Science of the Total Environment 520 (2015) 120–126

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Seasonal and spatial variability of nitrosamines and their precursor sources at a large-scale urban drinking water system Gwen C. Woods a, Rebecca A. Trenholm a, Bruce Hale b, Zeke Campbell b, Eric R.V. Dickenson a,⁎ a b

Southern Nevada Water Authority, P.O. Box 99954, Las Vegas, NV 89193-9954, USA Denver Water, 1600 West 12th Avenue, Denver, CO 80204-3412, USA

H I G H L I G H T S • • • • •

Nitrosamines were examined in a drinking water system utilizing chloramines. Only N-nitrosodimethylamine (NDMA) was detected throughout the system. Seasonal variations indicate that greater NDMA forms in cooler temperatures. Evidence is provided that NDMA sources may originate from the distribution system. Results provide insight into NDMA linked to changes in water quality.

a r t i c l e

i n f o

Article history: Received 9 January 2015 Received in revised form 2 March 2015 Accepted 3 March 2015 Available online 22 March 2015 Editor: D. Barcelo Keywords: N-nitrosodimethylamine Full-scale Distribution system Seasonality Disinfection by-products Occurrence

a b s t r a c t Nitrosamines are considered to pose greater health risks than currently regulated DBPs and are subsequently listed as a priority pollutant by the EPA, with potential for future regulation. Denver Water, as part of the EPA's Unregulated Contaminant Monitoring Rule 2 (UCMR2) monitoring campaign, found detectable levels of Nnitrosodimethylamine (NDMA) at all sites of maximum residency within the distribution system. To better understand the occurrence of nitrosamines and nitrosamine precursors, Denver Water undertook a comprehensive year-long monitoring campaign. Samples were taken every two weeks to monitor for NDMA in the distribution system, and quarterly sampling events further examined 9 nitrosamines and nitrosamine precursors throughout the treatment and distribution systems. NDMA levels within the distribution system were typically low (N 1.3 to 7.2 ng/L) with a remote distribution site (frequently N200 h of residency) experiencing the highest concentrations found. Eight other nitrosamines (N-nitrosomethylethylamine, N-nitrosodiethylamine, N-nitroso-di-npropylamine, N-nitroso-di-n-butylamine, N-nitroso-di-phenylamine, N-nitrosopyrrolidine, N-nitrosopiperidine, N-nitrosomorpholine) were also monitored but none of these 8, or precursors of these 8 [as estimated with formation potential (FP) tests], were detected anywhere in raw, partially-treated or distribution samples. Throughout the year, there was evidence that seasonality may impact NDMA formation, such that lower temperatures (~5–10 °C) produced greater NDMA than during warmer months. The year of sampling further provided evidence that water quality and weather events may impact NDMA precursor loads. Precursor loading estimates demonstrated that NDMA precursors increased during treatment (potentially from cationic polymer coagulant aids). The precursor analysis also provided evidence that precursors may have increased further within the distribution system itself. This comprehensive study of a large-scale drinking water system provides insight into the variability of NDMA occurrence in a chloraminated system, which may be impacted by seasonality, water quality changes and/or the varied origins of NDMA precursors within a given system. © 2015 Elsevier B.V. All rights reserved.

1. Introduction With the advent of the EPA's Stage 2 DBP rule, treatment facilities have increasingly converted from chlorine to chloramine disinfection to reduce regulated disinfection by-products (DBPs), namely ⁎ Corresponding author. E-mail addresses: [email protected] (G.C. Woods), [email protected] (E.R.V. Dickenson).

http://dx.doi.org/10.1016/j.scitotenv.2015.03.012 0048-9697/© 2015 Elsevier B.V. All rights reserved.

trihalomethanes (THMs) and haloacetic acids (HAAs). Nitrosamines are a class of unregulated DBPs that are prevalent with the use of chloramines (Krasner et al., 2013), are probable human carcinogens (USEPA, 1991), and are perceived to pose a greater health risk than currently regulated DBPs (Shah and Mitch, 2012). Five nitrosamine species are acknowledged by the EPA's third Contaminant Candidate List (CCL3) (USEPA, 2009), six are listed in the second Unregulated Contaminant Monitoring Rule 2 (UCMR2) (USEPA, 2012), and future EPA regulation of nitrosamines is a possibility. Among the most widely identified and

