UNINTENDED CONSEQUENCES OF CHLORAMINE HIT HOME

Transcription

1 UNINTENDED CONSEQUENCES OF CHLORAMINE HIT HOME Marc Edwards, Becki Marshall, Yan Zhang and Yoon-Jin Lee Department of Civil Engineering Virginia Tech Blacksburg, VA ABSTRACT The known drawbacks of chloramination include nitrification, elastomer decay, and required pretreatment steps for fish culture and dialysis patients. To date, there has been no explicit consideration of adverse consequences of chloramination on property and water quality in buildings. Specifically, the effect of chloramine on re-growth of bacteria during stagnation, plumbing failures and lead leaching are poorly understood. Potential problems with bacterial re-growth can arise in building plumbing systems after chloramines decay and form free ammonia. Autotrophic nitrifying bacteria convert the free ammonia to nitrite and nitrate while creating organic carbon in the form of biomass and soluble microbial products. The levels of organic carbon created by nitrifiers are sufficient to support the growth of heterotrophic (HPC) bacteria. In one water tested with long stagnation times, HPC reached cfu/ml using chloramines, whereas HPC was 1000x less when chlorine disinfectant was used. The decay of chlorine does not release nutrients for bacterial growth a significant advantage relative to chloramine in situations with low flow and long detention times. Because the water within home plumbing is rarely sampled for bacteria, the true extent of the problem is not detected by routine distribution system monitoring using flushed samples. A switch to chloramine may increase lead leaching, brass failures and pinhole leaks under at least some circumstances. Of these problems, pinhole leaks and brass failures have the largest potential economic consequence. For instance, a single re-plumb can cost an individual consumer 500x more than the median annual projected cost of the Stage 2 regulation. The adverse public health impacts of mold growth from pinhole leaks, lead leaching and bacterial re-growth deserve consideration. While these problems may eventually prove to be rare events, they have significant consequences for the unfortunate consumers who are impacted. If these events prove to be widespread, alternatives to chloramine will become more attractive despite higher initial cost to utilities. KEYWORDS Re-growth, nitrification, brass failures, copper pinholes, lead, chloramine INTRODUCTION The United States Environmental Protection Agency (US EPA) Stage 1 and Stage 2 Disinfectants and Disinfection By-Products Rule (D/DBPR) has been actively negotiated since 1992 (AWWA, 2003a). The estimated benefits of the new Stage 2 regulation range from $0-986 million per year while projected costs are $54-64 million per year (Federal Register, 2003). In systems that require new treatments to meet Stage 2 requirements, the estimated increase in mean annual water bill per household is $8.38/year (Federal Register, 2003). To support the science behind these regulations, well over $100 million in research has been conducted to better define the risks from DBPs, microbial pathogens, and to support collection of occurrence and treatment data (US EPA, 1999).

2 Chloramine use as a secondary disinfectant is increasing since it is often the lowest cost means of complying with the Stage 1 and Stage 2 regulations. About 30% of surface water treatment plants currently use chloramine versus 20% in 1990 (AWWA, 2003b; San Francisco, 2004). As many as 65% of surface water treatment plants might convert to chloramines after compliance with Stage 2 D/DPR (AWWA, 2003b). This is a profound shift in secondary disinfection practice. Certain drawbacks associated with a switch to chloramination were anticipated. For instance, chloramines are a less effective disinfectant compared to free chlorine, but the D/DBPR was crafted to ensure "that the reduction of potential health hazards of DBPs does not compromise microbial protection (Federal Register, 2003)." Chloramines also degrade elastomers and require new pretreatments for fish culture or dialysis patients (AWWA, 2003a). Nitrification is also a concern when chloramine is used. The free ammonia present in water or that is released when chloramine decays is necessary for growth of autotrophic nitrifying bacteria according to the following representative overall equation written in units of mass (Grady et al., 1999): NH O HCO C 5 H 7 O 2 N NO H 2 O H 2 CO 3 Nitrites formed as an intermediary in the above reaction can consume chloramine disinfectant that controls growth of heterotrophic bacteria (AWWA, 2003b; Powell et al., 2004). If total chlorine residuals drop below about 2 mg/l, flushing of the distribution system and reduced water storage is often required to prevent nitrification and maintain adequate levels of chloramine (Powell, 2004; AWWA, 2003b). Dead ends of distribution systems which are poorly flushed are prime areas for nitrification (AWWA, 2003b). Despite all of these problems, the switch to chloramine immediately reduces the concentration of potentially carcinogenic disinfectant by-products in water, thereby reassuring many in the water industry that the change was in the public interest. The Law of Unintended Consequences guarantees that, despite good intentions and the best of anticipatory research, additional advantages and disadvantages of chloramine will be realized. This work describes three potentially serious problems arising from chloramine use including bacterial regrowth, lead leaching, and plumbing failures. The most serious manifestation of each problem occurs in premise plumbing, defined herein as the portion of the water distribution system in buildings and homes. Premise plumbing is characterized by large pipe surface area to water volume ratio (Table 1), regular periods of stagnation, and variable microclimates of temperature and redox that provide ecological niches for a wide diversity of microorganisms (Edwards et al., 2004a, 2003). Plumbing code assures that every building is a dead end," and water commonly sits stagnant in the pipes and onsite storage such as water heaters for hours or days before use. Historically, the property (or service line) that demarcates the main distribution system from building plumbing systems has not been crossed for regulation. The notable exception is the lead and copper rule (LCR), which has reduced corrosivity of the public water supply in regard to leaching of lead or copper. However, homeowners still bear final responsibility to protect themselves from excessive lead or copper exposure (Edwards et al, 2004b), and there are no maximum contaminant levels enforced at the tap in the United States for water held stagnant within premise plumbing. This is surprising to many consumers since this water is frequently consumed.

