on Residuals Management for Desalting Membranes

Transcription

1 filtration BY AWWA MEMBRANE RESIDUALS MANAGEMENT SUBCOMMITTEE Committee Report: CURRENT PERSPECTIVES on Residuals Management for Desalting Membranes WATER SUPPLIERS RECOGNIZE THE CAPABILITY OF DESALTING MEMBRANES TO COVERT POOR- QUALITY SUPPLIES INTO POTABLE WATER, BUT MANY UTILITY MANAGERS AND ENGINEERS DO NOT KNOW HOW TO MANAGE THE RESIDUALS. T his article is the second part of a committee report on residuals management for membrane processes. The first part, Committee Report: Residuals Management for Low-pressure Membranes (AWWA, 2003), presented information about residuals management for microfiltration (MF) and ultrafiltration (UF). This article addresses residuals management for desalting membrane processes, which include the high-pressure (driven) membrane processes; nanofiltration (NF) and reverse osmosis (RO) processes; and the electrically driven process, electrodialysis (ED). 1 (For the purposes of this article, NF, RO, and ED are identified as desalting membrane processes, as they are used to remove dissolved contaminants, both inorganic and organic, from water supplies. 2 ) RO and ED were first used for potable water treatment in the 1960s, and NF followed in the mid-1980s. As of 2003, more than 230 desalting facilities (187 RO, 29 NF, and 18 ED) with a capacity of more than 25,000 gpd (94,625 L/d) operate in the United States (Mickley, 2004a). The use of desalting processes will expand as water suppliers recognize the capability of these technologies to convert saline or poor-quality water supplies into potable water. With the continuing installation of desalting membrane plants, residuals management is a growing issue. The main residual from desalting membrane systems is concentrate, a wastewater with elevated levels of dissolved solids and other contaminants from the source water. Management of residuals from desalting membrane processes is challenging because of the high level of total dissolved solids (TDS) and the large volumes (relative to conventional treatment processes). Some desalting membrane projects have been cancelled or delayed because of concerns about residuals disposal. Many regulatory agencies, utilities, and design engineers are unfamiliar with the characteristics, treatment, and disposal of desalting membrane residuals. This article responds to these challenges and offers a current perspective on management of residuals from desalting membrane processes. COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

2 TABLE 1 Treatment applications for desalting membrane processes Application Reverse Osmosis Nanofiltration ED & EDR Total dissolved solids reduction Seawater desalination E N/A N/A Brackish water desalting E F E Brackish water desalting (high silica levels) G F E Inorganic ion removal Fluoride removal E F E Hardness removal (softening, e.g., calcium, magnesium) E E E Ferrous iron removal E E G E, N/A Nutrient removal Nitrate and nitrite E F G Phosphorus E G E Radionuclides removal E G V Other SWDA-regulated inorganic chemicals (e.g., arsenic) E F V Taste and odor compounds E (V) G (V) P Organic compound removal Total organic carbon Disinfection by-product precursors E E P Color (organic type) removal E E P Synthetic organic chemicals removal (e.g., pesticides) E G (V) P Pathogen removal Bacteria, viruses, Giardia, Cryptosporidium E* E* N/A E excellent, ED electrodialysis, EDR electrodialysis reversal, F fair, G good, N/A typically not used because of economics or inadequate removal capability, P poor or no removal, SWDA Solid Waste Disposal Act, V varies depending on contaminant *For integral systems (absence of membrane imperfections and mechanical leaks; e.g., glue line and O-ring seal leaks), permeate disinfection should be provided. Three AWWA committees worked together to develop this report the Residuals Management Research Committee, the Membrane Process Committee, and the Water Treatment Plant Residuals Committee forming the Subcommittee on Membrane Residuals Management. The committee report on residuals management for membrane processes is a working document because desalting membrane processes and their application to drinking water treatment continue to evolve. The Subcommittee on Membrane Residuals Management intends to provide periodic updates as changes occur in the character and management of residuals. DESALTING MEMBRANE PROCESSES HAVE COMMON FUNCTIONS AND CHARACTERISTICS Membrane processes are characterized in two ways. Desalting membrane processes are often characterized by the driving force that separates impurities from water and by the type of membrane (porous or nonporous) used to effect contaminant separation. Pressure-driven membrane processes designed for particle removal, commonly called low-pressure processes, include MF and UF. Desalting membrane processes include two classes: pressure-driven and electrically driven. Pressure-driven processes include NF and RO, which use semipermeable membranes designed to separate dissolved substances from water using a pressure differential. The removal characteristics and operating pressures of NF and RO are summarized elsewhere (AWWA, 2003). The electrically driven processes include ED and its variant, electrodialysis reversal (EDR). ED/EDR membranes are ion exchange membranes that use an electrical potential to separate impurities from water. Table 1 lists the relative performance of each process in treating drinking water. ED/EDR are electrically driven processes. The ED/EDR processes use an electrical potential to force dissolved ions through ion exchange membranes that are highly impermeable to water. The water is purified by the movement of contaminants not the movement of water across the membrane. The ED process (Figure 1) places alternating pairs of cation (+) and anion ( ) membranes between positively and negatively charged electrodes. When a voltage is applied, a current causes cations to move in the direction of the negatively charged electrode (cathode). The cations move through the cation membrane but are restrained by the anion membrane. The anions move in the direction of the positively charged electrode (anode) through the anion membrane, but are restrained by the cation membrane. Water flows in all channels, and the flow from channels where con- 74 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

