Investigation of Chloramines, Disinfection Byproducts, and Nitrification in Chloraminated Drinking Water Distribution Systems

Abstract

Four chloraminated drinking water distribution systems (CDWDSs) required to maintain numeric versus “detectable” residuals were spatially and temporally sampled for water quality and associated trihalomethane (THM) and haloacetic acid (HAA) formation. Monochloramine decreased from entry point (EP) to maximum residence time (MRT) samples while THMs and HAAs initially increased and then stabilized or slightly decreased. Subsequently, EP and MRT samples were used in laboratory-held studies to further evaluate disinfectant residual stability, chloramine speciation, and nitrification occurrence. MRT water exhibited a faster monochloramine concentration decline compared to EP water, indicating a decreasing disinfectant residual stability from increasing water age through distribution. Using a simple technique based on published inorganic chloramine chemistry, samples were also investigated for nondisinfectant positive interference (NDPI) on total chlorine measurements. NDPI concentrations represented up to 100% of the total chlorine concentration when total chlorine concentrations decreased to 0.05 mg-Cl₂/L, indicating little to no effective disinfectant residual remained.

Introduction

In the early 20th century, chloramines started gaining popularity in drinking water disinfection for being more persistent in distribution systems than free chlorine and causing fewer taste and odor issues (Kirmeyer et al. 2004). After the US Environmental Protection Agency (USEPA) established the Stage 1 and Stage 2 Disinfectants and Disinfection Byproducts (DBP) Rules (USEPA 1998, 2006), many utilities switched to chloramines for secondary disinfection to reduce the formation of regulated DBPs (AWWA 2013). With the increasing application of chloramination, the importance of understanding chloramine chemistry became progressively evident. Researchers have evaluated the effects of pH, temperature, reaction time, and the dose of free chlorine and free ammonia on the formation, speciation, and subsequent fate of inorganic chloramines, proposing chloramine reaction schemes (Crittenden et al. 2005; Duirk et al. 2005; Jafvert and Valentine 1992; Vikesland et al. 2001). Based on these reaction schemes, freely accessible web-based applications were developed to enable water treatment professionals to explore inorganic chloramine chemistry (Wahman 2018).

In comparison, much less is known about organic chloramines. Organic chloramines can form from the reaction of dissolved organic carbon (DOC) or dissolved organic nitrogen (DON) with inorganic chloramines or free chlorine (Deborde and von Gunten 2008; How et al. 2017). The complex precursor organic compounds lead to many different species of organic chloramines with varying formation and decay kinetics. Organic chloramines behave analogously to and cannot be distinguished from inorganic chloramines in commonly used drinking water methods [e.g., N, N-diphenylenediamine (DPD) colorimetric (Jensen and Johnson 1990b) and amperometric titration methods (Jensen and Johnson 1990a)] (How et al. 2017; Wahman and Pressman 2015). The difficulty in distinguishing organic chloramines, which are weaker disinfectants than their inorganic counterparts (Amiri et al. 2010; Donnermair and Blatchley 2003), creates a positive interference that leads to an overestimation of effective disinfectant residuals (Wahman and Pressman 2015). Recently, USEPA Method 127 (an indophenol colorimetric method) was developed for the specific determination of monochloramine (i.e., the effective disinfectant) that is not subject to organic chloramine interference (USEPA 2021).

