coliform detections. Conversely, HPC levels in the SnCl 2 loop were significantly lower than in the control during the same period.

Transcription

1 A pipe-loop system with five parallel lines composed of new ductile-iron and lead pipes was operated for 13 months to investigate the effects of corrosion control chemicals on lead release and bacterial regrowth. Four of the pipe loops were treated with corrosion control chemicals orthophosphate, polyphosphate, a blend of orthophosphate and polyphosphate, and stannous chloride (SnCl 2 ) and one served as an untreated control. The pipe-loop system received chloraminated and filtered surface water from a full-scale lime-softening plant and was monitored for total lead, dissolved lead, heterotrophic plate count (HPC) bacteria, and coliforms. Total lead concentrations in each of the treated loops were significantly lower (95th percentile confidence level) than in the untreated control, and orthophosphate (1 mg/l as P) consistently yielded the lowest lead concentrations of all of the chemicals tested, including SnCl 2 (0.125 mg/l). Nevertheless, none of the chemicals was able to consistently maintain total lead concentrations below the 15-µg/L action level for the 8-h stagnation time sample in the new lead pipes used for this study. In addition, all of the phosphate-containing chemicals caused HPC populations to exceed those in the untreated control loop during the Sampling taps at the effluent end of the system facilitate sampling of the five pipe loops. warm summer months, raising concerns about regrowth and the potential for increased coliform detections. Conversely, HPC levels in the SnCl 2 loop were significantly lower than in the control during the same period. BY RAYMOND M. HOZALSKI, ELIZABETH ESBRI-AMADOR, AND CHE FEI CHEN Comparison of stannous chloride and phosphate for lead corrosion control H istorically, lead (Pb) has been considered a convenient and suitable material for the conveyance of water (Akers & Fellow, 1979). Lead is malleable, which makes it easy to form into pipes, and it is durable and relatively resistant to corrosion. Because of its toxicity, however, lead is no longer installed in plumbing systems in the United States. Lead can accumulate in the body over time with health effects that include kidney damage, impaired cognitive performance, impaired reproductive function, anemia, elevated blood pressure, and delayed neurological and physical development (Tate & Arnold, 1990). Lead is also classified as a probable human carcinogen. Lead can enter drinking water from various sources including lead service connections, lead pipes in homes, lead-based solder used to join copper pipes, and brass faucets (Lee et al, 1989). Therefore, the US Environmental Protection HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

2 FIGURE 1 Schematic of the pipe-loop system showing one of five parallel lines Filtered water Chemical feed tank Chemical feed pump Static mixer Pressure gauge Needle valve Solenoid valve Sampling tap Feed pump Flow meter 3-in. (75-mm) ID x 40 ft (12-m) ductile-iron pipe in. (12.5-mm) ID x 25 ft (7.6-m) lead pipe Solenoid valve Valve To waste ID inside diameter TABLE 1 Sample pipe-loop operating schedule for a lead pipe* Time Flow cycle 4:30 a.m. On 5:30 a.m. Off 1:30 p.m. On 2:00 p.m. Off 4:30 p.m. On 5:00 p.m. Off 7:30 p.m. On 8:00 p.m. Off 10:30 p.m. On 11:00 p.m. Off 1:30 a.m. On 2:00 a.m. Off *The schedule for each pipe was staggered a few minutes for ease of sampling. Begin 8-h stagnation period. End 8-h stagnation period. Agency (USEPA) established the Lead and Copper Rule (LCR) in 1991 to protect the public from lead exposure via drinking water. The USEPA established the action level (AL) for lead at 15 µg/l as Pb. The LCR requires that water utilities monitor for lead, treat the water to minimize lead concentrations when the AL is exceeded, and replace lead connections when treatment approaches are ineffective. Even though lead pipes are no longer installed in the United States, distribution systems in many cities still have lead service connections that can release lead into the water. A recent highly publicized problem involving elevated lead concentrations in tap water in Washington, D.C., serves as an unfortunate example of the potential severity of the problem (Renner, 2004). According to a 1990 survey, there were approximately 6.4 million lead connections and 3.3 million lead service lines in US distribution systems, and approximately 61,000 lead service lines were being removed per year (Economic & Engineering Services, 1990). Thus, lead in drinking water supplies is likely to be a problem in many systems for the foreseeable future. BACKGROUND Lead corrosion. The most common oxidant for lead in drinking water is dissolved oxygen (DO), and the oxidized form of lead encountered is typically Pb(II) (Schock et al, 1996). However, Pb(IV) can also occur in highly oxidized environments such as drinking water distribution systems (Renner, 2004; Wysock et al, 1991). In fact, it is hypothesized that in the case of the Washington, D.C. system, conversion from free chlorine to chloramine as the terminal disinfectant resulted in a shift from Pb(IV) minerals to Pb(II) minerals and in a subsequent increase in lead release (Renner, 2004). Lead has the ability to form a variety of complexes under the chemical conditions commonly encountered in drinking water (Schock et al, 1996; Wysock et al, 1991). For example, Wysock et al (1991) reported that the scale on lead service lines removed from a water distribution system consisted of lead carbonate (PbCO 3 ), calcium carbonate (CaCO 3 ), Pb(II) oxide (PbO), Pb(IV) oxide (PbO 2 ), hydrocerussite [Pb 3 (CO 3 ) 2 (OH) 2 ], and unidentified solids containing aluminum, silica, iron, potassium, and sodium. Factors affecting lead corrosion and the concentration of lead found in drinking water include age of the plumbing material, stagnation time of the water inside 90 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

