Enhanced coagulation:

Transcription

1 NATURAL ORGANIC MATTER Enhanced coagulation: its effect on NOM removal and chemical costs Coagulation ph appeared to be a determining factor for maximum NOM removal. Gil Crozes, Patrick White, and Matthew Marshall C Several bench-scale studies were conducted to evaluate the effectiveness of enhanced coagulation for controlling chlorination by-products and to assess the impact of coagulation ph preadjustment on total organic carbon (TOC) removal and plant operating costs. Tests were conducted on a variety of surface water sources with TOC values ranging from 2 to 11 mg/l. An iron-based coagulant, such as ferric chloride, was consistently more effective than alum in removing natural organic matter (NOM). Coagulation ph appeared to be a determining factor for maximum NOM removal when ferric chloride was used as a primary coagulant. Typically, preadjustment of ph at a value of 6. ±.2 increased NOM overall removal to as much as 65 percent and reduced the coagulant dose by as much as 6 percent. Enhanced coagulation led to higher overall operating costs, but preadjustment of ph with sulfuric acid reduced costs by lowering the coagulant dosage as well as sludge production. hlorine is the most widely used oxidant in the United States for water disinfection. But chlorine reacts with the natural organic matter (NOM) that remains after treatment and forms halogenated by-products. Only some of these by-products have been identified, such as trihalomethanes (THMs) and haloacetic acids (HAAs), which are suspected to be human carcinogens. In 1979 the US Environmental Protection Agency (USEPA) set a maximum contaminant level (MCL) of.1 mg/l for total THMs in drinking water. More recently, USEPA proposed the Disinfectants/Disinfection By-products (D/DBP) Rule, which will reduce the THM MCL to.8 mg/l. In addition, HAAs will be regulated along with other DBPs at a proposed MCL of.6 mg/l. USEPA has also announced that as additional health risk data are collected, Stage 2 of the D/DBP Rule will be developed, which will revise the DBP MCLs. Long-term standards for THMs and HAAs may be 78 JOURNAL AWWA

2 Six surface water sources with TOCs ranging from 2 to 11 mg/l were analyzed using a standard jar-test apparatus. reduced to.4 mg/l and.3 mg/l, respectively, and are currently scheduled to be effective for large systems in 22. Alternatives generally available to water systems that exceed current or proposed MCLs include removing DBP precursors and using an alternative disinfectant for primary disinfection, distribution system protection, or both. Chloramines offer the significant advantage of virtually eliminating the formation of chlorinated by-products. However, the required contact time for inactivation of viruses and Giardia cysts to the level required by the Surface Water Treatment Rule (SWTR) is rarely available for chloramine postdisinfection at existing water treatment facilities. The use of chloramines as a residual disinfectant in the system is an effective method of limiting DBP formation during distribution of the water. Similarly, the use of chlorine dioxide can reduce the formation of chlorinated by-products during primary disinfection. The generation of chlorine dioxide and its reaction with NOM, however, leads also to the formation of by-products such as chlorite, a compound that is to be regulated under the D/DBP Rule at an MCL of 1. mg/l. This standard for chlorite and possible future limits on chlorine dioxide residual may limit the application of chlorine dioxide for primary disinfection of water that has a high disinfectant demand. Another alternative disinfectant to chlorine, chloramines, and chlorine dioxide is ozone. If used as a primary disinfectant for C X T compliance and followed by a chloramine residual in the distribution system, it can eliminate any significant contact between DBP precursors and chlorine. Although ozone is known to react with NOM to produce organic DBPs such as aldehydes and to increase the assimilable organic carbon (AOC) fraction, it is also known to react with bromide present in the water to form hypobromous (HOBr) acid. Depending on the ph of the water, HOBr in turn disassociates to form hypobromite ion (OBr ). HOBr and OBr can react with organic matter to produce brominated DBPs such as bromoform. Further, ozone may react with OBr - to produce bromate (BrO 3 ), a compound also scheduled to be regulated in the D/DBP Rule with an MCL of.1 mg/l. In some cases, high bromide concentration in the water supply may compromise the use of ozone for primary disinfection. Recognizing that chlorination will continue to be the most common disinfection process, enhanced removal of DBP precursors present in raw-water sources represents a valuable option for reducing the Enhanced coagulation is a valuable means of controlling DBP formation without requiring significant capital investments. potential for chlorinated by-products formation. The removal of NOM can be achieved by providing additional unit processes, such as granular activated carbon (GAC) and nanofiltration, or by enhancing existing coagulation, flocculation, and sedimentation processes. Optimizing the removal of DBP precursors for DBP control in conventional water treatment plants represents an economically attractive option. Enhanced coagulation is defined in the proposed D/DBP Rule as the addition of excess coagulant for TABLE 1 Enhanced coagulation requirements of the D/DBP Rule for percent TOC removal Alkalinity mg/l as CaCO 3 TOC mg/l < > % 3% 2% % 35% 25% >8 5% 4% 3% JANUARY