3 taminants are removed becomes the desalted product stream. Flow from channels to which salts are transferred becomes the concentrate. Feed pressures for ED/EDR are typically between 50 and 70 psi (345 and 483 kpa) to overcome head losses in the ED/EDR membrane modules or stacks. With ED, the direction of electrical charge is always the same. With EDR, the direction of charge is reversed several times hourly by reversing the electrical polarity, which changes the direction of ion movement through the membranes and electrically flushes fouling ions from the membrane surfaces to control the buildup of scale, improve recovery, and reduce the need for scale-inhibiting chemicals. The ED/EDR process is designed for desalting brackish water supplies and is not applicable to the removal of noncharged solutes, such as silica, or particles and pathogens. It is inefficient at removing dissolved organics, which are weakly charged. NF and RO are pressure-driven processes. NF and RO membrane processes are usually classified based on their salt rejection capabilities, as shown in Table 2. When NF and RO membrane processes are used for the removal of synthetic organics, particularly pesticides and endocrine disruptors, molecular-weight cutoff (MWCO) is a more appropriate means of classification. RO membrane MWCO is typically D, and NF membrane MWCO usually ranges from 200 to 2,000 D. NF and RO membranes can also be classified based on operating pressure. NF typically operates at lower feed pressures than does RO because of inherently greater permeability and the lower osmotic pressures of its feedwaters. RO feed pressures vary greatly, reflecting the wide ranges in membrane permeability and in feedwater TDS level and associated osmotic pressure. High TDS feedwaters, such as seawater, have an osmotic pressure of approximately 400 psi (2,758 kpa) and require feed FIGURE 1 Feedwater Source: Ionics Inc. Schematic of ED process showing cell pair arrangement and product and concentrate flow ANODE (+) Cl SO 4 Anion Membrane Cation Membrane Ca ++ SO 4 Anion Membrane Cation Membrane Ca ++ CATHODE ( ) pressures >800 psi (5,516 kpa). Some ultralow-pressure RO membranes operate at psi (517 1,034 kpa) on low TDS feedwater, which is similar to NF membranes. Because they are nonporous, NF and RO membranes effectively retain particles that can foul the membrane surface. In addition, particles can clog the NF/RO module flow channels if total suspended solids (TSS) and turbidity in the feedwater are too high. Operation of NF and RO systems on feedwaters containing >1 mg/l TSS, 1 ntu turbidity, or 5.0 silt density index (SDI) will significantly degrade performance and result in the need for frequent chemical cleanings. To avoid poor performance, NF and RO feedwater should have <0.2 ntu turbidity and <3.0 SDI. NF/RO have similar system configurations. NF and RO use flat-sheet membranes arranged in spiral-wound modules (Figure 2). A series of membrane modules is usually housed in horizontal pressure vessels, each containing six to eight modules. Multiple pressure vessels are staged, both in parallel and in series (Figure 3), to achieve the desired capacity and to maximize permeate recovery. In this arrangement, each module converts Product Water H 2 O Concentrate Cl H 2 O, CaSO 4, NaCl Na + Na + Product Water H 2 O Concentrate Cl H 2 O, CaSO 4, NaCl CaSO 4 calcium sulfate, ED electrodialysis, H 2 O water, NaCl sodium chloride, SO 4 sulfate ion Na + Cl Na + between 10 and 20% of its feedwater to permeate. Typical recoveries are 80 90% for NF plants and 65 85% for brackish water RO plants, depending on the concentration of scaling species. Those of seawater RO plants range from 40 to 60%. The basic components of a pressure-driven membrane system (Figure 3) include a pump, pressure vessels containing membrane modules, and a means to control the pressure applied to the pressure vessel and regulate permeate flow and recovery. ED/EDR system configurations differ. An ED system pumps feedwater through the membrane stacks that contain hundreds of anion/cation membrane pairs and associated feed channels, together with cathodes and anodes. A pump recycles a portion of the ED concentrate to the concentrate channels to maintain appropriate fluid velocity. In an EDR system, a series of electrically actuated valves are used to switch permeate and concentrate streams within the stacks in alignment with the electrode polarity reversal. Fouling causes lost productivity. Over time, membrane systems lose productivity from the accumulation of foulants on the membrane surface or COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

4 TABLE 2 FIGURE 2 Feedwater Feedwater Salt (ion) rejection capabilities of NF and RO membranes Conceptual rendition of spiral-wound NF/RO element Spiral-wound RO Module Source: Ionics Inc. NF nanofiltration, RO reverse osmosis within the feed channels. Fouling typically consists of a combination of the following, depending on the feedwater characteristics, membrane type, and system operating conditions: (1) inorganic colloidal or suspended particles, (2) inorganic scale from the precipitation of supersaturated salts, (3) organic materials, and (4) biological materials. Cleaning restores system performance. Desalting membrane systems typically add chemicals continuously to the feedwater to inhibit scale formation. Such chemicals include acid to increase carbonate solubility and antiscalants and dispersants that inhibit formation of precipitates. Chemical cleaning or clean-in-place (CIP) cleaning restores system performance through the use of appropriate low-ph, high-ph, and surfactant/detergent solutions, either individually or in combination, to dissolve, loosen, and remove membrane foulants. These solutions are typically prepared in batches in the CIP tanks and recirculated through the membrane train. The types of solutions Permeate Ion Rejection Membrane Process Monovalent % Divalent % Nanofiltration Reverse osmosis NF nanofiltration, RO reverse osmosis Permeate Product spacer RO membrane Feed spacer RO membrane Concentrate Density of dots: highest density concentrate, medium density feedwater, lowest density permeate and frequency of cleaning depend on the types of foulants, membrane characteristics, and cleaning goals. CIP performance is improved by using warm solutions and extended contact times and is typically performed every three to 12 months depending on the feedwater quality and system operating characteristics. RESIDUALS INCLUDE CONCENTRATE AND SPENT CLEANING SOLUTIONS Residuals from NF, RO, and ED systems result from the separation and CIP processes and include two types: concentrate, which contains dissolved and particulate contaminants removed from the feedwater and may contain chemicals from pre-treatment facilities and spent cleaning solutions, which contain high concentrations of the cleaning chemicals, plus feedwater contaminants