Due to the required presence of ammonia, an inherent challenge in chloramination practice is the occurrence of nitrification: the oxidization of ammonia into nitrite and eventually nitrate by nitrifying microorganisms. Based on known chloramine chemistry, inorganic chloramines decay with time, while organic chloramines both decay and continue to form (Lee and Westerhoff 2009); thus, organic chloramines become a greater part of the resulting chloramine mixture (i.e., total chlorine) as total chlorine residuals decrease (Baribeau et al. 2005). The resulting lack of effective disinfectant residuals can create a favorable condition for nitrifying microorganisms to metabolize ammonia and proliferate, leading to nitrification to a greater extent. Nitrification can further deplete disinfectant residuals and cause biological and chemical deterioration of water quality. Researchers (Robertson 2014; Wahman and Pressman 2015) have noted the problems associated with very low (i.e., “detectable” or “trace”) disinfectant residuals in distribution systems. Furthermore, a recent study (Roth and Cornwell 2018) using water from 21 distribution systems across 13 states found that maintaining a “trace” residual concentration did not satisfy the chlorine demand of the water, while a major DBP increase occurred when raising the residual from “trace” to 0.2–0.5 mg-Cl₂/L; further residual increases only led to modest DBP formation. Overall, more research is warranted to better understand the balance between providing a chlorine or chloramine dose that can maintain a numeric residual while minimizing DBP formation. In addition, results from USEPA’s third Six-Year Review of national primary drinking water regulations indicated that regulations requiring a “detectable” residual, as the USEPA (1989) defines in the Surface Water Treatment Rule (SWTR), are candidates for revision (USEPA 2016). In the current manuscript, “detectable” refers to the SWTR definition, which is not a specific, numeric value.

Overall, the study results aimed to provide practical information as the USEPA moves forward with evaluating whether to revise regulations requiring “detectable” residuals by providing information on NDPI and regulated DBP concentration (USEPA 2016). Specifically, there were three objectives in the current study. First, for four (three surface water and one groundwater) chloraminated drinking water distribution systems (CDWDSs) where state regulations require greater than “detectable” disinfectant residuals, spatial and temporal sampling was used to investigate water quality and associated concentrations of regulated trihalomethanes (THMs) and haloacetic acids (HAAs), allowing assessment of simultaneously maintaining a numeric residual and meeting DBP regulations. The collected water quality and DBP data also served as primary data for two related publications evaluating the presence of opportunistic pathogens in these same systems (Pfaller et al. 2021; Zhang et al. 2021). Second, the collected water samples were leveraged to conduct controlled, laboratory hold studies, allowing further evaluation of disinfectant residual stability, effective (i.e., inorganic chloramines) and noneffective (e.g., organic chloramines) disinfectant residual concentrations, and nitrification occurrence while removing the confounding impacts associated with distribution systems (e.g., unknown hydraulics, pipe wall reactivity, and varying water quality and treatment). Third, a simple technique based on published chloramine chemistry was used to investigate nondisinfectant positive interference (NDPI), presumably from organic chloramines, on total chlorine DPD measurements, allowing assessment of the effective disinfectant residual (i.e., inorganic chloramines) under varying residual concentrations and residence times (RTs).

Materials and Methods

Sample Collection, Handling, and Analysis

Water samples were collected quarterly from four full-scale CDWDSs described in Table S1. As shown in Fig. 1, three of the four systems utilize surface water (SW1, SW2, SW3) and one utilizes groundwater (GW) as source waters. Sampling locations in each CDWDS were selected based on the expertise of the water treatment plant (WTP) operators to include sites (1) at the system entry point (EP); (2) representing an average RT (ART); (3) near and downstream of a storage tank (Tank); and (4) at an extreme end of the distribution system, corresponding to the maximum RT (MRT). Because the GW CDWDS did not have any tanks in the distribution system, samples were collected from the EP, two ARTs (ART1 and ART2), and one MRT site.

Fig. 1.

Chloraminated drinking water distribution system (CDWDS) sampling plan for field and laboratory-scale studies. *Samples from these sites were used in the laboratory hold studies.

At the time of collection, samples were analyzed in the field by WTP personnel for total chlorine, monochloramine, free ammonia, and nitrite concentrations as well as pH and temperature. Total chlorine and monochloramine were analyzed following the standard Hach DPD and Indophenol methods (Hach 2019, 2022). Samples were also collected for the analysis of (1) the four regulated trihalomethanes (THM4), including bromodichloromethane (BDCM), bromoform (TBM), chloroform (TCM), and chlorodibromomethane (CDBM), following USEPA Method 551.1; and (2) nine haloacetic acids (HAA9), including monochloroacetic acid (MCAA), dichloroacetic acid (DCAA), trichloroacetic acid (TCAA), monobromoacetic acid (MBAA), dibromoacetic acid (DBAA), bromochloroacetic acid (BCAA), bromodichloroacetic acid (BDCAA), chlorodibromoacetic acid (CDBAA), and tribromoacetic acid (TBAA), following USEPA Method 552.3. Among the HAA9, the first five (HAA5) are currently regulated by the USEPA. To provide water for laboratory-scale hold studies, additional samples were taken in 1-L high-density polyethylene (HDPE) bottles. Water samples were shipped overnight on ice to the USEPA in Cincinnati, OH. A complete list of all analytical methods, instruments, and services used in this study is summarized in Table S2.