3 TABLE 2 TABLE 3 Alkalinity TOC Removal Water TOC Turbidity mg/l as Required Source mg/l ntu CaCO 3 ph percent Lake Houston and Trinity Missouri * Sacramento Big Sioux Weiser Mississippi NA *Presettled raw water Softened water Not applicable FIGURE 1 Water quality of sources tested Flocculation Flash Parameter Mixing Phase 1 Phase 2 Phase 3 rpm Time min G* s G T *G values are given at 22 o C for a 2-L square beaker, using a stirrer from Phipps and Bird, Richmond, Va Mixing conditions TOC removal by alum, ferric chloride, and a cationic synthetic polymer Ferric chloride Alum Catfloc DL Coagulant Dose mg/l Trinity water 8 percent; Lake Houston water 2 percent improved removal of DBP precursors by conventional filtration treatment. The rationale behind the introduction of this treatment technique requirement in the proposed D/DBP Rule was that only a very small fraction of the chlorination by-products and associated health risks have been identified; thus an increase in precursor removal would reduce overall known and unknown public health risks. A two-step process will be proposed for utilities with conventional treatment plants to demonstrate the installation of enhanced coagulation. The first step will be the use of a 3 X 3 matrix with required percentage removals of TOC based on raw-water alkalinity and TOC values (Table 1). Systems will be required to achieve the percentage removal of TOC between the raw water and the treated water prior to continuous disinfection, defined as the continuous addition of a disinfectant for the purpose of achieving a level of inactivation to meet the minimum requirements of the SWTR. If a conventional treatment plant cannot achieve the percent TOC reduction presented in the 3 X 3 matrix, the utility can go to a step 2 procedure to determine enhanced coagulation. This alternative enhanced coagulation procedure would require a coagulant dosage and a coagulation ph such that an incremental addition of 1 mg/l of coagulant results in a TOC removal of.3 mg/l or more. Also, step 2 would require enhanced coagulation to occur at ph values less than or equal to a maximum value determined by the raw-water alkalinity. The step 2 approach requires the approval of the primacy agency. This article evaluated enhanced coagulation not only to meet the TOC removals required under the D/DBP Rule, but also as a means to control DBP formation in order to meet the current THM MCL, the proposed short-term THM and HAA MCLs, and the long-term DBP standards. Increasing NOM removal in drinking water may also be beneficial for some other aspects of water treatment. For example, the chlorine demand can be significantly reduced. Also, lower levels of organic matter in the distribution system may reduce the potential for bacterial regrowth. As a result of higher coagulant dosages, the settled-water ph is further decreased, thus reducing the required C X T for primary disinfection, assuming the final ph adjustment for corrosion control purposes occurs downstream of primary disinfection. 8 JOURNAL AWWA

4 Implementation of enhanced coagulation in a conventional water treatment plant raises several issues. Increased coagulant dosages translate directly into higher sludge production. Existing sludge removal and sludge dewatering systems may be undersized. Chemical storage and feed facilities may not be adequate. Optimum turbidity removal may not be achieved under enhanced coagulation conditions. Enhanced coagulation will increase the overall cost of chemicals used for coagulation and for final ph adjustment. Background NOM encountered in waters can be in a particulate, colloidal, or dissolved form. Generally, particulate organic matter, which includes organic debris and microorganisms, is easily removed by coagulation, flocculation, and sedimentation in a manner similar to other turbidity-causing particles in water. Dissolved organic matter is defined as those organic molecules passing a.45-µm membrane filter. There is no widely accepted definition of colloidal organic matter in terms of size. Organic colloidal substances can include large molecule agglomerates as well as macromolecules passing through a.45-µm filter. What differentiates an organic colloid from an organic molecule in true solution is the capability of being coagulated by simple charge neutralization. Only a small fraction, 5 15 percent, of the dissolved organic matter naturally occurring in water can be characterized by gas or liquid chromatographic techniques associated with mass spectrometry. 1 The remaining higher-molecular-weight organics (85 95 percent) are usually referred to as humic substances. They are characterized on the basis of their water solubility at a given ph. 2 By definition, fulvic acids remain in solution in water at a ph of 1., whereas humic acids precipitate out of solution at this ph. Certain authors have further characterized NOM in terms of molecular weight. 3,4 Others supplemented the characterization by the use of a pyrolysis gas chromatographic mass spectrometric technique. 5 Humic and fulvic acids are anionic polyelectrolytes with a degree of ionization that depends on ph. Researchers performed a potentiometric titration of fulvic and humic acids and observed two inflection points, characteristics of phenol-oh groups (ph ) and of carboxylic groups (ph ). 6 FIGURE 2 Systems will be required to achieve a specific percentage removal of TOC between the raw water and the treated water prior to continuous disinfection Missouri presettled water TOC removal by alum and ferric chloride Ferric chloride Alum Coagulant Dose mg/l.1.3 mg/l of anionic polymer were added as a flocculant aid NOM removal through coagulation has been widely documented in the literature. 