and 25°C, and silica scale-like barium sulphate scale is difficult to dissolve during cleaning and must be avoided. Silica scaling usually occurs in the last stage of a membrane array (Table 2.9). It is usually associated with iron and alumina (e.g. clays, mullite and feldspars); metal hydroxides adsorb silica to form silica–metal complexes such as aluminium silicates that readily foul the membranes. Iron fouling can be prevented by (a) anti-foulants, or (b) removing iron by oxidation as by greensand filtration or iron filters. Alumina fouling is prevented by (a) pH control and coagulation prior to multimedia filtration, or (b) alumina compatible anti-scalants. If silicates are less than 5 ppm, iron should be less than 0.5 ppm and manganese less than 0.2 ppm, and when the silicates are 30–50 ppm, iron should be less than 0.05 ppm and manganese less than 0.02 ppm. In addition, hardness should be removed and alumina should be less than 0.05 ppm. A dispersant such as high molecular weight polyacrylate scale inhibitor is helpful in silica scale control by slowing agglomeration of silica-scale particles.

The solubility of silica is a function of pH and temperature. Lime softening and lime plus soda ash softening are most effective in removing silica. Other options are process related: (a) run the RO system at reduced recovery, (b) increase the feed water temperature – silica solubility increases with temperature, and (c) use silica inhibitors. Colloidal silica is difficult to remove by IX because it is not ionised, and can foul the resins when the levels are high. Colloidal silica flocculates easily in boundary layers resulting in severe fouling. It can be removed by UF membranes with a MWCO of up to 100,000 Da. Since the solubility of the silica increases below a pH of about 6.0 and above a pH of about 9.0, the actual solubility of silica in the concentrate stream is further affected by the pH of reject water.

Biofouling

Biological fouling by microorganisms has an adverse effect on membrane performance (Table 2.5). A biofilm is difficult to remove because it protects the microorganisms from shear forces and disinfection chemicals. Microorganisms – bacteria, algae, fungi, viruses, and higher organisms – can be regarded as colloidal matter (typical bacteria size is 1–3 μm). Biofouling can be complicated due to mutual interactions between different types of fouling; for example if iron oxide or a biofilm accumulates on the membrane surface it prevents the migration of ions (calcium, sulphate) and other solutes back into the bulk solution, as discussed in Chapter 1, resulting in a supersaturated condition. Thus gypsum (calcium sulphate) scale is formed beneath the primary foulant, and is not easy to remove as compared to the iron scale or the biofilm. The potential for biological fouling is higher with surface water than well water.

Biofouling has been investigated extensively [38,39]. Key characteristics are summarised below:

• Appropriate sampling of biofilms is necessary to identify biofouling. A biofilm has the following characteristics: (a) high content of water and organic matter (70–95%), (b) high numbers of colony-forming units and cells, (c) high contents of carbohydrates and proteins, (d) high content of adenosine triphosphate (ATP), and (e) low content of inorganic matter.

• RO membranes reject bacteria absolutely. However, microorganisms can be found in the permeate due to (a) leakage through O-rings, (b) microscopic imperfections in the membrane surface, (c) microbial contamination of feed water, and (d) growth of microorganisms from contaminated piping.

• Biofilm mode of growth enables microorganisms to survive and multiply even in extremely low nutrient habitats (5–100 ppb TOC).

• Generally biofouling is a slow process.

• The cumulative effects of membrane biofouling are (a) increased cleaning and maintenance costs, (b) a noticeable deterioration of product water quality, and (c) significantly reduced membrane life.

• Biofouling can lead to secondary mechanical deformation (“telescoping”) of SW membrane elements.

• Biofouling of membrane surfaces is invariably accompanied by some degree of mineral deposition.

• Biofilm can be considered as a dense gel layer, and dissolved minerals tend to accumulate in this layer and increase concentration polarisation.

• Biofouling causes flow losses due to constriction of the flow channel, increase roughness of the surface, and increase drag because of their viscoelastic properties.

• The first step in biofilm formation prior to microbial adhesion is the irreversible adsorption of macromolecules, which leads to a “conditioning film” (humic substances, lipopolysaccharides, microbial products). This conditioning film alters the effect of the membrane; the electrostatic charge and the critical surface tension may change.