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

monitored nitrosamines is N-nitrosodimethylamine (NDMA), a nitrosamine species found above the minimum reporting level in ~50% of samples in treatment systems utilizing chloramination as a primary disinfectant (UCMR2 database) (Woods and Dickenson, 2015; USEPA, 2012). The USEPA estimates that NDMA has a 10−6 cancer risk level of 0.7 ng/L in drinking water (USEPA, 1993). The US Office of Environmental Health Hazard Assessment has issued a health goal of 3 ng/L (CDPH, 2013), while Ontario (Canada) has a standard of 9 ng/L. Australia regulates NDMA at 10 ng/L (QPC, 2005), and the World Health Organization has issued a 100 ng/L guideline (WHO, 2008). It is becoming increasingly apparent in the water industry that no single disinfection technique is free from DBP formation, and that further research is necessary to better understand both the reduction of DBP formation and subsequent trade-offs necessary to keep all species minimized. Widespread occurrence data of nitrosamines, however, are still somewhat lacking, particularly studies that have examined temporal variability. Of studies that have been conducted, a survey of 36 drinking water systems in California found NDMA associated with ion exchange treatment as well as recycled water (CDPH, 2002). Quarterly data from 21 drinking water systems in the USA and Canada, likewise found that chloraminated systems produced more NDMA, and that levels were generally higher in distribution systems. This study did not discuss seasonal variability (Barrett et al., 2003). Other surveys include the examination of 38 drinking water systems in North America, wherein distribution samples were also found to contain elevated NDMA concentrations (Boyd et al., 2011), and a comprehensive study in Ontario, Canada from 1998 to 2009 (MOE, 2013). Data from 1998 to 2007 from the Ontario survey was examined previously (Russell et al., 2012), but information such as disinfectant use was not reported, and seasonality was not discussed. Finally, the USEPA's UCMR2 survey is spatially the most comprehensive survey to-date, with some 1199 utilities required to monitor for nitrosamines on a bi-yearly (groundwater) or quarterly (surface water) basis. Variability across regions is apparent from this dataset, but may, to some extent, be a product of regional differences in disinfection practices; seasonality was overall found to have no clear trend with NDMA occurrence (Woods and Dickenson, 2015). While important information is gleaned from the above-mentioned occurrence studies, more comprehensive analyses of individual water systems may render useful information such as regional variability with climate, detailed information about plant practices, and testing for nitrosamine precursors so as to better define the origins of precursors. The research presented here sought to capture a thorough investigation of nitrosamine occurrence at a full-scale urban drinking water treatment facility, and examine detailed occurrence both spatially and temporally. Characterization of nitrosamine precursors may enable utilities to manage source water supplies and treatment practices to minimize nitrosamine formation. Heavy loadings of NDMA precursors in utility source waters have been linked to wastewater-impact (Krasner et al., 2008; Schmidt and Brauch, 2008; Nawrocki and Andrzejewski, 2011; Shen and Andrews, 2011; Russell et al., 2012). Earlier research proposed that secondary amines present in wastewater or other natural waters may form a hydrazine intermediate which further reacts with monochloramines to form NDMA (Choi and Valentine, 2002; Mitch

121

and Sedlak, 2002). Other research has sought to address inputs of natural sources of NDMA precursors into raw water, but naturally-derived precursors appear to have lower NDMA yields than many of the anthropogenically-derived compounds tested (Dotson et al., 2009; Gerecke and Sedlak, 2003; Krasner et al., 2013). Materials from within the water treatment process may contribute significant NDMA precursors. In plants utilizing anion exchange treatment, resins with tertiary and quaternary amine functional groups contain NDMA precursors that may be released during treatment (Flowers and Singer, 2013; Kemper et al., 2009; Nawrocki and Andrzejewski, 2011). Recent research further highlights nitrosamine formation from organic polymer aids following oxidation with chlorination or chloramination (Kemper et al., 2010; Mitch and Sedlak, 2004; Park et al., 2009a,b; Wilczak et al., 2003). Coagulant polymer aids that contain quaternary ammonium cations, such as epichlorohydrin (Park et al., 2009a,b) and the widely-used polyDADMAC (Najm and Trussell, 2001; Park et al., 2009a,b; Wilczak et al., 2003), have been identified as high-yielding NDMA precursors. Finally, origins from distribution system materials, such as rubber gaskets and seals have been identified (Morran et al., 2011; Teefy et al., 2011), although relatively little research has been designated to distribution precursors. As part of the USEPA's second Unregulated Contaminant Monitoring Rule (UCMR2), all public water systems serving N100,000 people were required to monitor for six nitrosamines, including NDMA (Table 1). The urban public drinking water system, Denver Water, serves some 1.2 million people (Denver Water, 2014) and was included in this survey. Denver Water's water supply originates primarily from mountain snowmelt. Anthropogenic impacts such as wastewater influence are considered low, but the treatment plants do utilize the coagulant polymer aid polyDADMAC as well as chloramines for final disinfection. The quarterly UCMR2 results from three sites of maximum residency in the distribution system resulted in detectable levels of NDMA (≥2 ng/L) in all but 2 of the 12 samples, with values ranging from 2.0 to 4.7 ng/L (median value of 2.8 ng/L); 3 entry point sampling sites resulted in no detection of NDMA on any of the sampling dates. The other five nitrosamines included in the UCMR2 survey were not found at detectable levels. Following discovery of NDMA in distribution samples, Denver Water in collaboration with the Southern Nevada Water Authority initiated a more intensive monitoring survey to better understand the temporal and spatial variability of nitrosamines, as well as nitrosamine precursors. Biweekly sampling was conducted throughout 2013 at points of maximum residency in the distribution system (UCMR3 sites) as well as more extensive sampling campaigns conducted on a quarterly basis with samples from source waters and partially treated waters. During the quarterly sampling, 8 other nitrosamines of concern were measured (Table 1) and formation potential (FP) tests were performed to estimate nitrosamine precursor loadings throughout the treatment and distribution systems. To the best of our knowledge, this is the most temporally-intensive sampling campaign ever conducted for nitrosamines in a full-scale system. The results lend insight into seasonal effects on NDMA formation as well as evidence of NDMA precursors originating from within the distribution system, a category of precursors much less studied than source water precursors.

Table 1 Suite of 9 nitrosamines analyzed quarterly at Denver Water (Dec. 2012–Dec. 2013). Nitrosamine species

Cancer risk level (ng/L)

Included in CCL3

Included in UCMR2

MRL from UCMR2 (ng/L)

MRL from this study (ng/L)

N-nitrosodimethylamine (NDMA) N-nitrosomethylethylamine (NMEA) N-nitrosodiethylamine (NDEA) N-nitroso-di-n-propylamine (NDPA) N-nitroso-di-n-butylamine (NDBA) N-nitroso-di-phenylamine (NDPhA) N-nitrosopyrrolidine (NPYR) N-nitrosopiperidine (NPIP) N-nitrosomorpholine (NMOR)