removed during cleaning. Concentrate is produced continuously; spent cleaning solutions are generated intermittently (typically every three to 12 months). In addition, the volume of concentrate is much greater than the volume of spent chemical solution. For purposes of this article, the characteristics of the concentrate from NF, RO, and ED/EDR systems are considered similar, and their management is presented together. ED/EDR systems also produce a specialized waste stream of limited flow called electrode waste, which contains significant levels of hydrogen and chlorine gases that are typically stripped from the electrode waste stream using a degasifier (which is part of the EDR system). However, free chlorine will be present in the poststripper waste stream and at a much diluted concentration in the ED/EDR concentrate when the stripped electrode waste is mixed with the concentrate flow. Chemical cleaning residuals, however, are different for ED/EDR systems than those for NF and RO systems: NF and RO systems are typically cleaned with acid (mineral or citric) to remove inorganic foulants and alkaline solutions (typically caustic soda, often in combination with detergents/surfactants and sometimes chelating agents) to remove biofilms and organic foulants. ED/EDR systems are typically cleaned with concentrated hydrochloric acid and sodium chloride solutions (AWWA, 1995). ED/EDR systems are sometimes cleaned with chlorine solutions to remove biofilms and organics. NF and RO systems cannot use chlorine as it damages their membranes. Concentrate is related to feedwater quality. Because desalting membrane processes produce a high-purity stream by rejecting high percentages of contaminants in the feed stream, the the concentrate stream is primarily a more concentrated version of feedwater. As such, the characteristics of the concentrate are directly related to feedwater quality. For NF and RO, these contaminants and other constituents in the feedwater will be rejected by the membrane and will become part of the concentrate 76 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

5 FIGURE 3 A Basic components Feedwater Feed Schematics of NF/RO systems Feed pump Concentrate recycle (optional) B Two-stage array to achieve high recovery NF nanofiltration, RO reverse osmosis stream. Contaminant concentrations are typically four to 10 times the feedwater concentration and depend on the rejection characteristics of the membrane and the water recovery. If pretreatment is used, the feedwater to the desalting membranes will have lower levels of certain constituents and particles. However, the pretreatment may increase inorganic ions, such as sulfate, iron, and aluminum, if coagulants are used and may increase residual organics if polymer or sulfuric acid is used. Concentrate contains low concentrations of particles, typically <10 mg/l TSS, because of low NF/RO feedwater concentration of particles. For ED, ionized contaminants in the feedwater will be concentrated in the concentrate stream in the manner described for NF and RO. In addition, the ED concentrate will contain free chlorine from the generation of chlorine gas at the electrode. Concentrate quantity is a function of the amount of feedwater purified (or converted to permeate), as defined by the product water recovery. Table 3 shows typical recoveries for the various types of water sources treated by desalting membrane processes and associated concentrate quantity as a percent of feed. NF and brackish water RO systems operate at recoveries of 80 90% and 65 85%, respectively. Seawater RO systems operate at lower recoveries, typically 40 60%. NF and brackish RO residuals differ somewhat because NF salt rejection is lower. Therefore, for a given feed and recovery, NF concentrate is less saline than RO concentrate. Further, NF provides low rejection of monovalent ions (e.g., sodium and chloride) compared with multivalent ions (e.g., calcium and sulfate). Consequently, NF concentrate has a higher ratio of multivalent ions to monovalent ions than does feedwater. Lastly, because RO systems typically treat higher-salinity waters, the TDS levels in the RO concentrate are much higher, especially for seawater. As mentioned earlier, the concentration of contaminants is proportional to water recovery and membrane rejection, with concentration increasing as water recovery increases. Typical source water quality, water recovery, and residuals concentrations are included in Table 3, where the values represent the maximum concentrations for residuals streams based on a simplifying assumption that all of the salts are rejected (100% salt rejection). However, desalting membrane salt rejection is less than 100%; thus, the concentrate values in Table 3 are conservative. For high-quality source waters, such as surface water and groundwater with lower TDS, lower salt rejection membranes are commonly used. In these cases, contaminant levels in the concentrate stream are lower than shown in Table 3. The amount of particles or TSS in desalting membrane concentrates is low, typically <10 mg/l, because of the extensive pretreatment required to control fouling. The acid and scale inhibitor added to the membrane feedwater to control mineral precipitation will be Membrane module Concentrate Concentrate valve Permeate Concentrate Permeate concentrated in the desalting membrane concentrate. Scale-inhibitor levels in the concentrate are typically <20 30 mg/l and consist of phosphates or organic polymers (such as polyacrylates or dendrimers). For simplicity, the residuals stream concentration can be estimated based on the quality of the feedwater and a concentration factor related to the water recovery, as shown in Eqs 1 and 2. These equations are based on 100% rejection and therefore provide a conservative (high) estimate of residuals concentrations. Cr = Cf CF (1) in which Cr is the residuals (concentrate) concentration, Cf is the feed concentration, and CF is the concentration factor. CF = 1/(1 R) (2) in which R is the water recovery (expressed as a fraction or decimal). 3 For example, recoveries of 50, 75, and 90% COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