• Primary adhesion is a function of several factors: (a) microorganisms, (b) membrane surface property – charge and hydrophobic/hydrophilic, (c) fluid properties, and (d) bacteria growth phase. Many bacteria have a slight negatively charge, and have to overcome the repulsion barrier when they attach to slightly negatively charged membrane surfaces. Cell hydrophobic property, however, does not seem to be crucial in adhesion to polysulphone (PS) hydrophobic membrane.

• Primary adhesion occurs within a very short time; the highest rate is during the first 1 h followed by a slow plateau after 4–6 h. Dead cells adhere as fast as living cells. This means that destruction of bacteria is not enough, and dead cells must be removed.

• Higher cross-flow velocity provides higher shear forces, and thus, a thinner biofilm. However, the shear forces of the turbulent flow do not “penetrate” the viscous sub-layer and affect the bacterial monolayer. Fluid velocities in SW modules have very little effect on impeding the initial rate of microbial (or colloidal) foulant deposition, although they can reduce the thickness of the fouling layer.

• Surface roughness has a significant effect on biofouling as it: (a) increases the convective transport near the surface, (b) protects small particles from surface shear, and (c) increases the surface area of attachment.

• Polycations from pre-treatment enhance biofouling because of the electrostatic attraction to the slightly negative overall charge of microbial cells.

• A common cause of biofouling is the overdosing of flocculants (used in removing suspended solids during pre-treatment), which provide a suitable habitat for microbial growth.

• Chlorination of seawater has been observed to induce biofouling by degrading humic acid into biologically assimilable material, which supports the growth of a biofilm.

There are several risk factors that are also applicable to non-biofouling membrane operating conditions. These should be kept in mind when operating any membrane system:

• Feed-water characteristics – Temperature > 25°C, high amounts of organic and inorganic nutrients, large number of cells (> 10⁴ colony-forming units per ml), and high SDI.

• Operational characteristics – Infrequent monitoring of performance characteristics, use of microbial contaminated pre-treatment chemicals, lower cross-flow velocities, and long storage periods.

• System design – Extended piping runs, dead-legs and disinfected holding tanks.

In most cases, biofouling is assessed based on performance characteristics: (a) flux decline, (b) decrease in salt rejection, or (c) increase in feed-brine pressure drop as given in Table 2.5. These indirect methods, however, are not absolute indicators of biofouling, and cannot be used for any correlation. Common techniques and limitations for controlling biofouling include:

• Removal of the biofilm is essential for effective sanitisation because a biologically deactivated biofilm can cause biofouling.

• Chlorine is a biocide. However, it does not sterilise water, and after chlorination the surviving bacteria can grow especially between 25 and 35°C. Water has to be dechlorinated when using polyamide (PA) membranes.

• Activated carbon used for removing organics is often the source for colonising microorganisms.

• Cartridge filtration using 5.0 μm pore size cartridges are too large for removing bacteria.

• 254 nm UV irradiation is used for disinfecting water. Many types of bacteria are, however, resistant to UV light. Once a biofilm is formed UV light is not able to deactivate or remove the cells. UV treatment is limited to relatively clean water because of interference by suspended solids in water.

• Regular ozone treatment is effective for sterility, and is used in high-purity water systems.

• Hydrogen peroxide requires long contact times (a few hours), high concentrations (> 3%), or high temperatures (> 40°C). It is an oxidant and, therefore, not compatible with PA membranes.

• Detergents are effective in inhibiting biofouling in many cases. Alkaline cleaners are quite effective in controlling microbial slime. Biz® bleach is an effective cleaner but is an oxidant.

• Sodium bisulphite is effective in controlling biofouling and colloidal fouling. Typically, 500–1000 mg/l is dosed for 30 min every 24 h to control aerobic bacteria. The permeate contains 1–4% bisulphite during this shock treatment.

• Hot water sanitisation (HWS) is an effective control strategy.

• Preventive sanitisation is much more effective than corrective disinfection because a single attached bacteria is easier to kill and remove than a biofilm. Membrane life is shortened by extensive sanitisation.