7 9 As recently summarized by Randtke, 1 the major mechanisms by which naturally occurring organics can be removed by coagulation involve charge neutralization of colloidal NOM, precipitation as humates or fulvates, and coprecipitation by adsorption on the metal hydroxide. The coagulation and flocculation of colloidal NOM can be caused by a compression of the double layer surrounding the charged colloid, by charge neutralization, by bridging, or by colloid entrapment in the coagulant precipitate. The latter phenomenon is commonly referred to as sweep flocculation. Precipitation of NOM refers to the formation of an aluminum or iron humate or fulvate with a lower solubility product. Forming a substance with a lower solubility product causes its precipitation into a solid form. In coprecipitation, the soluble organic material is adsorbed onto the growing crystals of the metal hydroxide. Bonds involved in adsorption of NOM can include Van der Waals interactions, hydrogen bonding, hydrophobic bonding, ionic bonds, or dipole interactions. The degree of NOM removal through coagulation is affected by the nature and dosage of coagulants as well as the ph. Synthetic organic cationic polymers can achieve colloidal NOM charge neutralization and possibly participate in the precipitation of humic and fulvic acids. However, they do not provide a substrate for adsorption of the organic matter. In comparison, aluminum- or iron-based coagulants can achieve col- JANUARY

5 FIGURE FIGURE 4 SDSTHMs µg/l Ferric chloride Alum Sacramento TOC removal by alum and ferric chloride Coagulant Dose mg/l TOC as a surrogate for DBP precursors D/DBP Rule, Stage 1 D/DBP Rule, Stage Trinity water 8 percent, Lake Houston water 2 percent loidal destabilization, can form aluminum or iron humates and fulvates that would precipitate, and also provide a surface area for adsorption on the metal hydroxides. In addition, the formation of the aluminum or ferric hydroxide floc may also be accompanied by the entrapment of some colloidal or dissolved organics, as well as concurrently formed humates or fulvates. The effectiveness of a given coagulant to remove organics may vary with the active charge density, the floc surface area available for adsorption, and the nature of the bonds between the organics and the metal hydroxide floc. Researchers have shown the stoichiometric relationship between humic and fulvic acids and the coagulants. 6 Thus, the coagulant dosage will have a direct effect on the overall level of NOM removal. The coagulation ph affects both the inorganic coagulating species and the level of dissociation of the fulvic and humic acids. At lower ph values, the level of organics protonation increases, thus reducing the coagulant demand. Also, the coagulating species are more positively charged at lower ph. As a consequence, adsorption becomes more favorable, and the required coagulant dosage decreases. Under lowph conditions, it is likely that the mechanisms of charge neutralization and coprecipitation by adsorption are enhanced. The precipitation of humate or fulvate forms may be reduced, or the precipitation of organic acids may still take place but with fewer active sites for humates and fulvates formation. This article presents data from six bench-scale studies performed to evaluate enhanced coagulation for TOC removal in order to control DBP formation and meet the enhanced coagulation requirements of the D/DBP Rule. Both organic synthetic coagulants and inorganic coagulants, including alum and ferric chloride, were tested. For each raw-water source, the level of TOC removal necessary to meet the Stage 1 D/DBP requirements, the enhanced coagulation requirements of the D/DBP Rule, and the long-term anticipated standards for DBPs were evaluated. The corresponding dosage of coagulant was determined with and without coagulation ph adjustment by simulating DBP formation under conditions matching those expected in the distribution system. These data were used to determine the cost of enhanced coagulation to meet any or all of the three goals previously defined. Materials and methods The experiments were performed with six surface water sources with TOCs ranging from 2 to 11 mg/l at the time of the testing. Table 2 summaries the water quality encountered during testing at each site. Coagulation, flocculation, and sedimentation experiments were performed at the bench scale using a standard jar-test apparatus* with six rectangular 2-L jars (Gator jars). This equipment was calibrated to control mixing energy (e.g., velocity gradient G) at the bench-scale level. The calibration curve provided by the University of Florida showed the relationship between G and the rotation speed of the stirrer paddle for various temperatures. Once this relationship is known, the flocculation mixing energy can be controlled at the bench scale and extrapolated to full scale. A sampling tap located 1 cm below the surface of the water in the jar allowed sampling of small quantities of settled water for analyses. The flocculation mixing conditions are reported in Table 3 for a water temperature of 22 o C. *Phipps & Bird, Richmond, Va. 82 JOURNAL AWWA