0.7 2 0.2 5 6 7000 16 0.8 0.8

x

x x x x x

2 3 5 7 4

x

2

1.3 2.5 5.0 10 10 10 10 5.0 5.0

x x x x

122

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

2. Materials and methods 2.1. Site description Denver Water's water supply originates primarily from mountain snowmelt, is stored in mountain reservoirs on both sides of the Continental Divide and is carried via large tunnels and natural waterways to raw water reservoirs on the eastern side of the Divide. Three of these reservoirs are connected directly to 3 distinct metroarea water treatment plants with plant capacities of 185, 250 and 280 million gallons daily (Denver Water, 2014). Water from the plants is sent to some 30 treated water storage reservoirs with total capacity of more than 350 million gallons and through more than 3000 miles of pipeline for distribution to homes and businesses throughout the Denver area (Denver Water, 2014). The average daily consumption for 2013 was 165 MGD (personal communication with utility personnel). The 3 facilities employ conventional treatment of flocculation, sedimentation, filtration, fluoride addition, chlorine contact (~ 2 mg/L as Cl2) followed by ammonia addition (~ 0.5 mg/L as N) and pH adjustment (typical pH ~ 8). DOC concentrations for all 3 source waters are normally 2–3 mg/L. All 3 facilities use polyDADMAC as a coagulant aid (~ 1–3 mg/L at rapid mix) as well as a non-ionic polymer, polyacrylamide (as a filter aid) at points post coagulation. All 3 plants practice chlorine addition prior to coagulation, to the degree sufficient to detect a trace residual leaving the filters, translating into a period of chlorine pre-oxidation of roughly 2–8 h. Chloramines are used for final disinfection with a typical target residual of 1.5–1.8 ppm entering the distribution system. Two of the 3 plants receive their source waters from the South Platte River, with minor municipal, ranching, mining and recreational impacts. The most significant impact on this watershed is from the 2002 Hayman Fire of over 130,000 acres. The Foothills plant draws water directly from the river, at a location known as Strontia Springs Reservoir. The diversion point for the Marston plant is just downstream of the Foothills diversion. Under normal operations, this water flows into the 20,000 acre-ft impoundment known as the Marston Forebay, with up to a 6 month residence time before use in the plant. During periods where the Marston Forebay contains nonoptimal raw water (e.g., following algal blooms), water from the South Platte River flows directly to the Marston treatment plant, bypassing the Forebay. This alternative operational approach was used during 2 periods of this study (mid-Oct. 2012–late Jan. 2013; late Jul. 2013–early Aug. 2013). In periods of drought, waters entering the Forebay can be drawn from other sources that are somewhat influenced by recreational activities and wastewater. Waters from these other sources were used to supplement the Marston Forebay in 2012 and again from Feb–May, 2013. Denver's third water treatment plant, Moffat, receives its water from Ralston Reservoir with two sources: locally from Ralston Creek (influenced somewhat by mining) and most significantly, from a diversion on South Boulder Creek. This watershed is known to have minor impact from recreational activity, mining and pine bark beetle infestation. Moffat water is readily distinguished from water originating from the South Platte River using specific conductance; the former has a specific conductance typically b150 μS, whereas the latter is typically N250 μS. Finished water from all 3 plants may be blended at any point in the distribution system, though under normal operations, there is a high probability that water at any given point in the system is predominantly from a single plant. Water blended between South Platte River and Moffat sources has specific conductance typically between 150 and 250 μS. At any given time, one of the plants is offline, except when all 3 operate during maximum consumption in the summer or other periods of need. The 3 distribution sites chosen for UCMR2 monitoring have a high probability of being served predominantly by a given plant. These points are estimated to be the furthest points that satisfy

this condition within the distribution system. The UCMR3 sites, sampled within this study, are different from the UCMR2 sites, but were selected with the same intent. The new sites were chosen from Denver Water's 16 required Stage 2 DBP sites, which in turn were chosen from 44 Initial Distribution System Evaluation (IDSE) study sites monitored in 2008. The new sites were used in preference to the UCMR2 sampling locations based upon knowledge gained through the IDSE, indicating DBP monitoring objectives are best met at sites representing a compromise between long residence time and greater assurance that the majority of water would be from a given plant. 2.2. Sampling Monitoring was conducted from Dec. 2012 to Dec. 2013, with NDMA measurements performed at the 3 distribution sites every 2 weeks. On a quarterly basis, 6 other sites were monitored: one plant influent from each of the plants as well as one post-filtration sampling point (post-coagulation and polyDADMAC addition plus at a point of little-to-no chlorine residual). During the quarterly campaigns, 8 other nitrosamines were monitored at all sites and FP tests were run with chloramines to elucidate nitrosamine precursor loadings at each of the sites. Samples were collected in 1 L amber glass bottles along with field blanks, matrix spikes and experimental duplicates (for each batch collected). Sample bottles for background nitrosamine measurements contained 1 g sodium azide to inhibit biological growth and 80 mg sodium thiosulfate to quench residual disinfectant; these preservatives have previously undergone holding studies in Holady et al. (2012). Bottles for FP tests contained no preservatives. All bottles were kept cold, shipped overnight for laboratory analyses and analyzed within a few days (maximum holding time of 14 days). For quality assurance/quality control purposes, a DI blank, field blank, sample duplicate, and matrix spike were collected with each sample set. Previous research has shown that sample duplicates ranged from 6 to 12% relative standard deviation for NDMA and less than 20% for the other nitrosamines measured (Holady et al., 2012). 2.3. Nitrosamine measurements All nitrosamine extractions and analyses were conducted using a method described previously (Holady et al., 2012), which is essentially a modified version of USEPA Method 521 (USEPA, 2004). Briefly, samples were extracted with automated solid phase extraction (SPE) using activated charcoal SPE cartridges (Restek 521), and extraction procedures are detailed in Holady et al. (2012). Typically, 1 L of sample is extracted but to obtain a lower minimum reporting level (MRL), samples that were directly analyzed for nitrosamines (i.e., not undergone FP) were extracted in 2 L volumes (MRL: 1.3 ng/L). All quantitation was performed using isotope dilution with isotopically labeled nitrosamines for each analyte (Holady et al., 2012). Nitrosamines were measured via gas chromatography tandem mass spectrometry (GC– MS/MS) in positive chemical ionization (PCI) mode with liquid methanol and a Varian 4000 Ion Trap GC–MS/MS system on a DB-624 column (30 m × 0.25 mm × 1.4 μm). Whereas NDPhA breaks down in the injector to diphenylamine, diphenylamine measurements were used to analyze NDPhA. 2.4. Nitrosamine formation potential (FP) testing Nitrosamine FP tests were used to assess nitrosamine precursor content in samples. FP samples were collected at all 9 sample locations (raw, post-filtration and distribution sites for all 3 plants) on a quarterly basis. Using methods presented previously (Mitch and Sedlak, 2002, 2004; Mitch et al., 2003), 1 L of sample was dosed with phosphate buffer to pH 6.9 prior to addition of preformed monochloramines at a concentration of 140 mg/L as Cl2. Glass bottles were held at room temperature in the dark for 10 days and then quenched with sodium thiosulfate.