In-line coagulation

Polymeric coagulants are sometimes added in low dosages (< 10 ppm) to remove particles down to 0.5 μm particle size as compared to 10 μm without the coagulant upstream of a cartridge filter. Even very low amounts of cationic polymeric coagulants in RO/NF feed water can, however, foul negatively charged membranes. In addition, some organic coagulants are not compatible with acrylic-based anti-scalants used in RO/NF systems. Typical coagulant dosage ranges from 0.5 to 20 ppm. Inorganic coagulants generally require a higher dose than polymeric coagulants. The best technique for determining the proper dosage of coagulant is to feed the product in-line and measure the SDI in the filter effluent as a function of coagulation feed rate.

In-line dechlorination

Oxidising agents such as chlorine damage PA membranes by breaking down the polymer backbone as discussed in Chapter 6. The maximum allowable chlorine limit for PA membranes varies between 0 and 1000 ppm h. It is, therefore, mandatory that RO feed water is dechlorinated when using PA membranes. A reducing agent such as sodium metabisulphite (Na₂S₂O₅) is used to scavenge any traces of chlorine in carbon filter effluent water. Sodium bisulphite (NaHSO₃) is formed when sodium metabisulphite is dissolved in water. NaHSO₃ then reduces hypochlorous acid as per the reaction:

${\mathrm{NaHSO}}_3+\mathrm{HOCl}\to {\mathrm{NaHSO}}_4+2\mathrm{HCl}$

In theory, 1.34 mg of sodium metabisulphite (100%) is required to remove 1.0 mg of free chlorine. In practice, 3.0 mg is used. Similarly, it takes 1.46 ppm of sodium bisulphite and 1.77 ppm of sodium sulphite (Na₂SO₃) to remove 1 ppm of free chlorine. A minimum contact time of 5 s is required. Sodium metabisulphite also helps in reducing dissolved oxygen in water.

In-line pH adjustment

Membrane scaling is linked to system recovery and cross-flow velocity; generally speaking, for recovery up to 70%, and with or no iron present, acidification may be the only chemical pre-treatment necessary to prevent scaling by calcium carbonate. When water recovery is 75–80%, additional processes such as scale inhibitors and/or IX softening instead of acidification is required. The solubility of calcium carbonate depends on the pH of feed water. The equilibrium can be shifted to the right to convert calcium carbonate to soluble calcium bicarbonate [Ca(HCO₃)₂] by lowering the pH to 6.0 with sulphuric acid or hydrochloric acid as:

$\mathrm{CaC}{\mathrm{O}}_3+\mathrm{H}\leftrightarrow \mathrm{C}{\mathrm{a}}^{2+}+\mathrm{HC}{\mathrm{O}}_3$

Acid reacts with bicarbonate alkalinity to produce carbon dioxide. RO permeate is often high in anions and always acidic (pH < 7.0) due to the presence of dissolved carbon dioxide (carbonic acid). Carbon dioxide concentration can be several hundred ppm when using acidified feed [37]. Dissolved carbon dioxide is removed by tower decarbonation or by membrane degasification to reduce the loading on anion resins when EDI or mixed-bed IX is used for producing DI water. Membrane degasification is preferred in high-purity water systems to prevent contamination. Alternately, IX softening is used to remove calcium ions followed by raising the pH of the softened water to 8.3–8.5 by adding caustic soda. Raising the pH with sodium hydroxide converts the carbon dioxide to sodium bicarbonate, which is easily rejected by RO membrane.

Acid addition is determined by the LSI or SDSI of RO reject water. To control calcium carbonate scaling by acid addition alone, the LSI or SDSI in the concentrate stream must be negative as indicated in Table 2.6. When an anti-scalant (A/S) is used, the LSI can be 1.0. LSI is applicable when the TDS is less than 10,000 mg/l, whereas SDI is applied when the TDS is greater than 10,000 mg/l.

Sulphuric acid is commonly used but hydrochloric acid is preferred when the scaling potential is high due to CaSO₄, SrSO₄, and BaSO₄. Calcium sulphate is more soluble than BaSO₄ and SrSO₄. However, since calcium ion is present in natural water sources more abundantly than barium and strontium ions, CaSO₄ is a greater problem. Nevertheless, BaSO₄ and SrSO₄ scale is difficult to re-dissolve once precipitated. Hence, overdosing of sulphuric acid must be avoided. Acidification, however, has several limitations:

• Low pH increases fouling by natural organic matter (NOM) such as humic acids.