3 Table 1. Characteristics of U.S. Public and Private Transmission Systems (after M. Edwards, D. Bosch, G.V. Loganathan, A. M. Dietrich, 2003; Edwards, 2003) Characteristic Public Infrastructure Private Infrastructure Replacement Value 0.6 trillion $US > 0.6 trillion $US Pipe material Cement, Ductile iron, Plastic, Cast iron Copper, Plastics, Galvanized iron, Stainless steel, brass Total Pipe length (US) 0.97 million miles > 6 million miles Approx. Pipe Surface 0.26 cm 2 /ml* 2.1 cm 2 /ml* per Volume Water* Complete stagnation Relatively rare Frequent Disinfectant Residual Usually present Often completely absent after stagnation Flow Relatively consistent On/Off Temperature 0-30 C C Maximum cost over 30 yrs per consumer $ US Easily up to $25,000 per homeowner or millions for buildings Advocacy Water industry (WIN) # None #Water infrastructure network *Assumed 15.2 cm diameter for mains and 1.9 cm diameter for home plumbing Re-growth in Premise Plumbing Research for the D/DPR has not explicitly considered problems of microbial growth that can occur during stagnation or storage in premise plumbing. Previous research on re-growth of bacteria in water mains is a useful starting point to understanding problems that occur in these situations. To over simplify the conventional wisdom, the following equation is useful: Organic Carbon + Nitrogen + Phosphate + Trace Nutrients + O 2 Heterotrophic Bacterial Re-growth If high levels of chlorine disinfectants are present, re-growth via the above equation can be controlled even if all nutrients are present in adequate supply. But if disinfectant concentrations decrease to low levels or disappear, bacterial growth will proceed as long as all key nutrients are available. The first nutrient that is reduced to relatively low levels (and which therefore limits the maximum number of bacteria that can grow) is termed the limiting nutrient. A water is termed microbiogically stable if nutrients are present at low enough levels to prevent growth of undesirable levels of bacteria (Rittman et al., 1984; Sathasivan et al, 1999). This limiting nutrient approach is utilized successfully in many European countries to safely distribute drinking water with little or no disinfectant. In US practice, however, due to highly variable nutrient requirements amongst bacteria of interest (e.g., Rice et al., 1991) and a reliance on disinfection to control re-growth, such concepts do not always receive explicit consideration. Nonetheless, with some exceptions, growth of bacteria in typical US water supplies is limited by the concentration of organic carbon, and utilities therefore try to minimize introduction of organic carbon to the distribution system (LeChevalier et al, 1991; EPA, 2002; AWWA; 2003). A superficial evaluation suggests that addition of ammonia to water would not markedly increase potential re-growth, since nitrogen is already present in excess.

4 A closer examination reveals a fundamental flaw with this analysis. As described in the earlier equation, nitrification occurs autotrophically with production of organic carbon at a ratio of 87 ug C/mg NH 3 -N consumed. This organic carbon is in the form of new cell biomass and soluble microbial products that can be efficiently utilized by heterotrophic bacteria (Kindaichi et al., 2004; Rittman et al., 1994). Consequently, heterotrophic bacteria are found in nitrifying cultures without any external organic carbon supply (Okabe et al., 2002). In practice, production of organic carbon in nitrifying bacteria new cell biomass is in the range of ug C/mg NH 3 -N (Tarre et al., 2004), but this underestimates total organic carbon created from nitrification since it does not consider soluble microbial products (Rittman et al., 1994). A typical 4 mg/l (as Cl 2 ) level of chloramine disinfectant can release 1 mg/l free NH 3 -N as it decays. Using the lowest yield cited above, 25 µg/l organic carbon is created if 1 mg NH 3 -N is utilized in nitrification. In a study of Dutch distribution systems with no or low concentrations of disinfectants, less than 10 µg/l of organic carbon (AOC) was sufficient to limit re-growth. AOC levels below 50 ug/l have been cited as desirable to control coliforms in disinfected U.S. distribution systems (LeChevalier et al, 1991). Reasonable correlations have been cited between AOC and bacteria levels in distribution systems that did not use chloramines (LeChevalier et al, 1991; van der Kooij et al., 1992), whereas no correlation was reported in one system using chloramine (Gibbs et al., 1993). It is possible that growth of bacteria in systems with chloramine is less dependent on AOC in the water supply, since organic carbon can be produced during nitrification. The potential problem with organic carbon production is magnified considered cycling of nitrogen in reactions with plumbing materials. For instance, nitrate produced by nitrifiers can be converted back to ammonia in reactions with metallic iron and lead surfaces (Westerhoff et al., 2003; Edwards et al., 2004b; Dudi, 2004). Additional organic carbon would be created each time ammonia is cycled by nitrifiers. In water main distribution systems, utilities often start to have nitrification problems below 2 mg/l (as Cl 2 ) chloramine residual (Yang et al., 2004). Surveys of utilities using chloramines have revealed that 63% report some degree of nitrification based on distribution system monitoring (Wilczak et al., 1996). Since premise plumbing systems are dead ends, with long detention times, higher temperatures and lower chlorine residuals than in the distribution system itself, utilities may not be detecting the true extent of the nitrification problem. Indeed, while routine bacterial monitoring by utilities often uses taps located within buildings, most standard protocols require flushing for 3-5 minutes before collecting samples. The water that is sampled is therefore representative of bacterial concentrations in the water mains. Three additional areas of concern are noteworthy in regard to premise plumbing. First, recent experience with packed bed granular activated carbon (GAC) for chloramine removal has indicated that once nitrifying bacteria are established (Fairey et al., 2004), their growth cannot be controlled even with chloramine residuals as high as 4 mg/l. It is therefore highly likely that nitrifiers will grow on GAC used in whole house or point of use filters when chloramines disinfectant is used. If so, these devices could serve to generate organic carbon that would support high levels of bacteria re-growth. Secondly, even if re-growth in homes was not a significant public health problem at one time, changes in consumer behavior might make such risks more significant. Specifically, there has been a noteworthy trend in the U.S. to decrease water heater temperature to minimize scalding and save energy. Australian standards require hot water systems to be maintained at a minimum of 60 C to control Legionella, with installation of a special mixing valve at points of delivery to prevent scalding (Spinks et al, 2003). In contrast, the U.S. EPA and utilities in the US recommend that consumers