6 FIGURE 5 Prior to performing the actual simulated distribution system (SDS) DBP formation experiments, the chlorine demand of the settled water sample was determined. Because chlorine residual decay is related to initial chlorine concentration, three different chlorine dosages were applied to the water sample. The chlorine residual decay experiments were performed at the average system temperature of the water in 25 ml amber glass bottles precleaned to USEPA standards. The ph of the water was adjusted and buffered to match conditions in the system. A control blank was set up using ultrapure water chlorinated at the highest chlorine dosage used for each water. Chlorine residual was measured by using the DPD method at 4, 24, 72, and 168 h and at the SDS detention time. When chloramines were used as residual disinfectant in the system, SDS chlorination time was selected to match the average detention time in the plant clearwell (C X T contactor) under low-flow conditions to simulate worst-case conditions. For utilities using free chlorine in the distribution system, a chlorine residual of.2.5 mg/l at the end of the system was targeted under SDS conditions. For utilities using chloramines in the distribution system, the chlorine dosages used for the SDSDBP formation experiment were selected to provide a chlorine residual of at least 1 mg/l after the selected holding time at the holding temperature. It was anticipated that utilities would maintain a combined chlorine residual of at least 1 mg/l in the system. Under this scenario, a free chlorine concentration of 1 mg/l could be used for C X T calculation in the storage reservoir. Three chlorine dosages varying at a.5-mg/l increment were selected for the SDSDBP formation analysis using data from the chlorine decay tests. All holding bottles were new amber glass bottles with PTFE lids precleaned to USEPA standards. The selected free chlorine dose was added to a 25-mL bottle for each water sample. Each bottle was capped, headspace-free, to avoid stripping any halogenated compounds. The baseline temperature was maintained constant throughout these experiments. After the requisite holding time, the three incubated samples for each water were analyzed for free chlorine residual, and the values were recorded. For a sample with a chlorine residual close to the target, THM and HAA samples were collected in sample bottles containing sodium sulfite and ammonium chloride, respectively, as a preservative to quench the remaining free chlorine in solution and stop further chlorination reactions. These sample bottles were immediately refrigerated at 4 o C until shipment in a cooler with frozen blue ice to the laboratory. Halogenated compounds were extracted within 24 h after sampling and analyzed within 72 h following extraction. The following methods were used during benchscale testing and the DBP investigations. Chlorine residual (mg/l). Chlorine residual measurements were performed using the DPD colorimetric method adapted from method 45-CL G in the seventeenth edition of Standard Methods, 11 Impact of coagulation ph on TOC removal ph 8. Natural ph ph 7. ph 6. Natural ph Enhanced coagulation requirement, DBP control (Stage 1) DBP control (Stage 2) 2 4 Ferric Chloride Dose mg/l Raw-water TOC 6. mg/l; SDS with chloramines in distribution system; Trinity water 8 percent, Lake Houston water 2 percent Removal of NOM can be achieved by providing additional unit processes, such as GAC and nanofiltration, or by enhancing existing coagulation, flocculation, and sedimentation processes. and were checked using the amperometric titration method according to method 45-CL D. 11 The concentration of the stock chlorine solution was measured using the amperometric titration method. DBPs (µg/l). The method used for THM analysis consisted first of a liquid liquid extraction (LLE) as described by USEPA method A capillary gas chromatograph (GC) equipped with an electron capture detector (ECD) was then utilized for measure- JANUARY

7 FIGURE Effect of coagulation ph on TOC removal from presettled Missouri river water Natural ph 6.5 ph 6. ph 5.5 Natural ph DBP control (Stage 2) DBP control (Stage 1) Enhanced coagulation requirement bottles precleaned to USEPA standards, capped with a PTFE lid, and preserved with five drops of 95 percent phosphoric acid. Turbidity. Turbidity was measured with a portable turbidimeter* using method 214-A. 11 ph adjustment. In order to determine the experimental conditions for coagulation ph adjustment, a raw-water titration was performed using a sulfuric acid solution of.1 5. N, depending on the buffering capacity of the raw water. The ph to which the water naturally dropped after addition of a given coagulant at a given coagulant dosage was plotted on the titration curve. Using this composite curve, the amount of acid necessary to bring the ph from this value to the targeted ph value was determined. The correct amount of sulfuric acid was added to the 2-L jar prior to addition of coagulants and under rapidmixing conditions. The ph was finally checked at the end of flocculation in the jars.. FIGURE Ferric Chloride Dose mg/l Presettled raw-water TOC of 3. mg/l; SDS with chloramines in distribution system Effect of coagulation ph on TOC removal from Sacramento water No ph adjustment Coagulation ph of 6. Natural ph DBP control (Stage 1) Enhanced coagulation requirement DBP control (Stage 2) ment of the analytes. HAAs were analyzed using an acidic, salted ether LLE, requiring esterification with diazomethane prior to GC ECD analyses. Total and dissolved organic carbon (mg/l). Nonpurgeable total organic carbon (TOC) analyses were conducted using the persulfate ultraviolet (UV) light oxidation method as described by method 531 C. 11 TOC samples were collected on-site in amber Results and discussion Data obtained at three selected sites were used to compare three types of coagulants a medium-molecular-weight synthetic cationic polymer with a medium charge density, ferric chloride, and alum. Figure 1 shows settled-water TOC data for