Laboratory-Scale Hold Studies

Experimental Setup

For laboratory-scale hold studies (Alexander et al. 2019, 2020), samples from the EP and MRT sites of each CDWDS were transferred into a series of 60-mL chlorine-demand-free amber vials (Summers et al. 1996), sealed headspace free, and kept in the dark at room temperature (20–25°C) for temporal analysis. Sampling occurred every day during days 0–3, every other day during days 5–15, and twice per week during days 15–45. For each sampling event, two vials (one EP and one MRT) from each CDWDS were sacrificed. Samples were analyzed for total chlorine, monochloramine, free ammonia, and nitrite concentrations along with pH and temperature. In addition, dichloramine and NDPI were estimated using a simple technique specified in the following section. Sampling continued until day 45 or until the total chlorine concentration depleted to or below the method detection limit of 0.02 mg-Cl₂/L (Hach method 8167) for two consecutive analyses, whichever occurred first.

Dichloramine and NDPI Estimation

Dichloramine and NDPI concentrations were estimated using commonly used Hach methods and sample pH cycling (hereafter referred to as the pH cycling method, see Table S2 for related analytical methods). As the bromide ion concentrations in the waters were low (i.e., <0.05 mg/L, Table S1), the dominant inorganic chloramines present were monochloramine and dichloramine (i.e., bromamine and bromochloramine concentrations were negligible). Monochloramine and dichloramine are innately unstable. Monochloramine is prone to auto-decomposition at neutral pH but becomes more stable with increasing pH (Jafvert and Valentine 1992; Vikesland et al. 2001). In contrast, increasing pH decreases dichloramine formation and increases dichloramine loss (Jafvert and Valentine 1987, 1992; Wahman 2019). Therefore, increasing the pH of a sample containing monochloramine and dichloramine results in dichloramine removal and minor changes in monochloramine concentration, and this chemistry was the basis for the pH cycling method.

Practically, the pH cycling method consists of the following steps on a sample split into two 30-mL subsamples: (1) on the first subsample, analyze for total chlorine $(C_{total1})$ and monochloramine $(C_{mono1})$ concentrations; and (2) on the second subsample, (a) adjust pH to between 10 and 11, with 10 M sodium hydroxide solution, (b) store in the dark for 10 min to allow dichloramine to decrease to negligible levels, (c) adjust pH to below 9.5 with 6 M hydrochloric acid, and (d) analyze for total chlorine $(C_{total2})$ and monochloramine $(C_{mono2})$ concentrations. The basis for the 10-min reaction time in the pH cycling method [step 2(b)] results from the major reaction causing dichloramine decay at high pH conditions (Jafvert and Valentine 1992; Wahman 2018), which is base-catalyzed dichloramine hydrolysis reaction to the unknown intermediate “I” [Eq. (1)]. Eq. (2) is the corresponding reaction rate expression where the reaction rate constant $(k_{DI})$ is 110 M⁻¹s⁻¹ (Jafvert and Valentine 1992). At pH 10, the hydroxide ion concentration, [OH⁻] is 10⁻⁴M, and the reaction rate expression becomes first-order [Eq. (3)]. Eq. (4) represents dichloramine loss in a batch system (i.e., samples) where $C_0$ is the initial dichloramine concentration and $C$ is the dichloramine concentration at time $t$ in seconds

$${\text{NHC1}}_2+{\mathrm{H}}_2\mathrm{O}\to \mathrm{I}$$ (1)

$$r=k_{DI}[{\text{NHC1}}_2][{\text{OH}}^{-}]=(110{\mathrm{M}}^{-1}{\mathrm{s}}^{-1})[{\text{NHC1}}_2][{\text{OH}}^{-}]$$ (2)

$$r=(0.011{\mathrm{s}}^{-1})[{\text{NHC1}}_2]$$ (3)

$$C∕C_0={\mathrm{e}}^{(-0.011\mathrm{s}-1)t}$$ (4)