5 reduce their water heater temperature to 48 C to save energy and prevent scalding (US EPA, 2004). Indeed, many water heaters in the U.S. have safety devices designed to prevent higher temperatures. There is concern that the trend to lower temperatures will increase problems with re-growth relative to higher temperatures that were once present, especially for legionella and mycobacterium avium (Borella et al., 2004; EPA, 2002). Increased use of low flow showerheads might also alter consumer exposure to endotoxins and bio-aerosols. Finally, the decreased reactivity of monochloramine versus free chlorine is often considered a major advantage, since it allows a disinfectant residual to be delivered to extreme ends of the distribution system. This allows for better disinfection of iron surfaces than free chlorine under flowing water conditions (LeChevalier et al, 1990). Considering the analysis presented earlier, if a circumstance ever arose in which monochloramine decayed more quickly then chlorine, a purported benefit would quickly become a disadvantage. That is, when monochloramine decays, it creates free ammonia which can lead to creation of organic carbon via nitrification. In contrast, complete loss of free chlorine residual forms chloride that does not increase re-growth. Corrosion in Premise Plumbing EPA cost:benefit analysis and research on Stage 1 and Stage 2 has not considered corrosion problems that occur within premise plumbing. Premise plumbing is frequently comprised of PEX, copper pipe, lead pipe, lead solder and leaded brass not regularly used in the main distribution system. The estimated replacement value of private plumbing systems in the US is on the order of $1 trillion dollars (Edwards et al., 2004a). A small increase in the corrosivity of water towards private plumbing materials could cost consumers billions of dollars per year (Edwards et al., 2004a, c). The likelihood of increased copper pinhole leaks resulting from enhanced coagulation and other treatments was anticipated well before Stage 1 was implemented (Edwards et al., 1994), and these treatments may be contributing to several recent outbreaks of pinhole leaks in the U.S. (Edwards et al., 2004c). At one large utility that aggressively addressed a pinhole leak problem experienced by its customers and sought to define its extent, as many as 27-53% of homes are thought to have developed at least one pinhole leak (Edwards et la., 2004c). A recent review (Edwards et al., 2004b) described earlier studies in which changes from chlorine to chloramine disinfection caused brass failures within buildings (Larson et al., 1956; Ingleson et al., 1949; anonymous, 1951). Nitrogen species are also involved in stress corrosion failures of brass (Edwards et al., 2004b). If the new regulations were to cause increased pinhole leaks in copper tube or brass failures, the cost of the new regulations to individual consumers may be higher than anticipated. While the cost of finding and repairing a single leak can range from $75-500, after two leaks plumbers recommend a complete re-plumb at a typical cost of $1,000-10,000. A $200 leak repair and a $5,000 re-plumb exceeds the median estimated annual cost of the stage 2 regulation by factors of 24 and 600, respectively. If water damages, increased insurance costs and possible mold remediation were considered, the disparity would be even greater. Newly re-plumbed systems have also been documented to fail in as little as 6 months in circumstances where the water was highly aggressive, so this is not always a one time expense. Recent changes in building practice or water treatment practice might also create premise corrosion problems in systems using chloramines where they would not have existed previously. For instance, to reduce water loss, hot water recirculation systems are now commonplace in many newer buildings. If

6 chloramines were shown to cause problems with hot water recirculation plumbing, unreasonable rates of failure might result in those instances. Likewise, increased use of aluminum anodes in water heaters, release of aluminum to water after pipeline rehabilitation by cement lining, lower NOM, higher phs, and even coagulant selection can dramatically alter pinhole leak frequency (Edwards et al., 2004c; Rushing et al., 2004). Edwards et al., 2004b also pointed out previous cases in which chloramines increased lead leaching from lead solder and brass materials relative to free chlorine (Portland, 1983; Lin et al. 1997). Indeed, work conducted more than a century ago found that nitrification increased leaching of lead to potable water (Garret, 1891). If use of chloramines increased leaching of lead or copper, a potential detriment to public health would be realized that would reduce the net benefits of the regulations. Increased growth of mold resulting from leaks can also have human health implications. PROBLEMS ARISING AT LEAST IN PART FROM CHLORAMINATION To fully assess the benefits and detriments of a given treatment change such as chloramination, it is necessary to consider long term impacts that occur in premise plumbing. The sections that follow describe three case studies of problems that are believed to arise in part from chloramination. Washington, D.C.: Lead corrosion, pinhole leaks and increased re-growth The water for Washington DC is produced by the Washington Aqueduct and sent to consumers through transmission lines owned by the DC Water and Sewer Authority (DC WASA). The changeover from free chlorine to chloramines in November of 2000 was initially considered highly successful based on traditional bacteria and disinfection by-product monitoring. Within a few months, however, non-traditional measurements revealed serious problems with the changeover. Specifically, a survey of plumbers working in areas served by the Washington Aqueduct (Edwards et al., 2001a) revealed an increase in copper tube failures in hot water recirculation lines. Later sampling also revealed dramatic increases in lead leaching to water in some areas of the WASA system (Edwards et al., 2004b). Subsequent laboratory and field testing leaves little doubt that the switch from chlorine to chloramine was the cause for increased lead leaching (Edwards et al., 2004b). The clear implication is that the corrosivity of the water was increased by the switch to chloramine. Samples of water were also collected from the WASA distribution system on a regular basis during a time period that chloramine was dosed (without phosphate inhibitor) and transported to our laboratory at Virginia Tech for use in various experiments. To examine the effects of chloramine versus chlorine on bacterial re-growth during stagnation in copper tubes, one water was "boosted" to a 3.7 mg/l chloramine residual (as Cl 2 ) using free chlorine. Typical free chlorine was < 0.1 mg/l, total ammonia was 1.06 mg/l, and free ammonia was 0.3 mg/l in this sample. Another portion of the water was created with free chlorine disinfectant by breakpoint chlorination to 0.3 mg/l free chlorine. Because breakpoint chlorination completely destroys ammonia and forms chloride, additional chloride was added to the chloraminated water using NaCl. Thus, the main difference between the two waters is presence of chloramine versus free chlorine disinfectant. The ph was adjusted to 7.4 using CO 2 or NaOH as necessary. Samples of each water were placed into 3 foot sections of type M copper tube. The water was held within each tube for three different stagnation times including 8 hours, 24 hours and 3 days, before the water was replenished as would occur during flow. Each condition was tested in triplicate making 18