dosages of these three coagulants ranging from 2.5 to 8. mg/l, when treating a blend of 8 percent of Trinity water and 2 percent of the Lake Houston water from Texas. The TOC data points shown on the y- axis represent the TOC of the raw water at the time of the experiment. Figure 1 shows that as expected, only a small fraction of the TOC, about 1 mg/l, was amenable to coagulation by the synthetic cationic polymer. This fraction probably corresponds to the particulate and colloidal NOM that could be removed as a result of charge neutralization. A significantly larger fraction of TOC was removed by the inorganic coagulants. This shows that the mechanism of coprecipitation by adsorption of the NOM on the metal hydroxide is significant. At similar coagulant dosages, expressed as dry weight of coagulant, ferric chloride consistently outperformed alum. Several explanations can be provided that account for the observation. First, ferric chloride presents roughly two times more active positive charges per dry weight unit of coagulant than hydrated aluminum sulfate. Thus, colloid destabilization and the formation of humates and fulvates can both be expected to be achieved with ferric chloride at half the dosage of alum, as would turbidity removal. It should be remembered, however, that this mechanism is believed to play a minor role in NOM removal, and the active charge density of the coagulant is not prevalent in the context of enhanced coagulation. Furthermore, a standard commercially available ferric chloride solution (typically ranging from 4 to 45 percent) is more acidic than a 5 percent alum solution. Also, the alkalinity consumed during the formation of the metal hydroxides is two times higher for ferric chloride than alum. As a result, 84 JOURNAL AWWA

8 FIGURE 8 for a similar coagulant dosage, the coagulation ph will be lower with ferric chloride than with alum. As previously discussed, the lower ph, by increasing the protonation of the humic substances and increasing the positive charge of the coagulating species, reduces the coagulant demand and favors the adsorption of organics onto metal hydroxides. Finally, it is likely that ferric hydroxide floc and aluminum hydroxide floc present significant differences in terms of specific surface area, surface charge, and the extent of the most active adsorption sites. Although literature data are limited, it appears that the specific surface area of hydrous iron oxyhydroxides and aluminum hydroxides are of the same order of magnitude, ranging from 16 to 23 m 2 /g for iron floc 12 and ranging from 2 to 4 m 2 /g for alum floc. 13 However, because of the higher concentration of active metal in ferric chloride solution and the higher molecular weight of iron, a similar dose of ferric chloride is likely to produce 2.8 times more metal hydroxide by weight than a similar dosage of alum. As a result, the surface area for NOM uptake is comparatively greater for ferric chloride. TOC removals by alum and ferric chloride are also given in Figures 2 and 3 for Missouri and Sacramento river waters, respectively. In both cases, ferric chloride was superior in terms of NOM removal. Figures 2 and 3 also show that the overall fraction of NOM amenable to coagulation by ferric chloride is greater than for alum. This shows a greater affinity of a fraction of NOM for ferric hydroxide floc than for aluminum hydroxide floc. This difference in adsorption capacity is likely due to more active adsorption sites on the ferric hydroxide floc Effect of coagulation ph on TOC removal from Big Sioux water No ph adjustment Coagulation ph of 6.5 Natural ph DBP control (Stage 1) DBP control (Stage 2) Enhancement coagulation requirement Ferric Chloride Dose mg/l SDS with chloramines in the distribution system The coagulant performance comparison showed ferric chloride to be the better-suited coagulant for enhanced coagulation. As a result of the coagulant performance comparison, it appeared that ferric chloride is the better-suited coagulant for enhanced coagulation. Ferric chloride offered the advantages of reduced sludge production, a greater acidity that would reduce the cost of coagulation ph adjustment, and the potential for removing a greater fraction of NOM amenable to coagulation. Although polyaluminum chloride could be wellsuited for enhanced coagulation because of the high specific surface area of its floc (about 1, m 2 /g), 13 this product was not retained for this study. Also, ferric chloride was preferred over ferric sulfate because of its higher concentration of iron in the commercially available product, eventually translating into lower cost, and because liquid ferric chloride is more acidic. Optimum floc settleability and turbidity removal were obtained for the higher dosages of coagulant by using an anionic polymer as a flocculant aid. The flocculant aid was added at a typical dose of.15 mg/l after 1 min of flocculation, and significantly contributed to lowering the settled-water turbidities. As previously mentioned, enhanced coagulation has two primary goals. The first is to achieve the percentage TOC removal required for conventional treatment plants by the D/DBP Rule. This TOC removal goal is reported for each water source studied in Table 2. The second goal of enhanced coagulation is to remove sufficient DBP precursors to allow the use of free chlorine as a primary disinfectant, and potentially as a residual disinfectant in the distribution system, while meeting the DBP MCLs proposed in Stage 1 of the D/DBP Rule as well as their anticipated revision in Stage 2 of the rule. For each of the six water sources studied, the maximum settled-water residual TOC allowable to achieve each of these goals was determined by establishing a relationship between TOC and SDS DBP concentrations. TOC was then *21P, Hach Co., Loveland, Colo. POL-EZ 692, Calgon Corp., Pittsburgh, Pa. JANUARY