6 FIGURE 4 Decision tree for disposal of desalting membrane residuals SPENT CIP SOLUTIONS Neutralize ph to acceptable range Neutralize Cl 2 as needed (ED/EDR only) Levels of TSS, organic compounds and TDS acceptable? No Remove contaminants? No Yes Yes Discharge to sanitary sewer (obtain IPP permit and pretreat as needed) or haul off-site (subject to other regulations) Spent CID Solution Issues Handling and disposal of concentrated solids, either discharge to a receiving water or disposal on land; disinfection and disposal of concentrated pathogens, as required RESIDUALS Discharge to receiving stream as permitted Treated water Solids and concentrate Treatment Land disposal; sanitary sewer; haul off-site would correspond to concentration factors of 2, 4, and 10, respectively. CIP residuals reflect the cleaning solution and foulants removed. Cleaning solutions were developed to address specific contaminants that cause fouling and loss of performance, and CIP residuals reflect the chemical characteristics of both the spent cleaning solution and material removed from the membrane system during CIP. Reactions with foulants will tend to raise the ph of acid solutions and lower that of basic ones. Types of solutions. Table 4 lists the typical cleaning formulations developed to remove various types of CONCENTRATE If needed, adjust ph to acceptable range Neutralize chlorine (ED/EDR only) as required Concentrate Issues On-site disposal: remove water and land disposal of precipitated solids Discharge to surface water: obtain NPDES permit; treat concentrate as needed (e.g., H 2 S,DO) Deep well disposal; obtain needed permits Discharge to sanitary sewer; obtain IPP permit; pretreatment as needed Volume and discharge rate of residuals for disposal; flow conditions for NPDES permit (e.g., 10-year low flows, tidal conditions, average flow); sanitary sewer system capacity; climate and availability of land disposal CIP clean-in-place, Cl 2 chlorine, NPDES National Pollutant Discharge Elimination System, DO dissolved oxygen, ED electrodialysis, EDR electrodialysis reversal, H 2 S hydrogen sulfide, IPP Industrial Pretreatment Program, TDS total dissolved solids, TSS total suspended solids foulants. The list reflects formulations and chemicals found to be effective by one NF/RO membrane supplier. Cleaning is more art than science, and as such, each membrane supplier has developed specific solutions or formulations based on laboratory and field experience with its product. Many companies catering to the membrane industry have developed proprietary cleaners to optimize cleaning efficiency for specific fouling situations; however, listing such products is beyond the scope of this article. Solution volumes. The volume of cleaning solution and the amount of waste solution are functions of the following: the membrane process and module configuration/system characteristics, the size and capacity of the membrane train or the portion of it to be cleaned, the number of membrane modules to be cleaned at the same time or in a given batch, the system design conditions (flux, recovery, staging, and manifolding), the cleaning system design (piping diameter and lengths), and the frequency of cleaning. Typically, the cleaning solution volumes generated during a CIP of NF and RO are approximately 3 gal/100 sq ft (1.2 L/m 2 ). (This volume does not include rinse water volumes.) Typical cleaning solution volume is estimated by adding the total empty vessel volume and pipe volume. Spent cleaning solutions, which may be diluted with rinse water (feed or permeate), can contain detergents, surfactants, acid, caustic, or other chemicals used to remove foulants from the membrane system. The spent cleaning solution may need to be treated before disposal. Typically, the spent cleaning solution volume is an extremely small percentage of the treated flow (less than 0.1%). RESIDUALS MANAGEMENT FOR DESALTING MEMBRANES IS DEFINED BY US REGULATIONS Regulations affect residual disposal. Several regulatory programs affect the disposal of desalting membrane residuals, including the Clean Water Act (CWA; PL as amended), the Underground Injection Control (UIC) Program, the Resource Recovery and Conservation Act (RCRA) for any solid waste residuals, and regulations that protect groundwater. Disposal options for desalting membrane residuals and associated regulatory and permitting agencies are listed as follows. Disposal to surface water discharge requires a National Pollutant Discharge Elimination System (NPDES) permit. 78 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

7 TABLE 3 Typical desalting membrane system design parameters Fresh Brackish Parameter Surface Water Groundwater Groundwater Seawater Feedwater total dissolved solids mg/l ,000 30,000 40,000 Water recovery % of feed Concentrate quantity % of feed Concentrate total dissolved solids 1,330 2,660 (85%) 2,660 3,330 (85%) 2,000 40,000 (75%) 60,000 80,000 (50%) (at example recovery) mg/l (%) Concentration factor* *Ratio of total dissolved solids in concentrate to total dissolved solids in feed, assuming 100% salt rejection Sewer discharge requires a permit issued by the local sewer agency to meet its sewer ordinance and the CWA Industrial Pretreatment Program (IPP) requirements, as stipulated in the agency s NPDES permit. Concentrate disposal by land application (e.g., percolation pond, rapid infiltration basin, landscape and crop irrigation) must comply with federal and state regulations to protect groundwater, public health, and crops/vegetation. Land application requires a permit from state agencies. Construction of evaporation ponds is subject to state requirements for pond construction. Concentrate disposal by subsurface injection (or deep well) is regulated by the UIC program of the Safe Drinking Water Act. The related construction, monitoring, and other permits are issued and enforced by the US Environmental Protection Agency (USEPA) region or state agency that has primacy. RCRA regulates the disposal of solids, such as precipitated salts and sludges; if such solid wastes do not pass the Toxicity Characteristic Leaching Procedure (TCLP) test, they are considered a hazardous waste (Type D) and must be handled accordingly. The most important regulation pertaining to the disposal of desalting residuals are those related to the CWA, including the NPDES program. Under the CWA, desalting membrane residuals are regulated as industrial wastes. However, USEPA has not established specific regulations for the disposal of water treatment residuals, including membrane residuals. For surface water discharge, an NPDES permit is required pursuant to the CWA its antidegradation policy prevents the relaxation of discharge limits for contaminants specified in an NPDES permit, particularly if the receiving water is designated as sensitive or impaired. If a wastewater treatment plant (WWTP) currently has a TDS discharge limit, combining high TDS concentrate from NF/RO with the existing discharge may not be allowed. As the CWA requires, the USEPA has established minimum requirements, including specific surface water quality criteria. Any