7 tubes total. The typical temperature was 27 +/- 4 C as per natural fluctuations that occur within a building in July. After two weeks, the water from the tubes after the indicated stagnation time was sampled for HPC bacteria and TOC (Figure 1). Figure 1. TOC measured after stagnation in rigs using chlorine and chloramine (above), and log HPC count (below). Error bars denote 90% confidence interval. TOC (mg/l) Chlorine Chloramine Initial TOC Stagnation Time (hours) log HPC (cfu/ml) Stagnation Time (hours) In the case of 8 hour stagnation, water from the copper pipe dosed with 3.7 mg/l (total chlorine as Cl 2 ) chloramine had slightly lower levels of HPC than did the system dosed with 0.3 mg/l (as Cl 2 ) free chlorine. However, in pipes with a 24 and 72 hour stagnation, bacteria levels increased markedly in the pipe dosed with chloramine. Indeed, HPC from the pipe after 72 hours stagnation approached 10 6 cfu/ml. In the corresponding condition with free chlorine, HPC never rose above 10 3 cfu/ml. TOC in the water decreased slightly with time in the system dosed with chlorine as would be expected based on previous research (Edwards et al., 2003b), but increased slightly with time in the system dosed with chloramine. In the case of 72 hours stagnation, TOC levels emerging from the pipe with chloramines were higher than those going into the pipe at > 90% confidence. One possible explanation for the increased organic carbon is autotrophic bacterial growth in the system with chloramines. This also can explain the high levels of HPC bacteria. The above experiment was continued for an additional week with dosing of 1 mg/l phosphate (as P) inhibitor. Bacterial levels changed only slightly in the system dosed with chlorine (Figure 2). With an 8 hour stagnation time, bacteria levels were also low when chloramine was used. But with 72 hours stagnation, HPC counts rose to 5 x 10 6 for the system dosed with chloramine plus phosphate. TOC

8 levels rose from an initial value of 1.8 to 2.4 mg/l during stagnation in the system dosed with chloramine after 72 hours stagnation, but were always less than 1.7 mg/l when chlorine was used. The 0.6 mg/l increase in organic carbon is 7 times higher than that expected based on a yield of 87 ug C/mg NH 3 -N consumed, and it may reflect sloughing of pre-existing biofilms or cycling of nitrate to ammonia. In any case, the re-growth situation in stagnant pipes is markedly different than in flowing distribution systems. Chlorine did not cause a problem with bacteria levels during long term stagnation in this altered water sample from DC WASA, whereas chloramine did so. The creation of nutrients during decay of chloramine is a likely explanation for the large disparity in performance. While the effect of phosphate is preliminary, it appears to have widened the disparity between the two disinfectants. As a point of practical comparison, a few months after DC WASA started dosing of orthophosphate to control lead leaching, coliform standards were exceeded for the first time since 1996 (Cohn, 2004). It is possible that increased re-growth was contributing to the problem in the water distribution system. A final test was conducted in which the stagnation time was abruptly changed to 8 hours for all conditions using chloramine. For the duration of this test (2 weeks) in the pipe that originally was stagnant for 72 hours, final chloramine residual was always below 0.25 mg/l and HPC was always above 10 5 cfu/ml. In the pipe consistently held at 8 hours stagnation, the residual was always above 2 mg/l. Biofilms formed on the pipe during a period of stagnation could not be readily controlled with restoration of the chloramine residual, consistent with the observation of Fairey et al., 2004 for GAC. The implication is that a period of a few days stagnation in a building might set up a detrimental condition that increases bacteria in the water for an extended period of time. Figure 2. TOC measured after stagnation in rigs using chlorine and chloramine (above), and HPC count (below). Orthophosphate was dosed at a level of 1 mg/l PO 4 -P. Chlorine Chloramine TOC (mg/l) Initial TOC log HPC (cfu/ml) Stagnation Time (hours)