9 FIGURE FIGURE 1 No ph adjustment Coagulation ph of Effect of coagulation ph on TOC removal from Weiser water Natural ph DBP control (Stage 1) Enhanced coagulation requirement Ferric Chloride Dose mg/l Effect of coagulation ph on TOC removal from softened Mississippi water Natural ph ph 8. ph 7.5 ph 6. Natural ph DBP control (Stage 1) DBP control (Stage 2) Ferric Chloride Dose mg/l Softened-water TOC of 7.4 mg/l prior to enhanced coagulation with ferric chloride; enhanced coagulation requirement not applicable used as a surrogate to DBP precursors. An example of this approach is presented in Figure 4, for the blend of Trinity and Lake Houston waters. Once the settled-water TOC goals were established for each water source, the coagulation ph was adjusted to various levels below the natural ph drop caused by the coagulant in an attempt to reduce the primary coagulant (i.e., ferric chloride) dosages and thus reduce sludge production and potentially the cost of enhanced coagulation. In some cases several coagulation ph values were tested, whereas in other cases only one coagulation ph was evaluated. Data are presented in terms of settled-water TOC as a function of ferric chloride dose and coagulation ph in Figures 5 1. For the blend of Trinity and Lake Houston waters, the D/DBP Rule enhanced coagulation requirement and the DBP MCL of Stage 1 could both be met when the settled-water TOC was reduced to <4. mg/l and chloramines were used in the distribution system. This settled-water TOC could be achieved using 4 mg/l of ferric chloride with no ph adjustment or 2 mg/l of ferric chloride with reduction of the coagulation ph to below 7. by addition of acid. For this water quality, meeting the anticipated DBP MCLs of Stage 2 required 4 mg/l of ferric chloride and a coagulation ph as low as 6.. The Missouri experiences wide TOC variations; under high TOC conditions (e.g., 8. mg/l), the enhanced coagulation requirement of 25 percent TOC removal was easily met. Also, DBP control to meet the Stage 1 standards was effective by lowering the finished-water TOC below 3.5 mg/l if chloramines are used as a residual system disinfectant. However, during a low TOC period (e.g., 3 mg/l) the most stringent requirement for the Missouri proved to be the percentage TOC removal required by the D/DBP Rule. As shown in Figure 6, a ferric chloride dose of 6 mg/l was necessary to achieve that goal without ph adjustment with acid. The required coagulant dose was reduced to 3 mg/l by adjusting the coagulation ph to 6.5 and was reduced to 2 mg/l by further lowering the ph to 6.. The Sacramento is typically characterized by a low TOC of 2. mg/l and low alkalinity. Because of the low raw-water TOC, a treatment process optimized for turbidity removal would not need to practice enhanced coagulation for additional DBP control or to meet the DBP MCL of Stage 1. However, the finished-water quality, including THMs and HAAs under optimum turbidity removal, does not appear to qualify this utility for exemption from the enhanced coagulation requirements. The required percentage removal of TOC by enhanced coagulation is as high as 4 percent under the D/DBP Rule. Figure 7 shows that this goal could be met at a ferric chloride dose of 2 mg/l with no coagulation ph adjustment. Because of the already low coagulation ph of 6.6, the required coagulant dose could only be further reduced to 15 mg/l by adjusting the coagulation ph to the lower value of 6.. Under these two conditions, most of the NOM fraction amenable to coagulation was removed. Additional coagulant, with or without ph adjustment, had very limited return in terms of further TOC removal. Practicing enhanced coagulation as required by the D/DBP Rule would also allow 86 JOURNAL AWWA

10 FIGURE 11 Coagulation Chemical Cost $/mil gal Chemical costs of enhanced coagulation to meet Stage 1 or enhanced coagulation requirements of the D/DBP Rule Trinity Optimal turbidity removal With ph adjustment Missouri Sacramento Water Source No ph adjustment Big Sioux water utilities using the Sacramento water to meet the DBP MCLs anticipated in Stage 2. In the case of the Sacramento, coagulation ph adjustment offered little advantage, essentially because of the low buffering capacity of the water and the low raw-water TOC. The Big Sioux typically experiences high TOC and is a relatively well-buffered water. TOC removal data obtained with ferric chloride are given in Figure 8. Coagulation ph adjustment was only evaluated at a ph of 6.5 because it was estimated that targeting lower ph values would translate into prohibitive coagulation costs. Figure 8 shows that although the enhanced coagulation requirement was met at a ferric chloride dose of 5 mg/l (which could be reduced to 4 mg/l by depressing the ph to 6.5), sufficient DBP precursor removal could not be achieved to meet the DBP MCLs of Stage 1 under SDS conditions. In this case, SDS conditions for DBP formation evaluation included the use of chloramines as a residual disinfectant in the distribution system in order to control THM and HAA formation. As shown in Figure 8, the required finished-water TOC to meet the long-term DBP standards at the time of testing was beyond the reach of enhanced coagulation. Indeed, for ferric chloride doses above 6 mg/l, incremental Weiser Mississippi coagulant dosages translated into little additional TOC removal. The fraction