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

123

b100 h, respectively. The DS-FH site is both physically far away from the Foothills plant and also experiences low local water use in relation to the scale of infrastructure, resulting in slow flow. Previous research has illustrated the slow kinetics of NDMA formation from NDMA precursors and chloramines, such that long residency times result in enhanced NDMA formation (Krasner et al., 2013). This scenario would appear to be the case at DS-FH, but examination of NDMA-FP reveals yet further details about precursors present (see Section 3.2). To decipher any seasonal trends or temperature influence, DS-FH, the site with consistently detectable NDMA data, was examined more closely. At first glance, there appears to be no clear trend (Fig. 1). If NDMA occurrence is further plotted against the water temperature data in the distribution system (Fig. 2), however, a relationship becomes apparent; at temperatures lower than ~10 °C, NDMA concentrations are predominantly at their highest, and as temperatures increase above 10 °C, NDMA concentrations decline. The statistical examination of a relationship between temperature and NDMA concentrations results in a negative correlation at a statistically significant level (p b 0.05), following removal of 2 outlier values from January (a pronounced dip in NDMA levels in Fig. 2). Previous research has demonstrated a similar effect on source waters with pH ~ 8 to 9 (Krasner et al., 2012). Though at pH ~ 7, temperature was found to be positively correlated with NDMA formation, at higher pH's, increasing temperatures were shown to decrease NDMA formation. This phenomenon was attributed to prechlorine oxidation being more effective at destroying NDMA precursors in warmer waters at pH 8 or 9. Alternatively, at pH 7, fewer deprotonated amines were presumably available for effective destruction/deactivation (Krasner et al., 2012). Denver Water source waters are typically greater than 7 and pH values at the DS-FH site were generally greater than 8 throughout the duration of this study (data provided by utility personnel). Denver Water also practices pre-chlorination during which contact with chlorine prior to ammonia addition is typically anywhere from 2–8 h. During warmer temperatures, the Foothills plant is typically treating at higher treatment rates than in winter, effectively reducing the pre-chlorine contact time, and presumably would have even lower NDMA formation at DS-FH if production rates were similar to winter months. A seasonal trend is therefore apparent, with lower NDMA concentrations during summer months and higher NDMA concentrations in the winter, when pre-chlorination is less effective at destroying NDMA precursors.

3. Results and discussion 3.1. Nitrosamine occurrence The year-long sampling campaign was concluded in Dec. of 2013. All 9 nitrosamines (Table 1) were monitored on a quarterly basis at points in the source waters, at points in treatment just post-filtration, and at the distribution sites for all 3 treatment plants. From the quarterly monitoring, only NDMA was found at detectable levels throughout the entire system. Previous analysis, provided by nationwide nitrosamine sampling for the EPA's UCMR2 monitoring, illustrates that NDMA is typically much more prevalent than other nitrosamine species: less than 0.5% of samples collected nationwide were found to contain the other nitrosamine species tested (NDBA, NDEA, NDPA, NMEA, NPYR; Woods and Dickenson, 2015). In examining the occurrence of the remaining nitrosamine species analyzed in the current study, an intensive monitoring campaign in Ontario, Canada found that NPIP and NMOR occurred in 0.9% and 8.5% of samples, respectively (at 109 and 160 systems, respectively), while NDMA occurred 37.1% of the time (179 systems) (Russell et al., 2012). A further study at 38 systems in the U.S. and Canada had occurrence levels of 15.8% for both NMOR and NDPhA, while NDMA occurred 73.7% of the time (Boyd et al., 2011). It is therefore perhaps not surprising that NDMA is the prevalent species within the Denver Water system. For biweekly monitoring, NDMA was the only nitrosamine measured and the results are illustrated in Fig. 1. It should be noted that for any given monitoring event, distribution system sites are expected to have a high probability of water predominantly originating from the plant “associated” with that site. Further, there are periods when a given plant is offline, meaning that the data represent some unknown contributions from the other 2 plants (Fig. 1). Overall, the Foothills-associated site (DS-FH) had consistently higher NDMA concentrations than the other two sites, with 100% of samples demonstrating detectable levels of NDMA with a median value of 5.4 ng/L. The Moffat-associated site (DS-MF) had the second greatest occurrence with 80% detectable and a median value of 1.7 ng/L; the Marston-associated site (DS-MR) had only 44% detects with a median value of 1.6 ng/L. In examining distances between DS sites and associated treatment plants, the Foothills site is significantly further from its associated treatment plant than the other 2 sites. Communications with utility personnel confirm that the residency time for DS-FH, DS-MF, and DS-MR are on the order of N200, ~ 100, and

8 7

NDMA (ng/L)

6 5 DS-MF DS-MR DS-FH

4 3 2 1

12/31

12/3

12/17

11/19

11/5

10/22

10/8

9/24

9/10

8/27

8/13

7/30

7/16

7/2

6/18

6/4

5/7

5/21

4/23

4/9

3/26

3/12

2/26

2/12

1/29

1/1

1/15

12/18

12/4

0

Dates (2012-2013) Fig. 1. NDMA occurrence within the distribution system on a biweekly basis at 3 maximum residency sites. Bars at the top indicate when associated water treatment plants were offline. MF: Moffat, MR: Marston, FH: Foothills (detection limit ≥ 1.25 ng/L).