3 This overview shows the entire length of the pipeloop system at St. Paul (Minn.) Regional Water Services. the pipes, and water quality parameters such as temperature, ph, alkalinity, and corrosion inhibitors. In terms of pipe age, lead concentrations in new lead pipes are typically higher than in old pipes (Murrel, 1988). This is most likely because new lead surfaces do not have any protective layer of scale between the metal and the water. Stagnation time is also an important factor because lead concentrations in lead pipes continued to increase for 6 h following a rapid increase over the first 2 h (Bailey et al, 1986). Wong and Berrang (1976) reported that after 4 20 h inside lead-soldered copper pipes, lead concentrations could reach 50 µg/l or more. Alkalinity and ph significantly affect lead solubility. Increasing ph is often an effective strategy for decreasing soluble lead concentrations, with the lowest lead concentrations occurring at ph >8 (Schock, 1989; Karalekas et al, 1983). Alkalinity is important not only because of its ability to buffer ph but also because it reacts with lead to form a protective PbCO 3 scale layer. Sheiham and Jackson (1981) reported that at alkalinity <50 mg/l as CaCO 3, the lead concentration was highly sensitive to ph (7 8.5), but at alkalinity >100 mg/l as CaCO 3, the lead concentration was insensitive to ph. Patterson and O Brien (1979) suggested that the optimum ph for controlling lead concentrations was between 8 and 8.5 when the alkalinity was at least 20 mg/l as CaCO 3. Corrosion inhibitors. Many water utilities add corrosion inhibitors to the treated water to control lead concentrations, copper concentrations, or both to comply with the LCR. Corrosion inhibitors mainly work by forming a protective scale layer on the pipe walls. The scale layer typically reduces metal solubility and decreases the corrosion rate and the subsequent release of pipe material (e.g., lead) into the water. One approach for creating a low-solubility scale layer is to add a corrosion control chemical containing phosphate. The effectiveness of phosphate for lead corrosion control is attributed to its ability to form solids such as hydroxypyromorphite that are less soluble than PbCO 3 over a wide ph range. Phosphate is typically added to finished water as orthophosphate or as a combination of orthophosphate and polyphosphate. A thorough review of phosphate use for corrosion control can be found elsewhere (McNeill & Edwards, 2002); this article discusses only selected results concerning lead. Orthophosphate is commonly used and typically effective at reducing lead concentrations (McNeill & Edwards, 2002). Although a wide range of doses have been applied (<0.1 to >1 mg/l as P), doses in the range of 1 to 1.6 mg/l as P are recommended for limiting lead corrosion by-product release (Edwards et al, 1999). For example, an 80% decrease in lead concentration has been observed with a dose of 1 mg/l as P (Colling et al, 1988). Other researchers reported that lead concentra- TABLE 2 Description of pipe-loop treatments Pipe Loop Corrosion Inhibitor Dose SnCl 2 SnCl 2 * mg/l as SnCl 2 Ortho-P Orthophosphate 1 mg/l as P Poly-P Polyphosphate 1 mg/l as P Blend Orthophosphate 1 mg/l as P polyphosphate Control None None SnCl 2 stannous chloride *AS-811, A.S. Inc., Alameda, California C-9, Hawkins Chemicals, Minneapolis, Minn. C-5, Hawkins Chemicals, Minneapolis, Minn. C-4, Hawkins Chemicals, Minneapolis, Minn. HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

4 TABLE 3 Pipe-loop influent water quality during the study period Parameter tions decreased to <15 µg/l with a dose of 1.5 mg/l as P in pipe-loop testing with both new and old lead pipes (Cantor et al, 2000). Polyphosphate has been used for many years as a deposit inhibitor. Many utilities use polyphosphate to keep iron and calcium in a soluble form to avoid red water stains in customers fixtures and excessive CaCO 3 scaling in pipes (Cantor et al, 2000). Even though its main purpose is to sequester metals in solution, polyphosphate is also marketed as a corrosion inhibitor, most likely because of its tendency to revert to orthophosphate (Schock, 1989). However, many studies have shown that polyphosphate is not effective at reducing lead concentration levels and that it could increase them in many cases (Edwards & McNeill, 2002; Cantor et al, 2000; Dodrill & Edwards, 1995; Holm & Schock, 1991; Schock & Wagner, 1985). For example, in addition to orthophosphate, Cantor et al (2000) tested a 2:1 orthophosphate and polyphosphate blend (0.6 mg/l as P). Lead concentrations in the pipes treated with the blend varied from 190 to 810 µg/l as Pb and were consistently higher than lead concentrations in pipes not treated with corrosion inhibitor. A new corrosion inhibitor. Stannous chloride (SnCl 2 ) is a relatively new corrosion inhibitor that has only recently been approved for use in potable water distribution systems in some states at concentrations in the sub- to low-milligram-per-litre range. The chemical has not been certified for potable water use by NSF International, but it has been approved for use as an antioxidant and preservative in food products by the US Food and Drug Administration (USFDA, 2003; USFDA, 1982). One regulation specifies a maximum level of % as tin in food (USFDA, 1982), but a company recently was granted a variance to test SnCl 2 as a preservative of canned asparagus spears at concentrations up to 35 ppm (0.0035%) in the finished food (USFDA, 2003). Little is known about the effectiveness of SnCl 2 for lead corrosion control in potable water distribution systems, and the authors are not aware of any information available in the peer-reviewed literature. The mode of action of SnCl 2 as a corrosion inhibitor is unclear. When added to water containing DO or chlorine or both, it likely will be oxidized to the stannic or Sn(IV) form (Cotton & Wilkinson, 1980) because the standard electrode potential for the Sn +4 /Sn +2 redox couple is V. One possibility is that the Sn +4 combines with Pb +2 to form a low-solubility precipitate, possibly lead stannate (PbSnO 3 ). There are no known human health effects of ingesting drinking water dosed with SnCl 2, but laboratory studies have shown endotoxic effects and reduced cell viability in Escherichia coli cultures (Dantas et al, 1996) and mammalian cell cultures (Dantas et al, 2002). For example, SnCl 2 concentrations as low as 10.5 mg/l induced low but significant levels of deoxyribonucleic acid damage in human-derived K562 cells (Dantas et al, 2002). Other researchers have reported that SnCl 2 has mutagenic (Singh, 1983) and carcinogenic (Ashby & Tennant, 1991) characteristics. The low- or sub-milligram-per-litre concentrations added to drinking water Mean ± Standard Deviation Temperature o C 12.7 ± 7.7 (n* = 51) ph 8.54 ± 0.26 (n = 51) Alkalinity, total as CaCO 3 mg/l 40.0 ± 8.6 (n = 51) Total hardness as CaCO 3 mg/l 76.0 ± 7.2 (n = 51) Calcium hardness as CaCO 3 mg/l 50.7 ± 9.4 (n = 51) Magnesium hardness mg/l 25.3 ± 7.9 (n = 51) Total residual chlorine mg/l 3.41 ± 0.42 (n = 46) CaCO 3 calcium carbonate, n number of samples FIGURE 2 Influent water temperature throughout the 13-month testing Temperature C /6/01 10/6/01 11/6/01 12/6/01 1/6/02 2/6/02 3/6/02 4/6/02 Sampling Date 5/6/02 6/6/02 7/6/02 8/6/02 9/6/02 10/6/02 92 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