state that has USEPA-delegated authority (termed primacy ) to administer the NPDES permit program must establish water quality standards for the protection of the designated uses of the water body at least as stringent as the federal limits. Therefore, a surface water discharge of desalting membrane residuals is regulated by an NPDES permit, which meets the CWA requirements. The USEPA has not identified the water industry as a potential candidate for effluent guideline development. However, because of public comments and preliminary research conducted by the USEPA, which indicates that a significant number of water treatment facilities have nontrivial discharges, the USEPA has decided to complete an effluent guideline rule-making for the water industry within three years. Although the USEPA has not established effluent guidelines to date, the USEPA has placed water treatment facilities in the industrial sector. This classification can result in a negative public perception and make permitting more difficult. Better data on concentrate characteristics are needed to educate regulators and change public perception, which should simplify the permitting process. CURRENT DISPOSAL METHODS REQUIRE CAREFUL EVALUATION Successful disposal of NF, RO, or ED residuals is critical to implementing a desalting membrane system and therefore requires careful evaluation of available alternatives, regulations, capital and operating costs, and site-specific conditions. If residuals disposal is not possible or economical, a desalting membrane process will not be used. The evaluation of alternatives should include assessment of the longterm viability of each method. Regulatory changes, in particular, could adversely affect the continued operation of a desalting plant. Disposal of desalting membrane concentrate can be particularly challenging because of the large volume and high TDS concentration. These characteristics can limit the options that can be permitted and that are cost-effective, particularly in regions where surface water discharge is unavailable. A decision tree for considering residuals disposal alternatives COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

8 FIGURE 5 Number of Desalting Membrane Plants (through 2001) Geographical distribution of US desalting membrane plants* 114 Source: Mickley, 2004a FIGURE 6 Number of Desalting Membrane Plants (through 2001) for desalting membrane processes is shown in Figure 4. Since the first municipal RO facility was installed in 1976 in Cape Coral, Fla., using surface water discharge for residuals disposal, the locations and disposal methods for desalting facilities have expanded greatly. According to a recent survey (Mickley, 2004a) of approximately 95% of the municipal membrane desalting facilities, most of the 234 facilities are located in Florida, California, Texas, and Illinois (Figure 5). Surface water discharge is the most common method of concentrate 33 Florida California Texas Illinois Disposal methods for membrane desalting plants 106 Surface Water Source: Mickley, 2004a 63 Sewer 31 Subsurface Injection 19 Land Evaporation Pond Recycle Reuse disposal, followed by sanitary sewer discharge, and deep well injection (Figure 6). The membrane facility s capacity has a bearing on the selected disposal method (Figure 7). Surface water discharge is the most common method of concentrate disposal for <1-mgd (3.8-ML/d) capacity, while >6-mgd (22.7-ML/d) plants are almost evenly split between subsurface injection and surface water discharge. Nearly all plants using subsurface injection are in Florida, which has geology favorable to this method. Concentrate disposal includes many alternatives. Current disposal practices for concentrate from desalting membrane treatment plants include: surface water discharge, sewer, subsurface injection, evaporation pond, and land application. Surface water discharge. Surface water discharge is the most common disposal practice. This reflects the proximity of a large number of desalting plants to a river, lagoon, or ocean. Disposal costs are low if the length of discharge pipeline is reasonable and the discharge meets NPDES permit requirements. Permit limits may include TSS, TDS, salinity, or specific contaminants, such as nutrients (e.g., nitrogen and phosphorus compounds), arsenic, or barium. Some contaminants, such as arsenic, are priority pollutants, which can increase the complexity of obtaining a permit. Rarely can a high-salinity concentrate be discharged into a lowsalinity receiving water if it results in the salinity at the discharge point increasing by 10% or more compared with the upstream receiving water. Some facilities address high salinity by diluting the concentrate with surface water or groundwater, WWTP effluent, or cooling water. An NPDES permit may include limits on whole effluent toxicity (WET), a bioassay performed according to USEPA-approved protocols to assess acute, chronic, and bioaccumulative toxicity to receiving water biota. The bioassays use approved pollutant-sensitive species. Some groundwater-fed RO systems have produced concentrates that fail WET limits tests (Mickley, 2000). Most such cases in Florida were associated with high calcium levels, and some were complicated by toxicity from high fluoride levels. Toxicity caused by high levels of major ions is a correctable chemical imbalance, as opposed to toxic contamination from heavy metals or pesticides. For this reason Florida has exceptions for major ion toxicity when it is the only toxicity present in a concentrate. Concentrates are well-suited for surface water discharge because they contain increased concentrations of naturally occurring constituents (in 80 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