9 Maui Hawaii: Bacterial re-growth in homes and lead corrosion problems. The Maui Department of Water Supply (DWS) system had been using chloramines since In the spring of 2001, the Maui Department of Water Supply (DWS) exceeded the EPA LCR action limit for lead. A decision was made to control lead corrosion by dosing of zinc orthophosphate in June of While the addition of the phosphate decreased lead levels to below the EPA action limits in many cases, concerns emerged amongst consumers regarding possible health effects arising directly or indirectly from dosing the phosphate inhibitor. Specifically, the residents believed that problems of itchy skin, rashes, eczema, blurring and burning eyes, respiratory problems and throat irritation were temporally linked to dosing of the zinc orthophosphate corrosion inhibitor to the water supply. A randomized blinded survey supported the opinion that the symptoms may have been associated with dosing of the phosphate to the water (Rohner et al., 2004). On April 10, 2003, a decision was made to stop dosing zinc orthophosphate and to switch to orthophosphate alone. Later testing indicated that some areas in the distribution system using chloramines and phosphate still did not pass the LCR monitoring-- the 90th percentile lead was 41 ppb. The consumers expressed continued concern regarding the safety of the water. Some of the problems described by residents were not inconsistent with emerging understanding of problems such as hot tub rash or hot tub lung. Since these problems are caused by microbes (CDC, 2003; Gorman, 2002; Mayo Clinic, 2004; Tobin, 2002), and the original water had undetectable (< 3 ppb) levels of orthophosphate, it was deemed possible that dosing of phosphate triggered a problem with bacterial regrowth. A decision was made to review distribution system data, conduct some sampling in homes, and to execute pipe rig testing for lead corrosion and bacteria re-growth. A review of the distribution system bacterial monitoring data did not reveal marked increases in sampled HPC bacteria after the switch to phosphate, but these samples are all collected after standard flushing protocols. The first draw sample is the first liter of water that flows from the consumers tap after an overnight stagnation event. One first draw sample was collected for HPC s from the cold water line in a Maui home reporting health problems and it contained 3 x 10 6 cfu/ml. After flushing the tap 3 minutes the HPC s dropped to 1100 cfu/ml. Clearly, the extent of the microbial problem in homes was not being detected by traditional monitoring. Additional field-testing of first draw samples was conducted for microorganisms that can influence corrosion using biological activity (BART) test kits. The kits included those for acid producing bacteria (APB), heterotrophic aerobic bacteria (HAB), sulfate reducing bacteria (SRB), slime forming bacteria (SLYM), and nitrifying bacteria. These are semi-quantitative tests for presence of bacteria thought to influence corrosion. Field tests were also conducted for ph, free and total chlorine residuals. When chloramine and phosphate were in use, the homes in which consumers were complaining of health problems tended to have chronically low or undetectable total chlorine residuals and high levels of bacteria (> 10 4 cfu/ml) as quantified using the BART test kits. In some cases, the low chlorine residuals were at least partly due to private water storage and use of in-home granular activated carbon filters, but in other instances the homes with low residuals were simply at distant points of the distribution system. Since the orthophosphate had not solved the problems with lead, a decision was made to stop the dosing of orthophosphate and instead control corrosion by addition of soda ash with a target ph of 8.6. A shift was also made to switch to free chlorine disinfectant to control the bacteria. A subset of homes

10 was re-sampled using BART testing to make order of magnitude estimates of bacterial numbers present in collected samples (Figure 3). The sampled water was allowed to sit in the pipes at least 6 hours but not more than 12 hours before the sample was collected. Figure 2. Comparison of acid producing bacteria levels in first draw samples when chloramine and phosphate were in use versus when only chlorine was used (above). Levels of slime forming bacteria detected in samples (lower). 12,000 10,000 APB (cfu/ml) 8,000 6,000 4,000 2,000 0 Chloramine + P Cl2 + No P Chloramine + P Cl2 + No P Slyme (cfu/ml) Sample Location No or very low levels of free chlorine residual were detected after 5 minutes of flushing in sample location 1 and 3. It is useful to discuss some specifics related to these homes. Sample location 1 was using a whole house activated carbon filter (during sampling in Part 1) when no free or total chlorine was detected, but they had switched to a simple sediment filter during sampling reported in this work. The Cl 2 residual increased to only 0.2 mg/l after 5 minutes of flushing in this home. It is unclear whether the Cl 2 residual was being destroyed or influenced by the sediment filter. At sample location 3 no Cl 2 residual was detected even after 5 minutes flushing. Sample location 3 is on a private system which has substantial storage between the home and the DWS water main. Analysis of this and other samples proved that it is highly unlikely a significant chlorine residual ever reaches the home plumbing even when chloramine is used. The homeowner prefers the taste of water without such a residual. Slime forming bacteria had been detected at very high concentrations in homes when phosphate and chloramine were used as disinfectant (Figure 2). The concentration of these bacteria were two orders of magnitude (100X) lower during follow up sampling when chlorine disinfectant was used without phosphate. In fact, only sample location 1 with the home sediment filter had any detectable slime forming bacteria when chlorine was in use without phosphate. Results at sample location 3 are also noteworthy. Even though detectable chlorine disinfectant never reached this home (with chloramines

11 or chlorine), the high levels of slime forming bacteria were no longer present in the first draw sample. This is most likely due to removal of ammonia (switch to chlorine) and phosphate (change in corrosion control) from the water. The consumer health complaints and concentrations of lead leached to water decreased when free chlorine was used without phosphates. To better understand the problem with re-growth, pipe rig tests were conducted at the treatment plant. The treated water was adjusted to eight levels of ph, disinfectant and phosphate levels (Table 2). For comparison, some testing was done with modifications of waters listed in Table 2, but without chlorine residual disinfectant. Those conditions are designed to be representative of re-growth conditions that might occur in homes near the end of the distribution system after chlorine residuals have disappeared. Table 2. List of water conditions tested. Each disinfectant was present at a level of 2.4 mg/l total chlorine (as Cl 2 ). Orthophosphate was present at 1 mg/l as PO 4 -P. ph Water Quality Condition 7.7 Chloramine 8.3 Chloramine 8.9 Chloramine 9.5 Chloramine 7.7 Chloramine and Phosphate 7.7 Free Chlorine 8.9 Free Chlorine PVC and copper pipes were exposed to each water for 7 weeks with a 48 hour stagnation time. Thereafter, water samples from each rig were tested in triplicate for HPCs. The 48 hour stagnation time between water changes is likely to be longer than is typically found in many homes, but not unheard of given weekend trips. On the other hand, the chlorine residual levels tested were relatively high compared to those present in distant parts of the distribution system. Laboratory testing in PVC pipes supported the hypothesis that the addition of phosphate caused increased levels of bacteria when chloramines was used. Of the seven waters described in Table 2 (seven bars on the left side of Figure 4), the worst bacterial problem was in the system at ph 7.7 with phosphate and chloramine. This is the condition most representative of the water during the time that residents were complaining of health ailments. The tests designed to replicate conditions without chlorine residuals are also instructive (four right bars of Figure 4). Even without the addition of phosphate or ammonia from chloramine, there was rapid growth of bacteria in this water once chlorine residual is gone. Total chlorine measurements after 48 hours stagnation at ph 7.7 in the PVC pipes were also highly instructive. With free chlorine the residual was 1.02 mg/l, with chloramine the residual was only 0.51 mg/l, and with chloramine and phosphate the residual was below detection. The same trend was apparent in other rigs in which chloramine and chlorine were dosed at the same ph. Clearly, chemical and/or biological reactions during stagnation were causing chloramine to decay more rapidly than free chlorine. Given that chloramine is assumed to be much more stable than chlorine under nearly all circumstances, this is deserving of immediate follow-up study.