of NOM not amenable to coagulation by ferric chloride appeared to be significant in the range of 5 to 6 mg/l of carbon. As a result, enhanced coagulation was not considered as a viable means of controlling DBP formation. Testing was performed at an existing softening plant, and enhanced coagulation was only evaluated as an alternate process. In this case, softening should remain the main treatment process, and DBP control should be achieved by the use of an alternative primary disinfectant in combination with the use of chloramines in the distribution system. The Weiser also typically exhibits a high level of NOM with a raw-water TOC of 8.4 mg/l at the time of the experiments. TOC removal data in Figure 9 show that the enhanced coagulation requirement could be easily met at a ferric chloride dose of about 3 mg/l. Adjustment of the coagulation ph to 6. was necessary to slightly reduce the required coagulant dose by 5 to 1 mg/l. Under the test conditions, the settled-water TOC required to meet the DBP MCLs of Stage 1 appears to be as low as 2. mg/l, despite the use of chloramines in the distribution system. To reach that TOC level, higher dosages of coagulant and coagulation ph adjustment to a value of 6. or lower were necessary. Nonetheless, Figure 9 shows that meeting this goal could be marginal because the TOC removal curve becomes asymptotic to the x-axis at higher coagulant dosages, showing that the settled-water TOC goal corresponds fairly closely to the fraction of NOM not amenable to coagulation. The long-term HAA standard could not be met by simply practicing enhanced coagulation. When these standards are implemented, an alternative primary disinfectant should be used. Figure 1 gives the results of the evaluation of enhanced coagulation with softened Mississippi water. The treatment facilities at the plant where the testing was carried out include softening as the first stage of treatment and conventional flocculation, sedimentation, and filtration as the polishing treatment processes. Softening at a ph of about 1.6 typically achieves a TOC reduction of about 5 percent, thus meeting the enhanced softening requirement of the JANUARY

11 FIGURE 12 Coagulation Chemical Cost $/mil gal Chemical costs of enhanced coagulation to meet Stage 2 and enhanced coagulation requirements of the D/DBP Rule Trinity Optimal turbidity removal With ph adjustment Missouri Sacramento Water Source No ph adjustment Big Sioux D/DBP Rule with a relatively large comfort level. The rationale for testing enhanced coagulation as secondstage treatment was to further reduce the DBP precursor level in the finished water so that free chlorine could be used as a primary disinfectant followed by chloramines for distribution system protection. The enhanced coagulation requirements of the D/DBP Rule do not apply to the second-stage treatment, which was strictly evaluated as a means for DBP control. Figure 1 shows that coagulation ph adjustment was very beneficial in increasing TOC removal at a given ferric chloride dose. It should be emphasized that the efficiency and cost of ph adjustment of the softened water can be significantly affected by excessive calcium carbonate floc carryover from the softening basins. Thus it is critical to control the softener s sludge blanket and to minimize the calcium carbonate floc carryover in order to avoid its subsequent dissolution during the recarbonation stage, which would increase the buffering capacity of the water. In the case of the Mississippi water treatment facilities, the adjustment of the softened water ph is performed using carbon dioxide. However, bench-scale ph adjustment was done using sulfuric Weiser Long-term DBP standards could not be met for the Big Sioux and the Weiser. Mississippi acid. The second stage treatment using ferric chloride as a coagulant and coagulation ph adjustment met both the DBP MCLs proposed in Stage 1 as well as the anticipated long-term standards. When 4 mg/l of ferric chloride was used at a coagulation ph of 6., the settled-water TOC was as low as 3.2 mg/l, representing an overall TOC removal through softening and subsequent coagulation approaching 8 percent. For each of the six cases studied, the chemical cost of coagulation was evaluated with and without ph adjustment and was compared with the cost of coagulation for optimum turbidity removal. Figure 11 shows chemical cost data for enhanced coagulation optimized to meet the DBP MCLs of Stage 1 or the enhanced coagulation requirement, which is more stringent. Figure 11 shows that in most cases the cost of enhanced coagulation was higher than the cost of optimum turbidity removal. With the exception of the Trinity and the Sacramento, ph adjustment with sulfuric acid significantly reduced the cost of enhanced coagulation. Whether ph adjustment is economically attractive seems to depend on the rawwater TOC, the level of TOC removal to be achieved, and the buffering capacity of the water. Only the cost of coagulation was evaluated, and the costs given in Figure 11 do not reflect those for final ph adjustment. Given the significant difference between the coagulation ph (as low as 6.) and the optimum ph for corrosion control ( ), it is economically sound to use lime when possible because its cost is significantly lower than that of caustic soda. The relative cost of enhanced coagulation with and without ph adjustment (Figure 11) should not be affected by final ph adjustment costs because in most cases sulfuric acid addition was used to reduce the ferric chloride dose necessary to reach a given ph. Indeed it appeared that once a minimum amount of ferric chloride floc is provided, additional ferric chloride dosage results in incremental TOC removal only because the ph is further depressed. It is therefore typically cheaper to depress that ph with sulfuric acid than with ferric chloride. Figure 12 shows the chemical cost of enhanced coagulation to meet the long-term DBP standards or 88 JOURNAL AWWA