9 the RO feed), which typically have no toxicity. NPDES permitting requirements vary by state and site-specific conditions. It is imperative that the local NPDES permitting agency be contacted early to confirm acceptability of a proposed surface water discharge. For membrane facilities treating groundwater that is brackish or contains arsenic or radionuclides, the elevated TDS or contaminant concentrations may not meet discharge requirements for fresh surface waters within a reasonably sized mixing zone. In these cases, desalting membrane plants may be limited to areas with available brackish surface receiving streams or may need to remove the toxic contaminants before discharge. Dissolved gases or lack of oxygen can also be a concern for concentrate disposal. Concentrates from the treatment of many groundwaters have very low dissolved oxygen (DO) levels. Before discharge, DO levels must be increased to avoid negative effects on receiving stream biota. As with DO, if the groundwater contains hydrogen sulfide, hydrogen sulfide in the concentrate must be suitably reduced before discharge. ED concentrate typically contains free chlorine, which must be neutralized by using a reducing agent such as sodium bisulfite before discharge to the receiving stream. Sewer. Discharge to a sanitary sewer requires a permit from the local sewage agency, which may impose limitations to protect sewers and treatment plant infrastructure, the treatment process, and final effluent and biosolids quality. Sanitary sewer discharge of a small volume of concentrate usually represents a low-cost disposal method with limited permitting requirements. The adequacy of the sewer system and WWTP capacity and NPDES permit limits must be addressed at the onset of the project. The local sewer agency may impose a one-time fee to purchase capacity for the concentrate discharge. In addition, operating costs will be billed based on volume and pollutant loads. FIGURE 7 Desalting Membrane Plants % Disposal methods by membrane desalting plant capacity mgd ( ML/d) 1 6 mgd ( ML/d) > 6 mgd (22.7 ML/d) Surface Water Sewer Source: Mickley, 2004a TABLE 4 Foulant Type Subsurface Injection Evaporation Pond Land Recycle Cleaning Solutions Reuse System Typical nanofiltration and reverse osmosis cleaning formulations* Inorganic salts 0.2% HCl 0.5% H 3 PO 4 2% citric acid Metal oxides 2% citric acid 1% Na 2 S 2 O 4 Inorganic colloids (silt) 0.1% NaOH, 0.05% Na dodecyl benzene sulfonate, ph 12 Silica (and metal silicates) Ammonium bifluoride 0.1% NaOH, 0.05% Na dodecyl benzene sulfonate, ph 12 Biofilms and organics Hypochlorite, hydrogen peroxide, 0.1% NaOH, 0.05% Na dodecyl benzene sulfonate, ph 12 1% sodium tripolyphosphate, 1% trisodium phosphate, 1% sodium EDTA Source: Cleaning Procedures for FILMTEC FT30 Membranes, Dow Chemical Co. (1994) EDTA ethylenediaminetetraacetic acid, HCl hydrochloric acid, H 3 PO 4 phosphoric acid, NaOH sodium hydroxide, Na 2 S 2 O 4 sodium hydrosulfite *Oxidants, including hypochlorite and hydrogen peroxide, cannot be used with chlorine-sensitive membranes, including polyamide nanofiltration and reverse osmosis flat-sheet membranes except under specific conditions (e.g., high ph). Barium sulfate, calcium carbonate, calcium sulfate Subsurface injection. Subsurface injection alternatives, such as deep well injection and boreholes, are regulated by the UIC Program. Regulatory considerations include the receiving aquifer s transmissivity and TDS, and the presence of a structurally isolating and confining layer between the receiving aquifer and any overlying Underground Source of Drinking Water (USDW) that is considered to be any water-bearing formation that has <10,000 mg/l TDS. Subsurface injection is infeasible in areas subject to earthquakes or with geologic faults that can provide a direct hydraulic connection between the receiving aquifer and an overlying potable aquifer. Well design should allow for testing injection well COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

10 FIGURE 8 Capital Cost Source: Mickley, 2004a Relative capital cost of concentrate management options Evaporation pond Flow Rate mgd (ML/d) integrity. Monitoring wells in proximity to the disposal well are typically also required to confirm that vertical movement of fluid has not occurred. The capital cost for subsurface injection is higher than for surface water disposal, sewer disposal, and land application in which these alternatives do not require long transmission pipelines. Disposal to subsurface injection is usually restricted to large volumes of concentrates when economies of scale make the disposal option more costeffective and when soil transmissivity is good. Although the UIC Program does not restrict the use of deep wells, geologic characteristics are not appropriate for deep well injection in many Brine concentrator Spray irrigation Surface water Deep well injection Sewer areas of the US. In addition, a backup means of disposal must be available for use during periodic maintenance and testing of the injection well. Evaporation pond. Solar evaporation is a viable alternative in relatively warm, dry climates with high evaporation rates, level terrain, and low land costs. Regulations typically require an impervious lining and monitoring wells. With few economies of scale, evaporation ponds are usually used for small volumes of concentrates. Although evaporation ponds typically are designed to accommodate concentrate for the projected life of the desalting facility, precipitation of salts is expected and must be incorporated into the depth requirements of the pond or provisions must be made for periodic removal and disposal or beneficial use of precipitated salts. In addition, the ultimate fate of the concentrated salts and potential future regulatory implications should be considered for any evaporation pond project. Land application. Land application can be a beneficial reuse of water when membrane plant residuals are used for spray irrigation of lawns, parks, golf courses, or crop land. Associated factors include the availability and cost of land, percolation rates, irrigation needs, water quality tolerance of target vegetation to salinity, and the ability to meet groundwater quality standards. An assessment of the compatibility with target vegetation should be conducted, including a review of the sodium adsorption ratio (SAR), trace metals uptake, and other vegetative and percolation factors. Regulations governing groundwater quality and protection of drinking water aquifers should be investigated as early as possible to confirm the acceptability of this alternative. Significant concerns may arise if the concentrate contains arsenic, nitrates, or other contaminants regulated in drinking water. If allowed, the concentrate may be diluted to meet groundwater standards. When salinity levels in the residuals are high, special salttolerant species (halophytes) could be considered for irrigation. Land application also includes the use of percolation ponds and rapid infiltration basins and generally is used only for small volumes of concentrates. These options are frequently limited by availability of land and/or dilution water and may also be limited by climate in locations where year-round land application is not feasible. Other methods. Other concentrate management and disposal alternatives such as blending with WWTP effluent or power plant cooling water may facilitate concentrate disposal and may be used in combination with 82 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