5 At the influent end of the system, tanks and pumps feed the four corrosion control chemicals into the pipe loops. for corrosion control are 1 2 orders of magnitude lower than levels shown to have deleterious effects in laboratory cultures. STUDY PARAMETERS St. Paul Regional Water Services. St. Paul Regional Water Services (SPRWS) supplies drinking water to more than 400,000 customers in the city of St. Paul, Minn., and its suburbs. The raw water primarily comes from the Mississippi River and passes through a chain of lakes before entering the treatment plant. The water treatment system consists of lime softening, flocculation, sedimentation, disinfection (free chlorine followed by ammonia addition to form monochloramine), filtration, and corrosion control. The utility supplies an average of 50 mgd (190 ML/d) and has a finished water storage capacity of mil gal (515.5 ML). The distribution system consists of approximately 1,100 mi (1,770 km) of water mains and 92,000 water services. More than Corrosion inhibitor addition. Because SPRWS was exceeding the AL of 15 µg/l in some of its distribution system sampling sites (17 out of 99), beginning in December 1999, a corrosion inhibitor was added to reduce lead corrosion. A blend of approximately 70% orthophosphate and 30% polyphosphate was initially added at a concentration of 0.03 mg/l as P and then increased to 0.1 mg/l as P in January Three months later, the concentration of the chemical was increased to between 0.3 and 0.4 mg/l as P. A rare positive total coliform result was obtained during routine sampling in April A month later, the number of sites that tested positive for total coliform increased to eight. In response to the coliform results, SPRWS stopped feeding the phosphate blend May 12, In June 2000, the number of sites with total coliform decreased to six and then to one site in July. Switch to SnCl 2. The positive total coliform counts may have been caused by one or both of the following processes: (1) the addition of phosphate, a microbial nutrient, to the water stimulated biological growth, or (2) polyphosphate caused the release of corrosion products and biofilms that were attached to the pipe wall. SPRWS concluded that the increase in the phosphate corrosion inhibitor concentration had caused the problem because no other significant operational or water quality changes had occurred. Therefore, the water utility switched to SnCl 2 1 in the fall of The dose initially applied was mg/l as SnCl 2, which was increased to mg/l as SnCl 2 during the warmer summer months (June September). To date, the use of SnCl 2 appears to Lead can enter drinking water from various sources, including lead service connections, lead pipes in homes, lead-based solder used to join copper pipes, and brass faucets. 23,000 service lines in the system are composed of lead or a combination of lead and some other material. Although the finished chloraminated water is biologically stable, there is some concern about regrowth in the distribution system because the plant effluent has a moderate-to-high assimilable organic carbon concentration of 164 ± 24 µg/l and a high dissolved organic carbon concentration of 3 5 mg/l (Zhang et al, 2002). control lead release to acceptable levels without negatively affecting microbiological water quality. Despite some initial success with SnCl 2, SPRWS was interested in learning more about corrosion inhibitors, including orthophosphate. One of the key questions was whether orthophosphate alone (i.e., without polyphosphate) could ensure LCR compliance without causing adverse effects on water quality. HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

6 FIGURE 3 Comparison of total lead concentration (8-h stagnation time) in the pipe loops throughout the 13-month testing period SnCl 2 Polyphosphate Control Orthophosphate Blend AL /6/ /6/2001 Total Lead µg/l 11/6/ /6/2001 1/6/2002 2/6/2002 3/6/2002 4/6/2002 5/6/2002 6/6/2002 7/6/2002 8/6/2002 9/6/ /6/2002 Sampling Date AL action level, SnCl 2 stannous chloride The vertical line indicates the start of corrosion inhibitor addition Oct. 18, Study objectives. The main goal of this research was to identify corrosion inhibitors that would control lead release without causing coliform or other microbiological problems. Four corrosion control chemicals (orthophosphate, polyphosphate, an orthophosphate polyphosphate blend, and SnCl 2 ) were evaluated for approximately one year on a pipe-loop system that simulated the SPRWS distribution system. The pipe-loop system consisted of five loops four treated with corrosion inhibitors and one untreated control. Total and dissolved lead, heterotrophic plate count (HPC), and coliform concentrations were monitored for each loop throughout the study. The authors believe this is the first direct comparison of SnCl 2 and phosphate-based corrosion inhibitors for lead corrosion control over an extended period of time in a flow-through pipe-loop system. This study also provided an evaluation of the effects of corrosion inhibitor addition on not only Pb concentrations but also bacterial concentrations. MATERIALS AND METHODS Pipe-loop system. The pipe-loop system was based on the design of an AWWA Research Foundation pipe-loop model (Economic & Engineering Services, 1990) except that a section of ductile-iron pipe was added to each line to simulate the water main. The system consisted of five parallel loops mounted on a plywood frame. Each loop was made up of 40 ft (12 m) of 3-in. (75-mm) insidediameter (ID) ductile-iron pipe followed by 25 ft (7.6 m) of 0.5-in. (12.5-mm) ID lead pipe 2 (Figure 1). The iron pipe simulated the water main, and the lead pipe simulated lead service lines and in-home lead piping. The pipe-loop system was constructed at the SPRWS treatment facility in the summer of On Aug. 22, 2001, the iron pipes of each loop were inoculated with a suspension of E. coli in an attempt to seed the system and establish a biofilm on the pipes. The goal was to mimic a mature distribution system pipe in which biofilms and coliforms often reside. Flow of filtered water through the loops was initiated a week later. Preliminary data from the five pipe loops were collected before the start of corrosion inhibitor addition Oct. 18, Bacteria inoculation. An E. coli (strain K-12) culture stored in glycerol at 70 o C was thawed and then streaked onto an m-endo agar plate. After the plate was incubated 94 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