• Low pH lowers permeate quality due to higher TDS of feed water and increase in silica and carbonic acid.

• Permeate TDS increases when using hydrochloric acid because added chloride has a lower rejection than sulphate. Hence, acid treatment is usually used for carbonates and phosphates scale prevention.

Anti-scalant threshold treatment

Scale inhibitors or anti-scalants (A/S) are generally organic compounds containing sulphonate, phosphonate, or carboxylic acid functional groups and chelating agents such as carbon, alum and zeolites that sequester and neutralise a particular ion that may be formed. The majority of scale inhibitors can be classified as “threshold inhibitors.” In addition, chelating agents such as EDTA (tetra sodium salt of ethylene diamine tetra acetic acid) are used to control hardness (at pH > 6.0) and metallic ion deposits. Anti-scalants prevent mineral scaling by getting absorbed on the scale forming salt crystals thereby preventing the attraction of the supersaturated salt to the crystal surfaces. Since anti-scalants inhibit the growth of crystal, it does not grow to a size or a concentration large enough to precipitate out of the suspension. Many scale inhibitors also contain dispersants that keep the precipitates suspended in solution [14,40,41].

Anti-scalants are preferred to IX softening when the feed water hardness is less than 100 ppm because of cost and ease of operation. Although anti-scalants are very effective in preventing carbonate and sulphate scaling of membranes, they do not prevent scaling from occurring; rather they delay the formation of large crystals that form scales. The effectiveness of scale inhibition is approximately 30 min. A RO/NF system that uses anti-scalants must be designed with automatic flush cycle after shut down to prevent scaling by concentrated salts in the feed-reject channel above the membrane surface.

Anti-scalants can be used alone, but are almost always used with acid feeds. When acid is used with an A/S, the LSI value (of the reject stream) of 1.0 is acceptable (see Table 2.6), although some A/S manufacturers claim an LSI of 2.7 is acceptable when using their recommended product [42]. The A/S dosage ranges between 2 and 10 ppm depending on the scale-forming potential of the RO feed water, product water recovery and manufacturer’s recommendations. The main advantage of a higher LSI value is that the RO/NF system can be operated at higher recovery resulting in lower operating costs. From a process point of view, higher recovery results in higher salt concentration in the feed-reject channel that in turn results in sparingly soluble salts exceeding their solubility limits faster. Hence, high recovery is only an option when operating a second-pass RO unit or when the feed water is pure.

Although the solubility of silica is ~ 120 mg/l at neutral pH and 25°C, the solubility increases with temperature and at pH > 9.0 (also at pH < 6.0). For normal operation, silica solubility can be increased to 300 mg/l in the presence of a dendrimer A/S [42]. Dendrimers are highly branched polymers. Dendrimer-based anti-scalants do not contain phosphates and do not foul membranes even when the A/S concentration exceeds 100 ppm. Also, the membrane manufacturer’s recommendation the following saturation limits for other sparingly soluble salts when anti-scalants are used: BaSO₄ = 6000%, SrSO₄ = 800% and CaSO₄ = 230%.

Sodiumhexametaphosphate (SHMP) is a threshold agent derived from the dehydration of orthophosphoric acid or its sodium salt. It is used to inhibit the formation of calcium carbonate and metallic sulphate scale. It is mostly widely used because it offers good inhibition at a low cost. Depending on the concentration of calcium and sulphate, and depending on the CF (Equation 2.31), the dosage is in the range of 2–5 ppm. SHMP can prevent calcium sulphate precipitation up to 150% of the saturation limit. Organophosphonates are an improvement over SHMP since they are more resistant to hydrolysis, but are more expensive. They offer scale inhibition and dispersion ability similar to SHMP [14,40,41].

Polyacrylic acids (PAA) are good at both scale inhibition and dispersion, and are more effective than SHMP. Polyacrylic acids with high molecular weight distribution show the best dispersion ability at the cost of scale inhibition ability. However, precipitation may occur with cationic polyelectrolytes or multivalent cations such as aluminium or iron, resulting in fouling the membranes. Blend inhibitors are a combination of low (2000–5000 Da) and high molecular weight (6000–25,000 Da) of PAA or a blend of low molecular weight PAA and organophosphonates, giving excellent dispersive and inhibitor performance.