11 TABLE 5 Membrane desalting plant concentrate: treatment and disposal methods Treatment Methods Disposal Number % of No ph Air- Method of Plants* Total Treatment Aeration Adjustment Disinfection Degassification stripping Defoaming Surface water Sewer Subsurface injection Evaporation pond Land application Recycle Reuse Total *Three plants use multiple treatments: one plant uses ph adjustment and degasification, one plant uses ph adjustment and disinfection, and one plant uses disinfection and air-stripping. disposal methods previously mentioned. Permitting requirements for blending of residuals with treated WWTP effluent depend on the fate of the combined stream. Blending concentrate with large-volume cooling water streams from power plants using seawater for once-through cooling will greatly reduce concentrations of the discharge and facilitate permitting. The discharge must still comply with standard surface water discharge requirements. Nevertheless, blending either with wastewater effluent or power plant cooling water provides dilution to support the implementation of a membrane-desalting facility. Zero-liquid discharge systems such as thermal evaporators, crystallizers, and spray dryers are available to reduce residuals to a solid product for landfill disposal. However, the capital and operating (energy) cost of these systems is typically much higher than that of the desalting membrane facility, making this option infeasible except for very small concentrate flows. In certain situations, the highly concentrated brine from the brine concentrator may be sent to evaporation ponds instead of precipitating the solids. This option costs less than processing concentrate to solids. Use of high-recovery RO systems in front of the thermal evaporators can reduce costs for waters of limited hardness. The selective and sequential removal of salts followed by their use may offer promise to reduce zero-liquid discharge costs (Mickley, 2004). Reducing the cost of zeroliquid discharge systems is one of the major goals of the National Desalination Roadmap (US Bureau of Reclamation, 2003). Current research is evaluating the technical and economic feasibility of using desalting membrane concentrates as a feed stock for sodium hypochlorite generation and for solar energy ponds to recover energy by heat generation. This type of research may ultimately provide additional alternatives for managing desalting residuals. Concentrate treatment practices are illustrated in survey. In a US Bureau of Reclamation study survey (Mickley, 2004a), 112 of 141 recently built desalting plants provided information on the treatment of concentrate before disposal. The types of disposal and treatment along with the number of plants are summarized in Table 5. Two disposal alternatives exist for spent cleaning solutions. Spent chemical cleaning solutions may be handled and disposed of separately from the concentrate or blended with it. Before blending, it may be necessary to neutralize the cleaning solution s acidity or alkalinity to prevent unwanted reactions and ensure that the blended residuals are compatible with concentrate discharge regulations. Combining high-ph discharges may promote precipitation of supersaturated salts in the concentrate. For certain ED cleaning wastes, residual chlorine will need to be neutralized with a reducing agent. The disposal methods for cleaning solutions were investigated in a survey of 70 US membrane plants, including 49 NF/RO and 27 MF/UF facilities (Kenna & Zander, 2001). As shown in Table 6, the two most common methods for disposal of cleaning solution residuals from the NF/RO plants were sewer discharge and mixing with concentrate. Results from a 2002 US Bureau of Reclamation survey (Mickley, 2004a) agreed with those of the Kenna and Zander study (Kenna and Zander, 2000) and provided statistics on treatment and disposal of cleaning wastes from 110 recently built plants (Table 7). Spent cleaning solutions may be treated in the cleaning tank at the end of each cleaning step for small facilities or in separate tankage at large plants, given the latter s relatively low cost and the significant benefit of reducing membrane system downtime for the overall cleaning process. For ph adjustment, acid (e.g., sulfuric acid) or base (e.g., sodium hydroxide) is added to the COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER

12 TABLE 6 Disposal methods for chemical cleaning solutions at NF and RO plants in the United States spent solutions until the desired ph is reached. Relative costs of disposal alternatives are increasing. Since the development of desalting membranes in the 1960s, costs of desalting membrane treatment plants have declined, driven primarily by reduced costs for both membrane modules and systems, and improvements in membrane and system performance. However, with the possible exception of enhanced evaporation systems for use with ponds, costs associated with conventional disposal options have not decreased. As a consequence, the capital and operating costs associated with residuals management will continue to increase as a percentage of overall plant costs. This is particularly true for large-capacity plants with limited low-cost disposal options for high residuals flows. The Plants Employing Disposal Method Disposal Method NF (7 plants) % RO (42 plants) % Sewer Mixed with concentrate* Deep well injection 43 Ocean 14 Evaporation pond 5 Other (no data) 24 Source: Kenna & Zander, 2001 NF nanofiltration, RO reverse osmosis *For subsequent final discharge according to disposal option used for concentrate residual TABLE 7 Summary of spent cleaning waste disposal methods Treatment Methods Disposal Number Percent No ph Method of Plants* of Total Treatment Adjustment Settling Sewer Surface water Land application Subsurface injection Evaporation pond Recycle Hauling off site Total *One site included treatment by ph adjustment and settling. management costs of high residuals flows will, therefore, limit future decreases in membrane plant capital and operating costs (Mickley, 2004b). Moreover, future residual management costs are likely to increase because of growing difficulty in finding economical and environmentally sound means of disposal. A number of site-specific factors may limit the range of concentrate management options at a given site, including: suitability of the geological/ hydrological conditions for deep well disposal, level of dissolved solids and toxic ions in the concentrate, low seasonal flows in local surface waters, adequate capacity of the local sewers and WWTP for the concentrate volume, limitations in the local sewer ordinance, availability