124

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

8

25

7 6 5

15

4

10

3 Temp, FH finished

5

NDMA (ng/L)

Temperature (Celsius)

20

2

Temp, DS-FH

1

NDMA, DS-FH

0

12/31

12/17

12/3

11/19

11/5

10/22

10/8

9/24

9/10

8/27

8/13

7/30

7/2

7/16

6/18

6/4

5/7

5/21

4/23

4/9

3/26

3/12

2/26

2/12

1/15

1/29

1/1

12/18

12/4

0

Dates (2012-2013) Fig. 2. NDMA occurrence at DS-FH plotted with temperature data from water leaving the Foothills plant and water at the DS-FH site.

3.2. Nitrosamine-FP Nine nitrosamines (Table 1) were tested for nitrosamine formation potential (FP). Of the 9 nitrosamines, only NDMA was found at detectable levels before and after chloramine applications. Background NDMA measurements were compared to NDMA-FP results to verify if a correlation existed throughout this study. Fig. 3 displays the NDMA and NDMA-FP data from the distribution sites (note: detectable levels of NDMA were not found in raw and post-filtration samples and subsequently could not be compared to FP data). The data demonstrate a strong positive correlation (R2 = 0.81; p b 0.01), validating that the NDMA precursor (FP) test, is in this case a good proxy method for the potential of water to form NDMA. A similar trend was noted in previous research (Mitch et al., 2003). Fig. 4 illustrates NDMA precursors measured throughout the entire treatment and distribution systems from Dec. 2012 to Dec. 2013. The Marston plant, Foothills plant and distribution system were monitored for NDMA-FP in Dec. (18th, 2012), Mar. (26th), Jun. (18th) and Sep. (24th). The Moffat plant was only online for the Jun. sampling event but was also sampled (along with its associated distribution site) at other times of the year in Feb. (12th) and Dec. (3rd, 2013) so as to capture more temporal data. Overall, the data illustrate an increasing trend over the course of the year, with raw waters demonstrating a particularly clear increase. The Sep. and Dec. (2013) sampling events, however, follow catastrophic flooding and represent unusual and problematic

Background NDMA (ng/L)

In addition to illustrating a potential trend with temperature, the DSFH data elucidate a handful of other variabilities that coincide with significant water quality events. For example, the Moffat-associated site (DS-MF) had 2 elevated NDMA levels in May, 2013 (7th and 10th). Fig. 1 illustrates that May 7th coincides with the Moffat plant coming back online (May 6th) and the Marston plant going offline (May 7th). Changes in plant operations and subsequent distribution flow and source plant contributions are likely therefore linked to elevated NDMA levels. A further elevated data point occurred in Aug. (27th) at the DS-MR site. This date coincides with a change in Marston source waters by using water directly from the South Platte River following a Marston Forebay taste and odor concern in July. The plant intake was switched back to the Forebay on Aug. 14th, and was followed by the greatest NDMA concentration detected at DS-MR (Aug. 27th) over the duration of the study. During the taste and odor incident, copper sulfate was applied to reduce algal blooms (Jul. 31st), and further sampling by utility personnel revealed elevated levels of ammonia for at least 2 weeks following copper sulfate applications (personal communication with utility personnel). There is potential therefore for the elevated DS-MR NDMA concentration to be attributed to the algal bloom where elevated ammonia and algae cell lysing occurred in subsequent weeks. A final noteworthy event began on Sept. 9th, 2013 when persistent heavy rains led to catastrophic flooding along the Colorado Front Range. The Moffat plant experienced poor water quality and subsequently went offline on the 12th. The Foothills plant was likewise impacted by related water quality issues and deviated from typical plant practices. Coagulant doses, including polyDADMAC applications, increased by up to ~ 50%, and rather than free chlorine application early in treatment, potassium permanganate was instead used to satisfy demand. During this period, chlorine was not used in the process until after filtration, significantly reducing the opportunity for pre-chlorine oxidation to destroy NDMA precursors. Fig. 1 demonstrates that the DS-FH and DS-MF experienced elevated NDMA concentrations (compared to previous sampling date) in the weeks following the flood (Sept. 24th and Oct. 7th). When the Moffat plant returned online (Nov. 16th), Ralston Reservoir (its water source) was still experiencing unusually high TOC values (~7 ppm) and coagulant and polymer doses were ~2× higher than usual. The Nov. 19th and Dec. 3rd NDMA concentrations at the DS-MF site may have been impacted by both higher TOC and polyDADMAC inputs.

8 7 6 5 4 3 2 1 0

y = 0.65x - 8.2 R² = 0.81

12

14

16

18

20

22

NDMA-FP (ng/L) Fig. 3. NDMA measurements vs. NDMA-FP measurements from the DS-FH site.