7 TABLE 4 Results of a linear regression analysis of water quality parameters with temperature Parameter Total Lead Dissolved Lead Running Water Lead HPC (n = 29) (n = 20) (n = 22) (n = 32) Slope Slope Slope Slope Pipe Loop µg/l 1 o C 1 p Value µg/l 1 o C 1 p Value µg/l 1 o C 1 p Value cfu (100 ml) 1 o C 1 p Value SnCl Orthophosphate Polyphosphate Blend Control HPC heterotrophic plate count, n number of samples, p probability, SnCl 2 stannous chloride for 24 h at 35 o C, a single colony was removed from the plate using a flame-sterilized metal loop and placed into 500 ml of m-endo liquid medium. The medium was then incubated overnight at 35 o C to achieve a final cell density of cfu/ml. Approximately 100 ml of the culture was poured into each empty iron pipe through a hole made at the middle of the 40-ft (12-mm) section. Finished water from the treatment plant was dechlorinated by adding bisulfite and was then pumped into the influent of the pipes until the pipes were full. The diluted E. coli suspension was allowed to incubate inside the pipes for approximately one week before continuous water flow was initiated. System operation. Chloraminated and filtered water from a full-scale lime softening plant was pumped into the influent header of the pipe-loop system at 5 gpm (0.3 L/s) and then split into each of the five loops. The corrosion inhibitors were fed into the loops from 20-gal (76- L) tanks 3 using chemical feed pumps. 4 After inhibitor addition, the water passed through a static mixer before entering the ductile-iron pipe section. The total flow through each loop was approximately 1 gpm (0.06 L/s). Water flowed continuously through the ductile-iron pipes to simulate flow conditions in a main but flowed intermittently through the lead pipe following a schedule that simulated household water use (Table 1). Water flow through the lead pipes was controlled by automatic solenoid valves 5 and a programmable controller. 6 Corrosion inhibitors. Four corrosion inhibitors were tested: orthophosphate (ortho-p), polyphosphate (poly-p), a blend of orthophosphate and polyphosphate (blend), and SnCl 2 (Table 2). The polyphosphate stock solution was composed primarily of polyphosphate (79%) but also contained some orthophosphate (21%). The blend consisted primarily of orthophosphate (71%) with the balance being polyphosphate (29%). A comparison of costs at the time of the study showed that SnCl 2 ($12.09/mil gal [$0.32 ML] at a dose of mg/l as SnCl 2 ) was 80% more expensive than a phosphate-based corrosion inhibitor ($6.70/mil gal [$0.18 ML] at a dose of 1 mg/l as P) per unit volume of water treated. Sampling for lead and other water quality parameters. Approximately once a week, water samples for analysis of lead were collected from the effluent end of the lead pipes following an 8-h stagnation period. This stagnation period mimicked the stagnation of water inside household plumbing overnight, and the 8-h stagnation sample was analogous to the first-draw sample used for LCR compliance monitoring. At the end of the 8-h stagnation period, samples were collected by opening the tap from the lead pipe and closing the waste line valve. The water from the lead pipe was collected in plastic bottles 7 for subsequent analysis of lead and other water quality parameters. Before sampling, all bottles were cleaned by filling them with a 0.5 N nitric acid solution for 24 h or more and then rinsing three times with reverse osmosis filtered water, in accordance with method 3010 (Standard Methods, 1995). The first 60 ml of water was discarded because it comprised water from the tap and sampling line. Next ~250 ml of water was collected for analysis of temperature, ph, alkalinity, and hardness. Then ~125 ml of water was collected for lead analyses. Another 500 ml was collected for chlorine analysis. After the 8-h stagnation samples were taken, a final 50-mL sample was collected from the lead pipe 1 min or more after water flow resumed. This sample was called a running water sample and was used to determine the amount of lead present in water that was not permitted to stagnate in the pipe. Finally, each time the pipe loop was sampled, 1 L (0.3 gal) of the influent water was collected to measure temperature, ph, alkalinity, hardness, chlorine, and lead HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

8 FIGURE 4 Comparison of dissolved lead concentrations (8-h stagnation time) in the pipe loops throughout the 13-month testing period SnCl 2 Polyphosphate Control Orthophosphate Blend AL /6/2001 Dissolved Lead µg/l 10/6/ /6/ /6/2001 1/6/2002 2/6/2002 3/6/2002 4/6/2002 5/6/2002 6/6/2002 7/6/2002 8/6/2002 9/6/ /6/2002 Sampling Date AL action level, SnCl 2 stannous chloride The vertical line indicates the start of corrosion inhibitor addition Oct. 18, concentration. Temperature was measured immediately after sample collection with a glass thermometer. After all of the loops were sampled, the bottles were taken to the SPRWS laboratory for analysis of ph, alkalinity, hardness, and chlorine. Samples were collected weekly from the influent of the pipe loop system and from the effluents of the ductileiron pipes to determine HPC and coliform bacteria concentrations. Before the water samples were collected from the loops for microbiological testing, the copper taps were sterilized by flaming with a propane torch for at least 10 s. All microbiological samples were collected in 500-mL plastic bottles that were presterilized by autoclaving. After collection, microbiological samples were immediately taken to the laboratory for analysis. Pipe sampling and analysis. Samples of lead pipe were removed from the SnCl 2 loop and analyzed by X-ray diffraction (XRD) in an attempt to elucidate the mode of action of the chemical in controlling lead release. Immediately after the system was shut down, the pipes were drained. Approximately 26 weeks after the system was drained, a 6-in. (150-mm) section of pipe from the influent end was cut out and split in half to expose the interior pipe surface. Then ~1-in. (25-mm) pieces were cut from one of the pipe halves for analysis by XRD at the University of Minnesota Characterization Facility in Minneapolis. XRD analyses were performed with a microdiffractometer 8 with copper radiation, a graphite incident-beam monochromator, a 0.03 in. (0.8 mm) point collimation system, and a two-dimensional multiwire area detector. Cross sections of the lead pipe were held in a special sample holder, and the beam was focused on the deposit inside the pipe. Data were collected to cover a range of 28 to 87 (2 o ) with a 300-s collection time per frame. The frames were integrated with a step size of 0.04 (2 o ) and plotted versus intensity. Identification of the deposits was attempted using the library of powder patterns from the International Committee of Diffraction Data database and specialized software. 9 Lead analyses. From the 125-mL sample collected for lead analysis, 30 ml was passed through a syringemounted filter (0.45-µm membrane filter 10 ) into a 50- ml plastic bottle for analysis of the dissolved lead concentration. The remaining unfiltered sample was used to determine the total lead concentration. All lead sam- 96 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

9 FIGURE 5 Comparison of running water total lead concentrations in the pipe loops throughout the 13-month testing period SnCl 2 Polyphosphate Control Orthophosphate Blend AL /6/2001 Dissolved Lead µg/l 10/6/ /6/ /6/2001 1/6/2002 2/6/2002 3/6/2002 4/6/2002 5/6/2002 6/6/2002 7/6/2002 8/6/2002 9/6/ /6/2002 Sampling Date AL action level, SnCl 2 stannous chloride The vertical line indicates the start of corrosion inhibitor addition Oct. 18, ples were acidified with nitric acid to a concentration of 0.5%. The lead samples were stored at room temperature and analyzed within six weeks. Lead concentration was determined by a graphite furnace atomic absorption spectrophotometer equipped with an autosampler. 11 A quality control (QC) sample was analyzed before each set of water samples was analyzed. If the result from the QC sample was not within ±10% of the certified concentration, 12 the instrument was recalibrated. Spiked samples and duplicate samples were also analyzed periodically for quality assurance and QC purposes. The method detection limit (MDL) for lead was 1 µg/l as Pb. Microbiological analyses. HPC and coliform concentrations were determined using the membrane filter technique in methods 9215 and 9222 (Standard Methods, 1995). The filtering device, consisting of six individual funnels and filter holders attached to a vacuum manifold, was sterilized by autoclaving for 30 min before the samples were processed. After ml of sample had passed through the 0.45-µm-pore-size membrane filters, 13 each filter was removed and placed in a 47-mm (1.85-in.) disposable plate containing either m-coliblue 24 agar 14 (for coliform analysis) or R2A agar (for HPC analysis). For each water sample, duplicate coliform plates and triplicate HPC plates were prepared. The plates were incubated membrane-filter-side up at 35 o C for 24 h (coliform analysis) or h (HPC analysis). Other water quality analyses. Alkalinity was analyzed by titration with 0.02 N sulfuric acid according to method 403 (Standard Methods, 1995). Total and calcium hardness were analyzed by titration with EDTA according to methods 314 and 311, respectively (Standard Methods, 1995). Measurements of ph were made with a ph meter. Chlorine was determined by the DPD method (Standard Methods, 1995). Data analysis. All statistical analyses were performed using the data analysis package included with the software. 15 Comparisons of the effects of different corrosion inhibitors on various water quality parameters were made by performing paired t-test analyses. Linear regression analysis was used to evaluate the effects of temperature on lead concentrations and HPC. Probability (p) values were reported for the statistical analyses, and results with p values <0.05 were deemed statistically significant. HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