12 Figure 4. HPC counts in Maui water tested in PVC pipes. The seven conditions at the left (red) had a chlorine or chloramine residual as per Table 2. The four samples at the far right (blue) did not have a chlorine residual. Actual bacterial levels for samples on the right were likely to be higher than indicated, since they were above the maximum test range of 50,000 cfu/ml Olinda Treatment Facility Average HPC count (CFU/mL) Action Limit (500 CFU/mL) 1 ph 7.7 Chloramine ph 8.3 Chloramine ph 8.9 Chloramine ph 9.5 Chloramine ph 7.7 Chloramine + Phosphate ph 7.7 Free Chlorine ph 8.9 Free Chlorine ph 7.7 NH4 + Phosphate ph 7.7 Phosphate ph 7.7 NH4 ph 7.7 control Much lower levels of bacteria were found in water from copper pipes in the Maui water, as long as ph was maintained between 8.3 and 8.9 (Figure 5). At lower and higher ph ranges, there was a significant problem with HPC bacteria. Free chlorine also led to significantly lower bacterial counts at ph 7.7 and 8.9 when compared to the condition at the same ph using chloramine. As was the case in the Washington D.C. water, chloramine can cause increased problems with re-growth during long stagnation events relative to chlorine. Figure 5. HPC counts for Maui (Olinda) water from the copper pipes HPC Count (CFU/mL) Olinda Water action limit (500 ph 7.7 Chloramine ph 8.3 Chloramine ph 8.9 Chloramine ph 9.5 Chloramine ph 7.7 Chloramine+ Phosphate ph 7.7 Free Chlorine ph 8.9 Free Chlorine

13 It is uncertain if the bacteria arising from re-growth in homes pose a significant public health risk. It is understood that generic measures such as HPC are not sufficient to judge water safety (WHO, 2002), and studies of water with high bacteria after re-growth on GAC and other point of use devices have generally shown relatively low health risks (LeChevalier et al, 2004; Camper et al., 1985; Rollinger et al, 1987; Caulderon et al., 1987). Indeed, levels of heterotrophic bacteria commonly found in potable water usually represent less than 4% of the total human dietary intake (Stine et al., 2005). But it is also increasingly acknowledged that certain opportunistic pathogens such as Legionella, Mycobacterium Avium and Pseudomonas Aeruginosa do pose a significant public health threat (EPA, 2002). The risk from mycobacterium and legionella might be greatest from airborne bio-aerosols formed in hot water systems (EPA, 2002; Borella et al., 2004). England has started to implement first draw sampling for bacteria in public buildings, and the initial results do not indicate a significant problem relative to the levels of bacteria detected after flushing (Jackson et al., 2004). This might indicate that the high levels of bacteria cited in this work are extraordinarily rare, occurring only under unusual circumstances of higher temperature, infrequent water use and distance from the treatment plant. Or, it might be that the results from England are typical of utilities using chlorine in cooler climates. Other researchers have hinted at problems similar to those cited herein for Maui and in pipe rig testing using Washington D.C. water. For instance, Murphy et al. (1997a) noted rapid chloramine decay and nitrification in copper pipes, but this did not produce problems with culturable heterotrophic organisms (Murphy et al., 2003b). Powell et al. (2004) cited situations in which chloramine decay was more rapid than had previously been observed for free chlorine at a Florida utility. Additional research and vigilance is necessary, since it would be undesirable to worsen control of bacterial growth even in a small subset of homes and buildings susceptible to such problems. Anonymous Utility: Pinhole leaks and Brass Failures Within 18 months after switching to chloramine from chlorine, one utility detected marked increases in pinhole leaks and brass failures. Due to legal concerns this utility does not wish to be identified, but it did agree to allow pictures of failed specimens and general observations to be shared (Figure 6). Failures generally occurred in areas of relatively high water velocity and wall thinning was apparent. Some brass failures are by the same wall thinning mechanism, but other failures are due to stress corrosion cracking. CONCLUSIONS The adverse consequences of Stage 1 and Stage 2 regulations on home plumbing have not been given adequate consideration given the high economic value of the asset and possible impacts on individual consumers. Lead leaching, copper leaching and pinhole leaks can be costly and pose a public health problem. Chloramine is proven to have advantages for disinfection during flow conditions and on rapidly corroding steel surfaces. But if chloramine residual is lost in areas of the water distribution system or in buildings, growth of nitrifying bacteria can create organic carbon that may spur re-growth of heterotrophic bacteria. If enough organic carbon is produced, and further considering the obvious availability of nitrogen, undesirable levels of bacteria can grow in the water held in pipes during periods without flow. Complete loss of chlorine residual does not introduce nutrients to the water, which provides a relative performance advantage versus chloramine.