24

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

25

NDMA-FP (ng/L)

20 15

10 Raw water

5

Post-filtration

12/7

11/12

9/23

10/18

8/4

8/29

7/10

5/21

6/15

4/1

4/26

3/7

2/10

1/16

12/22

11/27

Distribution

0

Dates (2012-2013) Fig. 4. NDMA precursor estimates throughout the entire treatment and distribution systems.

water quality conditions (see previous section for more discussion). In looking at the sampling events from Dec. (2012) thru Jun., it would appear that increasingly warmer months may be contributing more NDMA precursors in raw, partially treated, and finished drinking water samples. This increase in NDMA-FP values interestingly coincides with a general decreasing trend in actual NDMA formation at the remote distribution site (DS-FH; Fig. 1). More sampling would need to be conducted to verify an increasing trend in NDMA precursors with temperature. The NDMA-FP data were further distinguished by plant and associated distribution sites (Fig. 5). These data demonstrate an overall increasing trend in NDMA precursors from raw to post-filtration to distribution sites. The Foothills plant and associated distribution site illustrate a particularly dramatic increase in precursors from raw to finished water. Although there are not enough data points from the individual plants (3–4 points at each site) to determine if the increasing trends are statistically significant, the overall system was found to have increasing NDMA precursors from raw to post-filtration to distribution sites throughout the duration of the study at a statistically significant level (2-tailed, paired t-test, p b 0.05). What is particularly interesting is the average NDMAFP at the 3 distribution sites, such that greater residency time in the distribution system appears to coincide with increasing NDMA precursors (i.e., Marston b Moffat b Foothills). Thus the higher NDMA concentrations at DS-FH in Fig. 1 are likely from longer chloramine contact time than the other distribution sites, but may also be a product of increased precursor contributions within the distribution system.

Raw water

25

Distribution

20 15 10

n=4 n=4 n=4

n=11 n=11 n=11

0

n=4 n=4 n=4

5

n=3 n=3 n=3

NDMA-FP (ng/L)

Post-filtration

Moffat

Marston

Foothills

Overall

Fig. 5. Average NDMA precursor estimates (FP tests) by individual plants and the overall system.

125

In examining the NDMA precursor test results for the distribution system, an important consideration has to be kept in mind. Distribution system water had been in the presence of chloramines for ~100 h (DSMF and DS-MR) or ~ 200 h (DS-FH; entering the distribution at a targeted 1.5 ppm) prior to additional chloramination from FP tests (140 ppm initial dose for 10 days). The initial research that generated this FP test (Mitch et al., 2003) had previously examined how varying concentrations of chloramines and duration of contact time affected FP results. Ideally a FP test provides conditions that convert all precursors in a given sample to NDMA and thereby indicates the “maximum potential” to form NDMA. Mitch et al. (2003) demonstrated on several wastewaters that with a 140 ppm dosage, all NDMA precursors had essentially reacted within 5 days. With natural watershed samples, however, they discovered that precursor reactions exhibited slower kinetics and would continue to form NDMA (albeit by continuing to maintain a daily residual of 70 ppm). Where source waters in this study are primarily concerned with watershed precursors, further analyses with simulated distribution system (SDS) tests were performed to elucidate whether the elevated NDMA-FP values from the distribution system were a product of additional precursors or simply a product of increased chloramine reaction time. FP tests were performed on Foothills finished waters with SDS tests performed with reaction times of 0, 100, 200 h. The results from this supplementary analysis provided insight into NDMA-FP results from the distribution system. When finished water samples underwent 100 h of SDS conditions (1.5 ppm initial chloramine dose, room temperature) followed by 10 days of FP conditions, there was 0.0 ± 8.8% difference in NDMA concentrations compared to the control (0 h SDS). However, when samples underwent 200 h of SDS conditions, there was a 15.7% ± 3.9% increase in NDMA formed. In comparing this data with the NDMA-FP results generated from the actual distribution system, DS-FH had an average of 45.9 ± 33.2% increase in NDMA after FP conditions over the course of the year. Removing abnormal data following the September flooding further raises this estimate to 59.7 ± 22.6%. From a qualitative standpoint, however, these data tentatively suggest that the extended contact time with chloramines prior to FP tests may have increased the NDMA-FP values, but that such conditions are not likely to account for all of the NDMA formed. Some NDMA precursors may have been added within the distribution system. More data points would be necessary to verify that the distribution system is contributing to higher FP values. The apparent increase in NDMA precursors from raw to partiallytreated waters is perhaps not surprising owing to extensive research that has previously shown that cationic polymer coagulant aids (such as polyDADMAC) form considerable NDMA following disinfection with chloramines (Najm and Trussell, 2001; Park et al., 2009a,b; Wilczak et al., 2003). Denver Water utilizes polyDADMAC prior to filtration. Evidence has been provided that small degradation products of polyDADMAC (including dimethylamine) are likely precursors that react with chloramines to form NDMA (Park et al., 2009a,b; Wilczak et al., 2003). It can therefore be expected that polyDADMAC contributes to increased NDMA-FP levels in waters post-filtration. A more surprising find, is the potential increase in NDMA precursors from filtered waters to waters in the distribution system. Relatively minimal research has so far been conducted regarding precursors originating from within the distribution system itself. One study found that new construction of an 8-km polyvinyl chloride (PVC) pipeline with rubber sealing rings and pipe-joining lubricant, followed by chloramines disinfection resulted in NDMA levels in excess of 100 ng/L (Morran et al., 2011). Further investigation of these materials revealed that NDMA originated from the rubber sealing rings (upon contact with chlorine, chloramines and even ultra-pure water), and rubber-lined gate valves were also found to release NDMA. Periods of stagnation were found to augment NDMA formation, and analysis of the pipeline after 2 years of operation provided data showing that NDMA was still being released in 10–25 ng/L higher levels than a parallel system (Morran et al., 2011). Another distribution study found elevated NDMA levels during routine monitoring in