10 FIGURE 6 Comparison of HPC in the pipe loops throughout the 13-month testing period 1,400 1,200 1,000 SnCl 2 Polyphosphate Control Orthophosphate Blend AL /6/ /6/ /6/ /6/2001 1/6/2002 2/6/2002 3/6/2002 4/6/2002 5/6/2002 6/6/2002 7/6/2002 8/6/2002 9/6/ /6/2002 HPC cfu/100 ml Sampling Date HPC heterotrophic plate count, SnCl 2 stannous chloride The vertical line indicates the start of corrosion inhibitor addition Oct. 18, RESULTS AND DISCUSSION Pipe-loop influent water quality. Temperature was the only influent water quality parameter that changed significantly during the 13-month testing period (Table 3 and Figure 2). Before the addition of any corrosion inhibitor, the ph of the influent to the pipe-loop system was 8.54 ± 0.26 (Table 3). However, the ph inside the orthophosphate loop (7.87 ± 0.23) was consistently less than the influent ph and the ph in the other loops (similar to influent) because of the low ph (<1) of the orthophosphate stock solution, which was made up of 36% phosphoric acid. Although not monitored as part of this study, DO in the full-scale treatment plant effluent ranged from 7 mg/l in summer to 13 mg/l in winter. The pipe loop influent residual chlorine concentration was relatively high (3.41 ± 0.42; number of samples [n] = 46) and stable, with no significant change after 8 h of stagnation time in the lead pipes (3.18 ± 0.37 to 3.45 ± 0.30; n = 13). The lead concentration was below the MDL (1 µg/l) for all influent samples. Lead. The lead pipes were new, and no corrosion inhibitor was added to any of the loops at the outset of the experiment. Thus it was not surprising that total lead concentrations for the 8-h stagnation time were similar in all of the loops and high ( µg/l) during this period (Figure 3). After corrosion inhibitor addition was started Oct. 18, 2001, lead concentration in the ortho-p loop declined rapidly from 151 µg/l October 16 to 65 µg/l October 26. Lead concentrations in the other loops did not exhibit a decrease until approximately nine weeks after corrosion inhibitor addition began. The elevated lead concentrations in the SnCl 2 loop during December (Figure 3) coincided with the initiation of chemical dosing and overfeeding of the SnCl 2 solution, resulting in a dosage that was tenfold higher than desired. It is not clear whether the elevated lead concentrations were attributable to the onset of chemical dosing or the elevated dosage. Lead concentrations in the SnCl 2 loop declined rapidly after this dosing problem was corrected. Lead concentrations in the control and SnCl 2 loops varied with the seasonal change in influent water temperature (compare Figures 2 and 3). Results of a linear regression analysis showed a positive correlation between influent water temperature and total lead concentration for the control and SnCl 2 loops but not the other loops (Table 4). Because of the influence of temperature on the lead and HPC results (Table 4), the experimental data were divided into two groups for subsequent comparisons and discussion: temperature <10 o C and temperature >15 o C. In addition, the analyses included only data collected after Jan. 3, 2002, when conditions in the system stabilized. 98 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

11 Total lead concentrations (mean ± standard deviation) ranged from 21.4 ± 6.45 µg/l (ortho-p) to 119 ± 20.6 µg/l (control) when the influent water temperature was <10 o C, and from 26.0 ± 11.3 µg/l (ortho-p) to 199 ± 50.6 µg/l (control) when the temperature was >15 o C (Table 5). The loop treated with ortho-p consistently had the lowest lead concentrations, whereas the highest lead concentrations were observed in the control and poly-p loops. A paired t-test analysis indicated that all of the loops in which a corrosion inhibitor was applied had significantly lower total lead concentrations than the control loop when the influent water temperature was <10 o C (p = to ) and when the temperature was >15 o C (p = to ) (Tables 6 and 7). From lowest to highest, the ranking of mean total lead concentrations in each of the pipe loops was as follows: (1) ortho-p, (2) SnCl 2, (3) blend, (4) poly-p, and (5) control. This ranking was consistent for both the cold and warm periods. Furthermore, a paired t-test analysis comparing the ortho-p and SnCl 2 loops indicated that the total lead concentration in the ortho-p loop was significantly lower (95th percentile confidence level) than in the SnCl 2 loop for both the cold and warm periods (Table 8). The dissolved lead concentrations for the 8-h stagnation time (Figure 4) followed the same trend as the total lead concentrations (Figure 3) except during the low temperature period, when the SnCl 2 loop had the lowest FIGURE 7 XRD pattern from the analysis of the interior surface of the lead pipe in the SnCl 2 loop Intensity counts 1, XRD X-ray diffraction Green vertical lines indicate lead (Pb) standard, and purple vertical lines indicate hydrocerussite [Pb 3 (CO 3 ) 2 (OH) 2 ]. 2 mean value. Dissolved lead concentrations ranged from 8.7 ± 4.7 µg/l (SnCl 2 ) to 96.2 ± 11.9 µg/l (control) when the influent water temperature was <10 o C and from 9.5 ± 7.2 µg/l (ortho-p) to 167 ± 44.7 µg/l (control) when the temperature was >15 o C (Table 5). Dissolved lead concentrations in all of the treated loops were lower than those in the control loop for both the cold water period (p = to 0.060) and warm water period (p = to ; Tables 6 and 7. The percentage of total lead in dissolved form for the SnCl 2 loop (47.4 ± 32.6%) and the ortho-p loop (45.3 ± TABLE 5 Pipe-loop effluent lead and HPC results (mean ± standard deviation) for both the cold and warm weather periods Parameter Total Pb Dissolved Pb Running Water Pb HPC* µg Pb/L µg Pb/L µg Pb/L cfu/100 ml Temperature Temperature Temperature Temperature Temperature Temperature Temperature Temperature <10 o C >15 o C <10 o C >15 o C <10 o C >15 o C <10 o C >15 o C Pipe Loop (n = 14) (n = 12) (n = 5) (n = 13) (n = 7) (n = 13) (n = 15) (n = 12) SnCl ± ± ± ± ± ± ± ± 50.4 Orthophosphate 21.4 ± ± ± ± 7.2 <1 < ± ± 99.0 Polyphosphate 105 ± ± ± ± ± ± ± ± 309 Blend 61.1 ± ± ± ± ± ± ± ± 305 Control 119 ± ± ± ± ± ± ± ± 69.6 HPC heterotrophic plate count, n number of samples, Pb lead, SnCl 2 stannous chloride HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