Anti-scalant treatment can, however, contribute to fouling largely due to either under-dosing or over-dosing; the former can result in scaling while the latter can lead to fouling [37,41]. Overdosing can also lead to biofouling and can result in complexes formed with hardness ions (since assimilable organic carbon content is a food source for microorganisms).

Basic controls

The basic level of control routinely used in the chemical process industries involves sequencing operations such as manipulating valves or starting/stopping pumps, instrument verification, data acquisition, on-line maintenance and fail-safe shut down procedures. The next level of computer control involves process control of parameters such as flow, pressure and temperature. RO/NF systems require both levels [43].

The simplest form of RO/NF system contrail entails the adjustment of only two variables: membrane inlet pressure and reject-flow rate, as shown in Figure 2.20. These functions affect the quantity and quality of RO permeate and indirectly potential for membrane fouling. The reject-flow control valve must be throttled in conjunction with the inlet-pressure control valve to achieve the desired productivity or product water recovery, i.e. yield.

Several more complex valve control sequences are illustrated in Figure 2.21. In a two-pass RO system (Figure 2.21a), the first-pass RO inlet-pressure and reject-flow control valves are adjusted simultaneously with the second-pass RO reject-flow control valve to achieve the desired recovery. In the case of a two-pass RO system (Figure 2.21b), where the feed to the secondary RO unit is re-pressurised, the secondary membrane unit inlet-pressure control valve is also adjusted. In order to reduce the size of the waste stream and increase overall product water recovery, the reject stream from the primary RO unit is sometimes further processed in a brine RO unit (Figure 2.21c). In such setups, the primary RO inlet-pressure and reject-flow control valves, the brine RO reject-flow control valve and the secondary RO reject-flow control valve are all adjusted simultaneously to achieve the desired recovery.

RO system performance (%recovery and %rejection) for a given membrane is a function of the operating conditions. Hence, the control system is designed to run the system at the membrane inlet pressure and recovery specified by the system manufacturer. Membrane performance, however, declines with time due to changes in the material properties of the polymer, so that control valve settings periodically require adjustments – for instance, raising the membrane inlet pressure. In addition, the control system may be used to monitor RO/NF pump inlet and outlet pressures, membrane reject and permeate pressures, membrane array pressure drop (ΔP), feed water temperature, pH, conductivity, chlorine level and permeate conductivity. Continuous monitoring of RO/NF performance, alarms, shutdown regime, data logging and performance trending are essential; for example, high ΔP across a membrane stage or an array is a good indicator of membrane fouling.

When specifying the high-pressure feed pump, membrane manufacturers assume that membrane flux will decline by about 20% in 3 years. The pump is, therefore, designed to provide feed pressure corresponding to the initial membrane performance and to compensate for expected flux decline. In the case of centrifugal pumps, the pump selected is oversized, and during operation the feed pressure is regulated by throttling (turning down the pump discharge valve) or by a variable frequency drive.

Control systems

A programmable logic controller (PLC)-based membrane system utilises remote input/output (I/O) functionality. The PLC communicates with various remote I/O control enclosures in the membrane plant, and performs all sequencing and inter-locking functions. In a fully automated system, independent (non-PLC) controllers such as PID (proportional plus integral plus derivative) or other tuning allow the operator to modulate control valves, e.g. feed water temperature, membrane inlet pressure and reject-flow rate [43]. The PID controllers also give the operator easy access for changing set point values, and supervise alarm conditions without the need for an operator interface screen. Independent PID controllers can allow the operator to run the system without a working PLC; if a single independent PID controller fails, only the particular parameter it was controlling cannot be adjusted.

Process control valves

Membrane feed-pressure control may be performed via two different techniques: proportional valve position or pump motor speed [43]. The valve solution is generally less expensive in initial capital but solid state control of pump motor speed reduces operating expense because electrical power savings can be realised over the life of the equipment. The long-term reliability of the motor speed solution is greater if done properly, but the process control is slightly better with the proportional control valve.