14

15 Figure 6. Representative failures from buildings after a utility switched to chloramine Copper thinning Brass Stress Crack Brass and Type K Copper Failure ACKNOWLEDGEMENT The authors acknowledge the financial support of the National Science Foundation under grant DMI Opinions and findings expressed herein are those of the authors and do not necessarily reflect the views of the National Science Foundation. REFERENCES Andrzej, W.; Jacangelo, J.G.; Marcinko, J.P.; Odell, L. H.; Kirmeyer, G. H.; Wolfe, R. L. (1996) Occurrence of Nitrification in Chloraminated Distribution Systems. Jour. AWWA, 88(7):74. Anonymous. (1951) Review of Current Investigation No 8: Waterworks Fittings. Research Group Report. Jour. Inst. Water Engrs. (Br.), 5:700.

16 AWWA (2003a). Summary of the Proposed Stage 2 Disinfection Byproducts Rule. Public Meeting Web Conferences. October 9 and 16, Accessed 1/3/2005 at Stage2 Presentation webcast ( ).ppt. AWWA (2003b). Nitrification. Accessed 1/1/2004 at Borella, P.; Montagna, M.T.; Roman-Spica, V. (2004) Legionella Infection Risk from Domestic Hot Water. Emerging Infectious Diseases., 10(3), 457. Portland Bureau of Water Works (1983) Internal Corrosion Mitigation Study Addendum Report. Portland, Oregon. Calderon, R.L.; E. W. Mood. (1987) Bacteria Colonizing Point-of-Use, Granular Activated Carbon Filters and Their Relationship to Human Health. EPA CR Camper, A.K.; LeChevallier, M.W.; Broadaway, S.C.; McFeters, G. A. (1985) Growth and Persistence of Pathogens on Granular Activated Carbon Filters. Appl. Environ. Microbiol., 50 (6), CDC (2003) What all pool staff should know about recreational water illness. A rash of hot tub rashes. Accessed 8/22/2004 at Cohn, D. Bacteria put D.C. Water in Breach. Washington Post. Friday, September 24, Page B01. Dudi, A. (2004) Reconsidering Lead Corrosion in Drinking Water: Product Testing, Direct Chloramine Attack and Galvanic Corrosion. Virginia Tech MS Thesis. Edwards, M. Corrosion Control in Water Distribution Systems. One of the Grand Engineering Challenges for the 21st Century. Edited by Simon Parsons, Richard Stuetz, Bruce Jefferson and Marc Edwards. Water Sci. and Technol. p. 1-8 V. 49, N. 2 (2004a). Edwards, M.; Abhijeet, D. (2004b) Role of chlorine and chloramine in corrosion of lead-bearing plumbing materials. Jour. AWWA., 96 (10), 69. Edwards, M.; Rushing, J. (June 27, 2001a) Investigation of Copper Pitting Corrosion in Homes of Washington Suburban Sanitation Commission Customers. Final Report. Edwards, M.; Boulay, N. (2001b) Organic Matter and Copper Corrosion By-Product Release: A Mechanistic Study. Corr. Sci., 43 (1), 1. Edwards, M.; Bosch, D; Loganathan, G. V. Dietrich, A. M. (May, 2003). The Future Challenge of Controlling Distribution System Water Quality and Protecting Plumbing Infrastructure: Focussing on Consumers. IWA Leading Edge Conference. Noordwijk, Netherlands. Edwards, M.; Rushing, J.C.; Kvech, S; Reiber, S. (2004c) Assessing Copper Pinhole Leaks in Residential Plumbing. In Scaling and Corrosion In Water and Wastewater Systems. Edited by Simon Parsons, Richard Stuetz, Bruce Jefferson and Marc Edwards. Water Sci. Technol. 49 (2), 83. Edwards, M.; Ferguson, J.F.; Reiber, S. (1994) The Pitting Corrosion of Copper. Jour. AWWA.. 86 (7), 74. Fairey, J.L.; Katz, L.E.; Speitel, G.E. (2004) Monochloramine Destruction in GAC Beds. Proceedings of the 2004 AWWA WQTC. 13 pages. Federal Register: August 18, 2003 (V.68, N.159)]. National Primary Drinking Water Regulations: Stage 2 Disinfectants and Contaminants; Disinfection Byproducts Rule; National Primary and Secondary Drinking Water Regulations: Approval of Analytical Methods for Chemical Contaminants; Proposed Rule. Garret, J.H. (1891). The Action of Water on Lead. London, England. H.K. Lewis.