12 TABLE 6 Results of a paired t-test analysis comparing each treated pipe loop with the control pipe loop for samples in which the influent water temperature was <10 o C Parameter Total Pb Dissolved Pb Running Water Pb HPC (n = 14) (n = 5) (n = 7) (n = 15) Mean Mean Mean Mean Difference* Difference* Difference* Difference* Pipe Loop µg/l Pb p Value µg/l Pb p Value µg/l Pb p Value cfu/100 ml p Value SnCl Orthophosphate Polyphosphate Blend *Between treated and control HPC heterotrophic plate count, n number of samples, p probability, Pb lead, SnCl 2 stannous chloride 22.0%) was lower than for the poly-p loop (84.2 ± 21.5%), blend loop (77.4 ± 20.3%), and control loop (85.1 ± 12.9%). The significant percentage of particulate lead in the water samples, especially from the SnCl 2 and ortho-p loops, suggests that plumbosolvency models are likely to underpredict total lead concentrations overall and lead exposure. According to a lead solubility diagram developed by Schock (1989), for the concentration of ortho-p used (1 mg/l as P) and the alkalinity of SPRWS water (22 54 mg/l as CaCO 3 ), dissolved lead concentrations were predicted to be in the range of ~15 24 µg/l as Pb at 25 o C. The observed dissolved lead concentration for the warm water period (9.5 ± 7.2 µg/l) was slightly below the predicted range, likely because the water was often colder than 25 o C during this period, and the experimental system may not have achieved equilibrium before each sampling. The solubility diagram suggests that either SPRWS alkalinity must be <30 mg/l as CaCO 3 for a dose of 1 mg/l as P of ortho-p or the ortho-p dose must be increased to 2 mg/l or more to decrease dissolved lead concentrations below the AL. The former approach was not tested, and preliminary results from an evaluation of the latter approach suggested that increasing the ortho-p dose to 2 mg/l as P does not provide any additional reduction in lead concentrations over a dose of 1 mg/l as P (data not shown). Increasing the ph in the ortho-p loop to the range of the other loops (ph 8 9) either by using zinc ortho-p or by adding a base such as sodium hydroxide may have also helped to decrease lead concentrations to below the AL. The effects of such a ph adjustment were not investigated. In addition to the samples collected after the 8-h stagnation time, zero-stagnation-time (i.e., running water) samples were also collected from the lead pipes. At a flow rate of 1 gpm (0.06 L/s), the hydraulic residence time in the lead pipe was 15 s. The ortho-p loop consistently had the lowest running water lead concentrations (Figure 5): <1 µg/l for influent water temperatures <10 o C and <1 µg/l for influent water temperatures >15 o C. The control loop consistently had the greatest running water lead concentrations: 4.1 ± 0.7 µg/l for influent water temperatures <10 o C and 10.7 ± 2.4 µg/l for influent water temperatures >15 o C (Table 5). No running water lead samples (n = 135) exceeded the 15 µg/l Pb AL. HPC and coliforms. Figure 6 shows HPC values in the pipe loops as a function of time. For the low-temperature period from Jan. 15, 2002, to Apr. 29, 2002, HPC ranged from 0 to 157 cfu/100 ml but was typically <15 cfu/100 ml. The mean HPC values in each of the treated loops during this period (Table 5) were statistically similar to the control loop according to a paired t-test analysis (Table 6). The situation changed dramatically during the warmer months (temperature >15 o C); compared with the control, all of the loops treated with phosphate had significantly greater HPC values (p = to 0.048; Table 7. The greater nutrient levels in those loops, coupled with the higher water temperatures, appeared to stimulate bacterial growth. HPC levels in the SnCl 2 loop were significantly lower (p = 0.019) than in the control for the warm water period (Table 7). This suggests that the SnCl 2 inhibited the growth of heterotrophic bacteria in the system, which is consistent with the demonstrated toxic effects of the chemical on E. coli (Dantas et al, 1996). Although the iron pipes were inoculated with E. coli at the beginning of the experiment, coliforms were detected infrequently in the loops. In addition, fecal coliforms were never detected in any of the loops during the entire study. Except for one day in August 2002, coliform concentrations did not exceed 7 cfu/100 ml (data not shown). Coliform occurrence in the loops did not appear to correlate with HPC or phosphate. In fact, the 100 MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