and cost of land, availability of dilution water for land applications, climate suitability for yearround land application, availability of land for evaporation ponds or other disposal methods, local value of water, and demand (amount and seasonal variations) for irrigation water. Some cost generalizations can be made in cases in which all disposal options are available, permitting is possible, and distances from the desalting membrane plant to the alternative management options are similar. Figure 8 shows the relative costs for concentrate management options as a function of concentrate flow based on the cost considerations listed previously. Table 8 provides costs for some of the concentrate management options as a function of flow rate. These costs are based on available information updated to 2004 US dollars and only represent relative costs in the absence of any site-specific information. Sanitary sewer and surface water discharges are the two most cost-effective methods of concentrate disposal, which explains their popularity. Subsurface injection, evaporation ponds, and spray irrigation could be competitive alternatives if local conditions are conducive. Typically, the zero-liquid discharge systems, such as brine concentrators, have the highest capital and operating costs. Under specific circumstances (e.g., cold climate, low evaporation and soil uptake rates, high land costs, and low power costs) the zero-liquid discharge systems could be cost-competitive with evaporation pond and spray irrigation alternatives. A more detailed breakdown of the capital and operating costs for the concentrate disposal methods is given in Mickley (2004a) and Watson et al (2003). Surface water discharge costs. Costs for surface water discharge are influenced by many site-specific fac- 84 DECEMBER 2004 JOURNAL AWWA 96:12 PEER-REVIEWED COMMITTEE REPORT

13 TABLE 8 Comparative capital costs for selected management options* Cost $ 1,000 Spray Irrigation Evaporation Pond Subsurface Injection** Brine Acres Acres Acres Acres Concentrator Flow (Hectares) $ (Hectares) $ (Hectares) $ (Hectares) $ $ at $ at Rate 2,500 ft 10,000 ft Capital Energy mgd at 2 ft/yr at 20 ft/yr at 0.5 gpm/acre at 2.0 gpm/acre ($ at ($ at Cost Cost (m 3 /d) (at 0.61 m/yr) (at 6.1 m/yr) (at 11 m 3 /d/km 2 ) (at 44 m 3 /d/km 2 ) 762 m) 3,048 m) $ $/yr , ,750 5,700 1, (37.85) (2.43) (60.98) (0.243) (12.20) (5.7) (72.73) (1.62) (18.18) (533.34) (1,737.80) , , ,000 1,750 5,700 2,000 1,230 (378.5) (24.3) (304.88) (2.43) (60.98) (56.7) (727.27) (14.2) (181.82) (533.54) (1,737.80) , ,200 1, , ,000 2,500 8,100 8,750 3,500 (3,785) (243) (1,829.27) (24.3) (365.85) (567) (7,272.73) (142) (1,818.18) (762.20) (2,469.51) ,400 2,800 8,500 14,900 6,850 (7,570) (48.6 (731.71) (853.66) (2,591.46) ,000 3,600 10,000 38,500 17,200 (18,925) (121.4) (1,829.27) (1,097.56) (3,048.78) *Only relative costs are given; site-specific costs may vary significantly. Costs are based on 2004 US$ and exclude the cost of conveying the concentrate to a site (Mickley, 2004). Costs exclude means of blending/dilution, pretreatment to meet water quality requirements, and monitoring wells. Costs exclude solids disposal and seepage monitoring. **Costs exclude pretreatment and standby disposal system. Based on power cost of $0.10/kW h; costs exclude solids disposal and disposal of possible small brine stream. tors and are difficult to generalize. The key cost factors are those of concentrate conveyance to the outfall, outfall construction and operation, concentrate treatment to meet NPDES permit limits, and environmental monitoring of the discharge. The concentrate conveyance costs are closely related to the concentrate volume and the distance to the discharge outfall. The outfall capital costs depend on outfall diameter, length, piping material, and diffuser system configuration. This cost could be eliminated if an existing outfall can be used. Outfall discharge operating costs are closely related to the need to aerate the concentrate before disposal or to treat to control WET. Environmental monitoring costs may be significant, especially if the discharge is near an impaired or sensitive water body or area of limited natural flushing. Sanitary sewer discharge costs. Sanitary sewer discharge conditions are usually site-specific. The key cost elements for this disposal method are the cost of conveyance (pump station and pipeline) and fees for connecting to the sanitary sewer and treatment/disposal of the concentrate at the WWTP. Sewer connection and treatment fees can vary significantly from none to several orders of magnitude larger than the conveyance costs. Sewer connection fees usually are related to the available capacity of the sewer facilities and the effect of the concentrate discharge on the WWTP s operating costs. Discharge to sewer is used more often with small- and medium-sized desalting plants because small concentrate volumes affect WWTP systems less. Subsurface injection costs. The key factors that influence subsurface injection costs are well depth and the diameter of the well tubing and casting rings. Several other cost factors are the need for concentrate pretreatment before disposal; pump size and pressure (which vary depending on the geologic conditions and depth of the injection zone); the size and configuration of the monitoring well system; and site preparation, mobilization, and demobilization. Disposal via deep well injection is expensive, but it has economies of scale for largecapacity desalting plants. Evaporation pond costs. The key cost variables for evaporation ponds include evaporation rate (climate), concentrate volume, land and earthwork costs, liner costs, and salinity of the concentrate, which determines the useful life of the ponds. The main cost variable is the evaporative area because evaporation rates are typically lower than soil uptake rates. (Disposal of the same volume of concentrate using evaporation ponds requires more land than disposal by spray irrigation.) Capital costs for evaporation ponds have few economies of scale and typically are only acceptable for small plants. The largest municipal plant discharging to evaporation ponds has a capacity of 1.5 mgd (5.7 ML/d); the others have capacities of <0.4 mgd (<1.5 ML/d). Land application costs. Land application or spray irrigation is usually cost-effective only if the concentrate can be blended with fresh water to reduce salinity to a level acceptable for irrigation or if the TDS of the feedwater and the membrane concentrate are low. Its feasibility also depends on the type of crops/vege- COMMITTEE REPORT PEER-REVIEWED 96:12 JOURNAL AWWA DECEMBER