17 Gibbs, R.A.; Scutt, J.E.; Croll, B.T. (1993) Assimilable Organic Carbon Concentrations and Bacterial Numbers in a Water Distribution System. Water. Sci. Technol Gorman, C. Has your shower become a water hazard? Time Magazine, July 1, Ingleson, H.; Sage, A.M.; Wilkinson, R. (1949) The Effect of the Chlorination of Drinking Water on Brass Fittings. Jour. Inst. Water Engineers (Br.). 3:81. Jackson, P.J.; Williams, N.M.; Rule, K.L.; Davis, L.J.; Blake, S.; Warburton, S.G.; Ellis, J.C. (2004) Quality of Drinking Water in Public Buildings. Final Report to the Drinking Water Inspectorate No: DWI Kindaichi, T.; Ito, T.; Okabe, S. (2004) Ecophysiological Interaction between Nitrifying Bacteria and Heterotrophic Bacteria in Autotrophic Nitrifying Biofilms as Detemined by Microautoradiography-Fluorescence in-situ Hybidization. Appl. Environ. Microbiol., 70 (3), Larson, T.E.; King, R.M.; & Henley, L. (1956) Corrosion of Brass by Chloramine. Jour. AWWA, (January 1956). LeChevalier, M.W.; Schulz, W.; Lee, R.G. (1991) Bacterial nutrients in drinking water. Appl. Environ. Microbiol., LeChevallier, M.W; Lowry,C.D; Lee, R.G; Gibbon,L.D.(1993) Examing the relationship between iron corrosion and the disinfection of biofilm bacteria. Jour. AWWA, 85(7), 111. LeChevalier, M.W.; Wade, T.J.; Shaw, S.; Very, D.A.; Calderon, R.L.; Colford, J.M. (2004) Results of the Big Wet: an Epidemiology Study of the Microbiological Quality of Drinking Water in Davenport, Iowa. Proceedings of the 2004 AWWA WQTC. 5 pages. Lesilie, C.P.; Grady, Daigger, G.T.; Lim, H.C. (1999) Biological Wastewater Treatment (2 nd ). Marcel Dekker, Inc. New York Lin, N-H.; Torrents, A.; Davis, A.P.; Zeinali, M. (1997) Lead Corrosion Control from Lead, Copper- Lead Solder, and Brass Coupons in Drinking Water Employing Free and Combined Chlorine. Jour. Environ. Sci. Health, 32(4), 865. Mayo Clinic, Hot Tub Lung. Accessed August 22, 2004: Murphy, B; O Connor, J.T.; O Connor, T.L. (1997a) Willmar, Minnesota Battles Copper Corrosion. Part 3. Public Works, 12, 37. Murphy, B; O Connor, J.T.; O Connor, T.L. (1997a) Willmar, Minnesota Battles Copper Corrosion. Part 2. Public Works, 11, 44. Okabe, S.; Naitoh, N.; Satoh, H.; Watanabe, Y. (2002) Structure and Function of nitrifying biofilms as determined by molecular techniques and the use of microelectrodes. Water Sci. Technol., 46 (1-2), 233. Powell, R.M. (2004) Implementation of Chloramination by a Florida Utility: The Good, the Bad and The Ugly. Proceedings of the 2004 AWWA WQTC. 16 pages. Rice, E.W.; Scarpino, P.V.; Reasoner, D.J.; Logsdon, G.S.; Wild, D.K. (1991) Correlation of Coliform Growth Response with Other Water Quality Parameters. Jour. AWWA., 83 (7), 98. Rittman, B.E.; Snoeyink, V.L. (1984) Achieving Microbially Stable Drinking Water. J. AWWA 76 (10), 106. Rittmann, B.E.; Regan, J.M.; Stahl, D.A. (1994) Nitrification as a Source of Soluble Organic Substrate in Biological Treatment. Water Sci. Tech., 30 (6), 1.

18 Rohner, A.L.; Pang, L.W.; Linuma, G.; Tavares, D.K.; Jenkins, K.A.; Geesey, Y.L. (2004) Effects of Upcountry Maui Water Additives on Health. Hawaii Medical Journal. Sept 2004, V.63, pp Rollinger, Y.; Dott, W. (1981) Survival of Selected Bacterial Species in Sterilized Activated Carbon Filters and Biological Activated Carbon Filters. Appl. Environ. Microbiol., 53 (4), 777. Rushing, J.C.; Edwards, M. (2004) Effect of Aluminum Solids and Free Cl 2 on Copper Pitting. Corr. Sci., 46 (12), San Francisco (2004). In Depth Chloramination Q&A: Statistics and General Information Accessed 1/1/2004 at Sathasivan, A.; Ohgaki, S. (1999) Application of New Bacterial Regrowth Potential Method for Water Distribution System a clear evidence of phosphorus limitation. Wat. Res., 33(1), 137. Spinks, A.T.; Dunstan, R.H.; Coombes, P.; Kuczera, G. (Nov, 2003) Thermal Destruction Analyses of Water Related Pathogens at Domestic Hot Water System Temperatures. The Institution of Engineers. 28th International Hydrology and Water Resources Symposium. Stine, S.W.; Pepper, I.L.; Gerba, C.P.(2005) Contribution of drinking water to the weekly intake of heterotrophic bacteria from diet in the United States. Water Res., 39(1), 257. Tarre, S.; Green, M. (2004) High-rate Nitrification at Low ph in Suspended- and Attached-Biomass Reactors Appl. Environ. Microbiol., 70 (11), Tobin, T.C. Shower Is Top Suspect in Illness. St. Petersburg Times (Florida, USA), August 18, Accessed 8/22/2004 at 18/State/Shower_is_top_suspect.shtml U.S. Environmental Protection Agency (1999) Stage 2 Microbial/Disinfection Byproducts Stakeholder Orientation Meeting. Final Meeting Summary. U.S. Environmental Protection Agency (2002) Health Risks from Microbial Growth and Biofilms in Drinking Water Distribution Systems. U.S. Environmental Protection Agency (2004) Energy Efficiency. Accessed at Van der Kooij. (1992) Assimilable Organic Carbon as an Indicator of Bacterial Regrowth. Jour. AWWA. 84 (2), 57. Westerhoff, P.; James, J. (2003) Nitrate Removal in Zero-Valent Iron Columns. Water Res., 37 (8), WHO. Heterotrophic Plate Count Measurement in drinking water safety management. WHO/SDE/WSH/ Report of an expert meeting, Geneva, April Accessed 8/22/04 at: 2.htm WHO. Heterotrophic Plate Count Measurement in drinking water safety management. WHO/SDE/WSH/ Report of an expert meeting, Geneva, April Accessed 8/22/04 at: 2.htm Yang., J.; Harrington, G.W.; Noguera, D.R. (2004) Nitrification Modeling in Pilot-Scale Chloraminated Drinking Water Distribution Systems. Proceedings of the 2004 AWWA WQTC. 5 pages.