12 the enhanced coagulation requirement. For the Big Sioux and the Weiser, the long-term DBP standards could not be met using enhanced coagulation. For the other cases studied, the overall cost of coagulation was further increased to meet more stringent DBP standards. With the exception of Sacramento water, ph adjustment provided for a significant reduction in the cost of coagulation as much as 5 percent for the Mississippi water. Conclusions The bench-scale studies evaluating enhanced coagulation showed that inorganic coagulants were superior to the synthetic organic polymer tested for NOM coagulation, because one of the major mechanisms of TOC removal involves coprecipitation by adsorption on metal hydroxides. Furthermore, ferric chloride proved to be consistently more effective than alum in removing NOM. Similar removal could be achieved at a lower coagulant dosage, and the fraction of NOM amenable to coagulation was greater for the ironbased coagulant. Coagulation ph appeared to be a determining factor for maximum NOM removal. Six case studies showed that enhanced coagulation is a valuable means of controlling DBP formation without requiring significant capital investments. The enhanced coagulation requirements of the proposed D/DBP Rule were met in all cases studied. Sufficient TOC removal was generally achieved by enhanced coagulation under SDS chlorination conditions to meet the DBP standards proposed in Stage 1 of the D/DBP Rule. However, in two of the six cases studied, the anticipated long-term HAA and THM MCLs could not be met solely by using enhanced coagulation. The installation of enhanced coagulation systematically increased the cost of coagulation, traditionally optimized for turbidity removal, as well as the cost of final ph adjustment for corrosion control. In most cases, however, coagulation ph adjustment with sulfuric acid reduced the cost of enhanced coagulation by reducing the required primary coagulant dose. A secondary advantage of lowering the coagulation ph is the diminution of sludge production. Acknowledgment The authors thank the following water utilities for their contribution to this study. The City of Houston Public Utilities Department and the staff of the East Water Purification Plant in particular; the Board of Public Utilities, Kansas City, Kan., and the Quindaro Water Treatment Plant staff; the City of Sacramento Water Department; the City of Sioux Falls Utilities; the City of Weiser Public Works, Idaho; the City of Minneapolis Department of Public Works and the staff at the Fridley Water Treatment Plant. The utilities identified here have not necessarily implemented enhanced coagulation with iron salts. Other issues such as soluble manganese contamination by liquid ferric chloride and scale up from bench to full-scale are still being investigated. Also, in some of the case studies, other methods for DBP control formation were selected. References 1. WATTS, C.D. Mass Spectrometric Identification of Non-volatile Organic Compounds; WRC Tech. Rept. TR11. Water Research Centre, Oxfordshire, England (April 1979). 2. AIKEN, R.G. ET AL. Humic Substances in Soil, Sediment, and Water. Wiley-Interscience, New York (1985). 3. SINSABAUGH, R.L. ET AL. Removal of Dissolved Organic Carbon by Coagulation With Iron Sulfate. Jour. AWWA, 78:5:74 (May 1986). 4. AMY, G.L. ET AL. Molecular Size Distributions of Dissolved Organic Matter. Jour. AWWA, 84:6:67 (June 1992). 5. BRUCHET, A. ET AL. Effect of Humic Substances on the Treatment of Drinking Water. Adv. in Chem. Ser., 219:93 (1989). 6. NARKIS, N. & REBHUN, M. Stoichiometric Relationship Between Humic and Fulvic Acids and Flocculants. Water Technol., 69:6 (1977). 7. SEMMENS, M.J. & FIELD, T.K. Coagulation: Experiences in Organics Removal. Jour. AWWA, 72:8:476 (Aug. 198). 8. HUNDT, T.R. & O MELIA, C.R. Aluminum Fulvic Acid Interactions: Mechanisms and Applications. Jour. AWWA, 8:4:176 (Apr. 1988). 9. DEMPSEY, B A.; GANHO, R.M.; & O MELIA, C.R. The Coagulation of Humic Substances by Means of Aluminum Salts. Jour. AWWA, 76:4:141 (Apr. 1984). 1. RANDTKE, S.J. Organic Contaminants Removal by Coagulation and Related Process Combinations. Jour. AWWA, 8:5:4 (May 1988). 11. Standard Methods for the Examination of Water and Wastewater. APHA, AWWA, and WPCF, Washington, D.C. (17th ed., 1989). 12. CROSBY, S.A. ET AL. Surface Areas and Porosities of Fe(III)- and Fe(II)-Derived Oxyhydroxides. Envir. Sci. & Technol., 17:12:79 (Dec. 1983). 13. BOTTERO, J.Y. & BERSILLON, J.L. Aluminum and Iron (III) Chemistry. Adv. in Chem. Ser., 219:425 (1989). About the authors: Gil Crozes is senior engineer at Montgomery Watson, 671 E. park Lane, Suite 2, Boise, ID Crozes is a graduate of the University Paul Sabatier (Toulouse, France) with a BA in biochemistry and of the National Institute of Applied Science (Toulouse) with an MS in environmental engineering and a PhD in water treatment and industrial processes engineering. He is a member of AWWA, and his work has been published by the Journal of Membrane Science. Patrick White is director of water treatment technology and Matthew Marshall is a PE, both at John Carollo Engineers, 5257 Fairview Ave., Suite 12, Boise, ID At the time this study was conducted, White and Marshall were employed by Montgomery Watson. JANUARY