13 TABLE 7 Results of a paired t-test analysis comparing each treated pipe loop with the control pipe loop for samples in which the influent water temperature was >15 o C Parameter Total Pb Dissolved Pb Running Water Pb HPC (n = 12) (n = 13) (n = 13) (n = 12) Mean Mean Mean Mean Difference* Difference* Difference* Difference* Pipe Loop µg/l Pb p Value µg/l Pb p Value µg/l Pb p Value cfu/100 ml p Value SnCl Orthophosphate Polyphosphate Blend *Between treated and control HPC heterotrophic plate count, n number of samples, p probability, Pb lead, SnCl 2 stannous chloride ortho-p loop had the lowest percentage of positive coliform samples (11.9%) of any of the loops with a high coliform concentration of 4 cfu/100 ml. Comparison of corrosion inhibitors. Among the corrosion inhibitors tested, orthophosphate (1 mg/l as P) consistently achieved the lowest lead concentrations (5 42 µg/l). Compared with the control, stannous chloride treatment (0.125 mg/l as SnCl 2 ) was also effective at reducing lead concentrations, but summer lead concentrations were typically five to seven times the AL. On the basis of this study s results, poly-p cannot be recommended for corrosion control because lead concentrations in the poly-p loop were only slightly lower than those in the control loop during most of the year. This conclusion is consistent with results reported by other researchers (e.g., Holm & Schock, 1991). The large percentage of ortho-p (71%) in the phosphate blend appears to be responsible for the decrease in lead concentrations in that loop, relative to the control and poly-p loops. On the basis of the lead results alone, the corrosion inhibitor recommended for lead corrosion control in this system would be ortho-p. Unfortunately, none of the corrosion inhibitors tested was able to consistently control total lead concentrations at or below the 15-µg/L AL for 8-h stagnation time samples. However, total lead concentrations for running water samples (i.e., no stagnation time) were consistently lower than the AL for all loops including the control loop. Therefore, flushing the pipes before withdrawing water for consumption is recommended for limiting lead exposure. In addition, the dissolved lead concentration for the ortho-p loop was typically below the AL (80% of samples). This suggests that point-of-use home filtration systems for particulate removal, in combination with ortho-p treatment, may be effective at maintaining lead concentrations below the AL. Furthermore, given the substantial amount of lead in dissolved form ( %) regardless of treatment and the failure of any corrosion inhibitor to consistently maintain lead concentrations below the AL, point-of-use treatment devices that incorporate a cation exchange system for lead removal (e.g., end-of-tap devices or water pitchers) are recommended for reducing lead exposures in homes with lead pipes, lead service connections, or both. The experiments described here were performed with a pipe-loop system constructed of new lead pipe. Although the 8-h stagnation lead concentrations in the SnCl 2 loop consistently exceeded the AL, the corrosion inhibitor appeared to be effective in the full-scale distribution system (data not shown). It appears that in the older lead pipes located in the full-scale system, the existing scale (in combination with inhibitor addition) was sufficient to maintain total lead concentrations below the AL. Therefore, use of new lead pipes in a pipe-loop-testing system may not be the best approach to mimicking full-scale system behavior. Corrosion control chemicals typically limit metal release by forming a low-solubility solid (such as hydroxypyromorphite) when ortho-p is used. Given the lack of information in the peer-reviewed literature on SnCl 2 for corrosion control, it is unclear how the chemical slows release of lead. To address this issue, XRD analyses were performed on samples of pipe removed from the SnCl 2 loop. The interior of the pipe surface contained a whitishyellow solid that consisted primarily of hydrocerussite (Figure 7). Although many peaks remain unidentified, no Sn-containing minerals were definitively detected. Possible explanations for the failure to detect Sn include the presence of a thin Sn-containing layer that was not detectable by the X-ray diffractometer, loss by dissolution after the pipe loop was shut down, or loss during removal and subsequent handling of the pipe section. Additional HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH

14 TABLE 8 Results of a paired t-test analysis comparing the SnCl 2 and orthophosphate pipe loops Parameter Total Pb Dissolved Pb Running Water Pb HPC Mean Mean Mean Mean Temperature Difference* Difference* Difference* Difference* oc µg/l Pb p Value µg/l Pb p Value µg/l Pb p Value cfu/100 ml p Value < (n = 14) (n = 5) (n = 7) (n = 15) > (n = 12) (n = 13) (n = 13) (n = 12) *Between SnCl 2 and orthophosphate HPC heterotrophic plate count, n number of samples, p probability, Pb lead, SnCl 2 stannous chloride research is needed to study the mode of action of SnCl 2 for control of lead release from lead pipes. Although ortho-p offered significant benefits in terms of lead corrosion control, its application greatly enhanced microbial growth in the pipe loops as measured by HPC. This is cause for concern because enhancing regrowth in the full-scale system could lead to the growth of opportunistic pathogens and interfere with pathogen monitoring (Geldreich et al, 1972). Although no increase in coliform concentrations following phosphorus addition were observed in this research, such an increase was observed in the full-scale SPRWS distribution system in spring This discrepancy was probably attributable to the lack of scale in the new ductile-iron pipes in the pipeloop system. Accumulation of scale and large tubercles in older iron pipes can afford bacteria (including coliforms) protection from fluid shear and disinfection, allowing them to survive for long periods of time and then proliferate when conditions allow. The SPRWS plant effluent has a moderate-to-high assimilable organic carbon concentration of 164 ± 24 µg/l and a high dissolved organic carbon concentration of 3 to 5 mg/l (Zhang et al, 2002). Thus, if ortho-p is selected for corrosion control, it may be necessary to reduce bioavailable carbon levels via biological filtration to minimize the potential for microbial growth. Although the addition of SnCl 2 reduced HPC relative to the control, the observed lead concentrations in the SnCl 2 loop of the pilot-scale system ( µg/l) were too high to recommend that chemical for lead corrosion control solely on the basis of these studies with new lead pipes. Nevertheless, given the potential for microbiological problems using phosphate-based inhibitors, SnCl 2 may be the only option for some utilities. CONCLUSIONS For the system and the water quality evaluated in this study, orthophosphate dosed at 1 mg/l as P provided REFERENCES Akers, C.J. & Fellow, R., Review of the Problem of Lead in Drinking Water. Envir. Health, 87:7:148. Ashby, J. & Tennant, R.W., Definitive Relationships Among Chemical and Mutagenicity for 301 Chemicals Tested by the U.S. NTP. Mutation Res., 257:229. Bailey, R.J. et al, Lead Concentration and Stagnation Time in Water Drawn Through Lead Domestic Pipes. Tech. Report TR 243. Water Research Centre, Medmenham, England. Cantor, A.F. et al, Use of Polyphosphate in Corrosion Control. Jour. AWWA, 92:2:95. Colling, J.H.; Whincup, P.; & Hayes, C.R., Measurement of Plumbosolvency Propensity to Guide the Control of Lead in Tap Waters. Jour. Inst. Water & Envir. Mgmt., 2:1:263. Cotton, F.A. & Wilkinson, G., 1980 (4th ed.). Advanced Inorganic Chemistry. John Wiley & Sons, New York. Dantas, F.J.S. et al, Genotoxic Effects of Stannous Chloride (SnCl 2 ) in K562 Cell Line. Food & Chemical Toxicol., 40:10:1493. Dantas, F.J. et al, Lethality Induced by Stannous Chloride on Escherichia coli AB1157: Participation of Reactive Oxygen Species. Food & Chemical Toxicol., 34:10:959. Dodrill, D. & Edwards, M., Corrosion Control on the Basis of Utility Experience. Jour. AWWA, 87:7:74. Economic & Engineering Services, Lead Control Strategies. AWWARF, Denver. Edwards, M. & McNeill, L.S., Effect of Phosphate Inhibitors on Lead Release From Pipes. Jour. AWWA, 94:1:79. Edwards, M.; Jacobs, S.; & Dodrill, D., Desktop Guidance for Mitigating Pb and Cu Corrosion By-products. Jour. AWWA, 91:5:66. Geldreich, E.E. et al, The Necessity of Controlling Bacterial Populations in Potable Waters: Community Water Supply. Jour. AWWA, 64:9:596. Holm, T.R. & Schock, M.R., Potential Effects of Polyphosphate Products on Lead Solubility in Plumbing Systems. Jour. AWWA, 83:7:76. Karalekas, P.C., Jr. et al, Control of Lead, Copper, and Iron Pipe Corrosion in Boston. Jour. AWWA, 75:2:92. Lee, R.G.; Becker, W.C.; & Collins, D.W., Lead at the Tap: Sources and Control. Jour. AWWA, 81:7: MARCH 2005 JOURNAL AWWA 97:3 PEER-REVIEWED HOZALSKI ET AL