Control of the reject-flow rate can only be achieved with the use of a proportional valve. Since the type of valve can affect the performance of the system, the valve should be capable of reducing the pressure (up to an order of magnitude bar, if necessary) while maintaining a smooth linear response throughout the required range of flows. In seawater RO applications, energy recovery from the high-pressure reject stream (> 55 bar g psig) is an important economical requirement [44,45]. High-pressure drop control is, however, complex. Some systems may be designed with a fixed pressure drop, using an orifice plate, upstream of the reject control valve. The orifice plate lowers the amount of pressure that must be dropped across the control valve. Although this reduces the sensitivity of the valve, it enables the control-system output signal to operate over a greater range, thereby, increasing the accuracy of control.

The difficulty with controlling membrane feed-pressure and reject-flow rate simultaneously is that one affects the other. This gets more complicated with a two-pass system. If the membrane feed pressure is reduced, the reject-flow rate also will decrease if all other parameters (feed water temperature, quality and membrane condition) in the system remain constant. The danger, then, is that the whole system may oscillate if the PID controllers are not properly tuned. Tuning one controller for fast operation and the other for relatively slow operation is often the solution. Fast operation generally is best applied to the membrane feed-pressure controller because during start-up this parameter must be adjusted first. Manual control of these valves is, therefore, often preferred.

Incorporating the controller start-up output condition is also necessary to prevent the controller from drifting towards 0% or 100% – i.e. fully closed or fully open, respectively – during idle periods; otherwise, if the RO/NF pump starts when the membrane feed pressure or the reject-flow control valve is closed, a sudden spike in initial pump discharge pressure can result in equipment damage. Conversely, if the membrane feed-pressure valve is fully open during startup, the membrane elements are exposed to excessive pressure, which may cause compaction of the membranes in the leading elements.

Product quality controls

Feed water parameters that affect membrane life adversely such as high pH, free chlorine and high temperature are shown in Figure 2.22. Often, automatic on/off valves are provided to divert the feed or the permeate stream to drain when the stream quality is not in the desirable range. For example, if the feed water conductivity is much higher than the system is designed for, membrane scaling may result. A much higher-than-design permeate conductivity, on the other hand, indicates membrane scaling, membrane element integrity, or membrane damage. Hence, a low rejection alarm is often provided. Similarly, when the permeate flow rate exceeds the design point (or the reject flow is less than the design value), the potential for membrane scaling is substantially increased. A high recovery alarm is also included to alert the operator to reset the reject-flow control valve setting or shut down the unit.

High-pressure drop across a membrane array is a strong indicator of membrane fouling due to constriction in the reject/concentrate flow channel above the membrane, and calls for membrane cleaning. Usually RO/NF plant designs incorporate an automatic 5-min flush cycle preferably with RO permeate before shutdown to protect the membranes from (a) scaling by sparingly soluble salts in the salt concentrated stagnant reject channel, and (b) biological fouling. Often the flush cycle is programmed in the PLC to run at a preset interval such as every 4 h. During flushing the high-pressure RO pump is off. Flushing minimises the likelihood of scaling and fouling but is not a substitute for feed water pre-treatment.

Safety features

RO control systems provide safety features for operating high-pressure RO feed water pumps. Pressures at pump inlet and discharge, membrane inlet, product output and reject output are, therefore, monitored by pressure switches and transmitters as shown in Figures 2.22 and 2.23. To ensure long-term safe operation of the equipment, pressure switches shut down the pump during low inlet pressure and high discharge pressure conditions. The high-pressure RO pump must be run in auto mode to ensure it shuts down automatically during an emergency. A relief valve is provided in the permeate line to protect membrane elements from damage due to any back-pressure downstream. Further, in a fully automated RO system, the membrane inlet pressure and the reject-flow control valves are of the air-to-open type. Air-to-close valves should not be used because it would result in excessive membrane pressure during start-up.

The following general process conditions are taken into consideration when designing a control system to ensure safe and reliable operation [14]:

• Product water back-pressure greater than the system pressure at any time will damage the RO membranes.

• Operate the RO unit at the lowest system pressure that produces the design flow rate and salt rejection.

• The initial minimum salt rejection is based on chloride ions and applies to each individual membrane element. The total %rejection may be lower depending on the array. Membrane deterioration may result in salt passage to double within 3 years.