126

G.C. Woods et al. / Science of the Total Environment 520 (2015) 120–126

the distribution system of a chloraminated treatment system; rubber gaskets used in a temporary storage tank were found responsible for the elevated NDMA concentrations (Teefy et al., 2011). To the best of our knowledge, no research has found conclusive evidence as to whether biological activity, such as nitrifying biofilms, are a potential sources of NDMA precursors, and these may also be potential contributors. 4. Conclusions A spatially and temporally comprehensive nitrosamine sampling campaign was undertaken at a large-scale urban drinking water system in Denver, Colorado to determine sources of nitrosamines. Over the duration of a year, NDMA was monitored at 3 sites in the distribution system and found to persist at higher concentrations at a remote site with over 200 h of residency. Further examination of the variability at this remote site revealed an apparent trend with temperature wherein lower temperatures resulted in higher NDMA concentrations, potentially attributable to less effective chlorine pre-oxidation during colder months. Though previous occurrence studies have failed to find seasonal trends in NDMA occurrence (Boyd et al., 2011; CDPH, 2002; Krasner et al., 2009; MOE, 2013; Russell et al., 2012; Woods and Dickenson, 2015), the current study, provides evidence of seasonal impacts on NDMA occurrence and corroborates previous bench-scale analyses of temperature and pH influences on NDMA formation (Krasner et al., 2012). Formation potential tests were further performed to estimate nitrosamine precursor loadings and were found to correlate well with actual NDMA formation in the distribution system. Estimates of NDMA precursors revealed that higher precursor levels were present in warmer summer months and following a catastrophic flooding event, during which elevated TOC was present and elevated levels of the polymer polyDADMAC were used. Seasonal influences and major water quality events therefore appear to be key influences on NDMA and NDMA precursor occurrence. Finally, at the most remote distribution site, it was noted that NDMA and potentially NDMA precursors were present at the highest concentrations of the entire system. Though previous research efforts have largely focused on raw water precursors and polymer aids used in treatment, data presented here provide evidence that further investigations into distribution system NDMA precursors may be warranted. Acknowledgments The authors would like to extend gratitude to Steve Lohman, Anita Hanagan, Heather Slater, Greg Zempel, and all the other Denver Water personnel involved with sample collection; special thanks is given to Heather Slater for the help in reviewing this manuscript. Appreciation is further extended to Dave Rexing, Jennifer Fuel Brett Vanderford, Janie Zeigler, Josephine Chu, Brittney Stipanov, Gurtar Kaur and all other SNWA personnel involved with laboratory analyses. References Barrett, S., Hwang, C., Guo, Y.C., Andrews, S.A., Valentine, R., 2003. Occurrence of NDMA in drinking water: a North American survey, 2001–2002. Proc. AWWA 2003 Annual Conference, Anaheim, Calif. Boyd, J.M., Hrudey, S.E., Richardson, S.D., Li, X.-F., 2011. Solid-phase extraction and highperformance liquid chromatography mass spectrometry analysis of nitrosamines in treated drinking water and wastewater. Trends Anal. Chem. 30, 1410–1421. CDPH (California Department of Public Health), 2002. Studies on the Occurrence of NDMA in Drinking Water. http://www.cdph.ca.gov/certlic/drinkingwater/ Documents/NDMA/NDMAstudies.pdf (accessed Mar. 30, 2014). CDPH (California Department of Public Health), 2013. NDMA and Other Nitrosamines — Drinking Water Issues. http://www.waterboards.ca.gov/drinking_water/certlic/ drinkingwater/NDMA.shtml (accessed Jan. 15, 2015). Choi, J., Valentine, R.L., 2002. Formation of N-nitrosodimethylamine (NDMA) from reaction of monochloramine: a new disinfection by-product. Water Res. 36, 817–824. Denver Water, 2014. http://www.denverwater.org (accessed Jun. 18, 2014). Dotson, A., Westerhoff, P., Krasner, S.W., 2009. Nitrogen enriched dissolved organic matter (DOM) isolates and their affinity to form emerging disinfection by-products. Water Sci. Technol. 60, 135–143.