15 significantly lower lead concentrations than the other corrosion control chemicals tested (SnCl 2, polyphosphate, and a blend of orthophosphate and polyphosphate). Nevertheless, none of these chemicals was able to consistently maintain 8-h stagnation time total lead concentrations below the 15-µg/L AL in the pipe-loop system. In addition to lead concentrations, the potential effects of corrosion control chemicals on microbiological water quality are also important. All of the phosphate-containing chemicals caused HPC levels to exceed those in the untreated control during the summer months, raising concerns about regrowth and increased coliform detections. On the basis of these HPC results, water utilities should exercise caution when adding corrosion control chemicals, so that reduced corrosion is not achieved at the expense of other water quality indicators. Therefore, utilities may want to consider using SnCl 2 if regrowth is exacerbated by the use of orthophosphate. ACKNOWLEDGMENT The authors thank St. Paul (Minn.) Regional Water Services (SPRWS) for providing financial support for this research project. In addition, the authors acknowledge the advice and assistance of SPRWS personnel including David Schuler, James Bode, Dan Finnegan, Jeff Wagner, and Ken Hoekstra. The authors also acknowledge the valuable input of Mike Semmens concerning the design and operation of the pipe-loop system. The authors thank Linda Sauer of the University of Minnesota Characterization Facility in Minneapolis for performing the X-ray diffraction analyses and helping with data interpretation. The authors also thank Carrie Bressler, Xiaojun Dai, Xin Ma, Eric Poppele, Li Zhang, and the anonymous reviewers for reviewing the manuscript and providing helpful comments and suggestions. ABOUT THE AUTHORS Raymond M. Hozalski (to whom correspondence should be addressed) is an associate professor in the Department of Civil Engineering, University of Minnesota, 500 Pillsbury Dr. SE, Minneapolis, MN, 55455; hozal001@umn.edu. Hozalski has a bachelor s degree in chemical engineering from Villanova University in Villanova, Pa., and master s and doctoral degrees from The Johns Hopkins University in Baltimore, Md. A recipient of the Abel Wolman doctoral fellowship, he has more than 14 years of experience in the water treatment field. Elizabeth Esbri- Amador was a master s candidate at the University of Minnesota during this study, and she is currently a project manager for McGhie Betts Environmental Services in Rochester, Minn. Che Fei Chen is a water quality specialist for St. Paul (Minn.) Regional Water Services. FOOTNOTES 1 AS-8111, ASI, Alameda, Calif. 2 Vulcan Lead Products, Milwaukee, Wis. 3 Lab Safety Supply, Janesville, Wis. 4 LMI, Milton Roy, Acton, Mass. 5 ASCO, Florham Park, N.J. 6 SLC 500, Allen Bradley, Milwaukee, Wis. 7 Nalge Nunc International, Rochester, N.Y. 8 Bruker-AXS, Madison, Wis. 9 JADE, Materials Data Inc., Livermore, Calif. 10 Fisher Scientific, Pittsburgh, Pa. 11 Spectra AA 300/400 Zeeman, Varian Inc., Palo Alto, Calif. 12 Analytical Products Group Inc., Belpre, Ohio 13 Millipore, Molsheim, France 14 Hach Co., Loveland, Colo. 15 Microsoft Excel 97, Microsoft, Redmond, Wash. If you have a comment about this article, please contact us at journal@awwa.org. McNeill, L.S. & Edwards, M., Phosphate Inhibitor Use at US Utilities. Jour. AWWA, 94:7:57. Murrel, N.E., Impact of Lead and Other Metallic Solders on Water Quality. Draft Rept., EPA Grant CR , Washington. Patterson, J.W. & O Brien, J.E., Control of Lead Corrosion. Jour. AWWA, 71:5:264. Renner, R., Plumbing the Depths of D.C. s Drinking Water Crisis. Envir. Sci. & Technol., 38:12:224A. Schock, M.R., Understanding Corrosion Control Strategies for Lead. Jour. AWWA, 81:7:88. Schock, M.R. & Wagner, I., 1985 (1st ed.). Corrosion of Solubility of Lead in Drinking Water. Internal Corrosion of Water Distribution Systems. AWWARF, Denver. Schock, M.R.; Wagner, I.; & Oliphant, R.J., 1996 (2nd ed.). Internal Corrosion of Water Distribution Systems. AWWARF, Denver. Sheiham, I. & Jackson P.J., The Scientific Basis for Control of Lead in Drinking Water by Water Treatment. Jour. Inst. Water Engrs. & Scientists, 35:6:491. Singh, I., Induction of Reverse Mutation and Mitotic Gene Conversion by Some Metal Compounds in Saccharomyces cerivisae. Mutation Res., 117:149. Standard Methods for the Examination of Water and Wastewater, 1995 (19th ed.). APHA, AWWA, and WEF, Washington. Tate, C.H. & Arnold, K.F., 1990 (4th ed.). Health and Aesthetic Aspects of Water Quality. Water Quality and Treatment (F.W. Pontius, editor). McGraw-Hill, New York. USFDA (US Food and Drug Administration), Canned Asparagus Deviating From Identity Standard; Temporary Permit for Market Testing. Fed. Reg., 68:117:36567, June 18. USFDA, Listing of Specific Substances Affirmed as GRAS Sec Stannous Chloride (Anhydrous and Dihydrated). Fed. Reg., June 25. Wong, C.S. & Berrang, P., Contamination of Tap Water by Lead Pipe and Solder. Bull. Envir. Contam. & Toxicol., 15:5:530. Wysock, B.M; Schock, M.R.; & Eastman, J.A., A Study of the Effect of Municipal Ion Exchange Softening on the Corrosion of Lead, Copper, and Iron in Water Systems. Proc AWWA Ann. Conf., Philadelphia. Zhang, M. et al, Biostability and Microbiological Quality in a Chloraminated Distribution System. Jour. AWWA, 94:9:112 HOZALSKI ET AL PEER-REVIEWED 97:3 JOURNAL AWWA MARCH