Flowers, R.C., Singer, P.C., 2013. Anion exchange resins as a source of nitrosamines and nitrosamine precursors. Environ. Sci. Technol. 47, 7365–7372. Gerecke, A.C., Sedlak, D.L., 2003. Precursors of N-nitrosodimethylamine in natural waters. Environ. Sci. Technol. 37, 1331–1336. Holady, J.C., Trenholm, R.A., Snyder, S.A., 2012. Use of automated solid-phase extraction and GC–MS/MS to evaluate nitrosamines in water matrices. Am. Lab. 44, 25–30. Kemper, J.M., Westerhoff, P., Dotson, A., Mitch, W.A., 2009. Nitrosamine, dimethylnitramine, and chloropicrin formation during strong base anion-exchange treatment. Environ. Sci. Technol. 43, 466–472. Kemper, J.M., Walse, S.S., Mitch, W.A., 2010. Quaternary amines as nitrosamine precursors: a role for consumer products? Environ. Sci. Technol. 44, 1224–1231. Krasner, S.W., Westerhoff, P., Chen, B., Amy, G., Nam, S.-N., Chowdhury, Z.K., Sinha, S., Rittmann, B.E., 2008. Contribution of Wastewater to DBP Formation. AWWA Research Foundation Pub., Denver, CO. Krasner, S.W., Westerhoff, P., Chen, B., Rittmann, B.E., Amy, G., 2009. Occurrence of disinfection byproducts in United States wastewater treatment plant effluents. Environ. Sci. Technol. 43, 8320–8325. Krasner, S.W., Lee, C.F.T., Mitch, W.A., von Gunten, U., 2012. Development of a Bench-Scale Test to Predict the Formation of Nitrosamines. Water Research Foundation Pub., Denver, CO. Krasner, S.W., Mitch, W.A., McCurry, D.L., Hanigan, D., Westerhoff, P., 2013. Formation, precursors, control, and occurrence of nitrosamines in drinking water: a review. Water Res. 47, 4433–4450. Mitch, W.A., Sedlak, D.L., 2002. Factors controlling nitrosamine formation during wastewater chlorination. Water Sci. Technol. Water Supply 2 (3), 191–198. Mitch, W.A., Sedlak, D.L., 2004. Characterization and fate of N-nitrosodimethylamine precursors in municipal wastewater treatment plants. Environ. Sci. Technol. 38, 1445–1454. Mitch, W.A., Gerecke, A.C., Sedlak, D.L., 2003. A N-nitrosodimethylamine (NDMA) precursor analysis for chlorination of water and wastewater. Water Res. 37, 3733–3741. MOE (Ministry of the Environment), 2013. Drinking Water Surveillance Program. http:// www.ontario.ca/environment-and-energy/drinking-water-surveillance-programdwsp-data (accessed Apr. 1, 2014). Morran, J., Whittle, M., Fabris, R.B., Harris, M., Leach, J.S., Newcombe, G., Drikas, M., 2011. Nitrosamines from pipeline materials in drinking water distribution systems. J. Am. Water Works Assoc. 103, 76–83. Najm, I., Trussell, R.R., 2001. NDMA formation in water and wastewater. J. Am. Water Works Assoc. 93, 92–99. Nawrocki, J., Andrzejewski, P., 2011. Nitrosamines and water. J. Hazard. Mater. 189, 1–18. Park, S.-H., Piyachaturawat, P., Taylor, A.E., Huang, C.-H., 2009a. Potential Nnitrosodimethylamine (NDMA) formation from amine-based water treatment polymers in the reactions with chlorine-based oxidants and nitrosifying agents. Water Sci. Technol. Water Supply 9, 279–288. Park, S.-H., Wei, S., Mizaikoff, B., Taylor, A.E., Favero, C., Huang, C.-H., 2009b. Degradation of amine-based water treatment polymers during chloramination as Nnitrosodimethylamine (NDMA) precursors. Environ. Sci. Technol. 43, 1360–1366. QPC, 2005. Public Health Regulation 2005, Current as at 1 March 2014. Queensland Parliamentary Council (https://www.legislation.qld.gov.au/LEGISLTN/CURRENT/P/ PubHealR05.pdf (accessed Feb. 15, 2015)). Russell, C.G., Blute, N.K., Via, S., Wu, X., Chowdhury, Z., 2012. Nationwide assessment of nitrosamine occurrence and trends. J. Am. Water Works Assoc. 104, E205–E217. Schmidt, C.K., Brauch, H.-J., 2008. N, N-dimethylsulfamide as precursor for Nnitrosodimethylamine (NDMA) formation upon ozonation and its fate during drinking water treatment. Environ. Sci. Technol. 42, 6340–6346. Shah, A.D., Mitch, W.A., 2012. Halonitroalkanes, halonitriles, haloamides, and Nnitrosamines: a critical review of nitrogenous disinfection byproduct formation pathways. Environ. Sci. Technol. 46, 119–131. Shen, R., Andrews, S.A., 2011. Demonstration of 20 pharmaceuticals and personal care products (PPCPs) as nitrosamine precursors during chloramine disinfection. Water Res. 45, 944–952. Teefy, S., Chan, C.C., Wong, W., Work, L., 2011. NDMA formation from gaskets used in water storage tanks. Proc. AWWA 2011 WQTC, Pheonix, AZ. USEPA, 1991. Integrated Risk Information System (IRIS). http://www.epa.gov/IRIS (accessed Feb. 13, 2015). USEPA, 1993. N-nitrosodimethylamine CASRN 62-75-9. Integrated Risk Information Service (http://www.epa.gov/iris/subst/0045.htm (accessed Feb. 13, 2015)). USEPA, 2004. Method 521: determination of nitrosamines in drinking water by solid phase extraction and capillary column gas chromatography with large volume injection and chemical ionization tandem mass spectrometry (MS/MS). EPA document no. EPA/600/R-05/054. National Exposure Research Laboratory, Cincinnati, OH (http:// www.epa.gov/nerlcwww/documents/m_521.pdf (accessed Feb. 15, 2015)). USEPA, 2009. Drinking water contaminant candidate list 3 — final. Fed. Regist. 74 (194), 51850. USEPA, 2012. Unregulated Contaminant Monitoring Rule 2 (UCMR2). http://water.epa. gov/lawsregs/rulesregs/sdwa/ucmr/ucmr2 (accessed Feb. 15, 2015). WHO (World Health Organization), 2008. Guidelines for Drinking-water Quality. 3rd ed. (http://www.who.int/water_sanitation_health/dwq/gdwq3rev/en (accessed Feb.15, 2015)). Wilczak, A., Assadi-Rad, A., Lai, H.H., Hoover, L.L., Smith, J.F., Berger, R., Rodigari, F., Beland, J.W., Lazzelle, L.J., Kincannon, E.G., Baker, H., Heaney, C.T., 2003. Formation of NDMA in chloraminated water coagulated with DADMAC cationic polymer. J. Am. Water Works Assoc. 95, 94–106. Woods, G.C., Dickenson, E.R.V., 2015. Evaluation of the final UCMR2 database: nationwide trends in NDMA. J. Am. Water Works Assoc. 107, E58–E68. http://dx.doi.org/10.5942/ jawwa.2015.107.0009.

Seasonal and spatial variability of nitrosamines and their precursor sources at a large-scale urban drinking water system.

Nitrosamines are considered to pose greater health risks than currently regulated DBPs and are subsequently listed as a priority pollutant by the EPA,...
720KB Sizes 8 Downloads 12 Views