From Eq. (4), the time needed for 99% dichloramine removal $(C∕C_0=0.01)$ is 419 seconds (7.0 min). The pH cycling method used a 10-min reaction time at a minimum pH of 10 to add a safety factor (43% increase over the estimated 7.0 min) to ensure the removal of dichloramine. Subsequently, a verification test was conducted with an anticipated worst-case scenario on a monochloramine/dichloramine solution (chlorine to ammonia-nitrogen mass ratio = 5.1, pH = 6.5) to verify if 10 min was long enough for dichloramine removal. The test results supported the applicability of the pH cycling method, and a detailed discussion of the results is provided in the Supplemental Materials (pH Cycling Method and Fig. S1).

Using the pH cycling method, NDPI $(C_{NDPI})$ and dichloramine $(C_{di})$ concentrations in the EP and MRT hold study samples (HSS) were calculated. Because EP-HSS and MRT-HSS contained free ammonia between 0.2 and 0.6 mg-N/L, free chlorine was negligible; therefore, it was reasonable to assume that the total chlorine in the first subsample $(C_{total1})$ was composed of monochloramine $(C_{mono1})$, dichloramine $(C_{di})$, and NDPI $(C_{NDPI})$ [Eq. (5)]. After pH cycling of the second subsample, the $C_{di}$ was expected to be removed through its rapid hydrolysis, resulting in the total chlorine $(C_{total2})$ consisting only of monochloramine $(C_{mono2})$ and NDPI $(C_{NDPI})$ [Eq. (6)]. Therefore, $C_{NDPI}$ was calculated using Eq. (6). Subsequently, $C_{di}$ was calculated using $C_{NDPI}$, $C_{total1}$, and $C_{mono1}$ [Eq. (5)]. It should be noted that the pH cycling may also destroy some organic chloramines as the stability of different classes of organic chloramines are affected by changes in pH (How et al. 2017), including faster degradation at pH greater than 10 due to base catalysis (Antelo et al. 1996) and decomposition rates being a linear function of hydroxide ion concentration (Szabó et al. 2020). Therefore, the pH cycling method results in a conservative (i.e., potentially biased low) estimate of NDPI as it presumably results mostly from organic chloramines

$$\text{Firstsubsample}:C_{total1}=C_{mono1}+C_{di}+C_{NDPI}$$ (5)

$$\text{Second subsample after pH cycling}:C_{NDPI}=C_{total2}-C_{mono2}$$ (6)

Simulation of Inorganic Chloramine Stability with a Web-Based Application (WBA)

Temporal EP-HSS and MRT-HSS measured monochloramine and dichloramine concentrations were compared to their simulated concentrations with a freely accessible WBA (accessible at https://usepaord.shinyapps.io/Unified-Combo/ or https://shiny.epa.gov/cfd/; source code downloadable from https://github.com/USEPA/Inorganic_Chloramine_-Formation_and_Decay_Application). Wahman (2018) provides a detailed description of WBA development, implementation, and instructions for use. Day 0 concentrations of EP-HSS and MRT-HSS from each CDWDS were entered as initial conditions, and the total organic carbon (TOC) was set to zero to only consider intrinsic inorganic chloramine instability in simulations. Therefore, the WBA simulations represent the baseline inorganic chloramine stability expected from theory (i.e., the maximum expected residual stability), and if experimental data are well represented by the WBA simulations, then other demand reactions are not impacting residual stability to a greater extent than is expected based on inherent inorganic chloramine chemistry.

Results and Discussion

Spatial and Seasonal Changes in Water Quality in Full-Scale Distribution Systems

Chloramine Residuals

Water quality analyses at the time of sampling provided an understanding of changes in water quality in each distribution system from the EP to the MRT (in increasing order of assumed RT and during each season) (Fig. 2). For reference, Table S3 summarizes the driving distances of sampling sites relative to the EPs and was used to assign relative RTs. Distribution systems that supplied treated surface waters (SW1, SW2, and SW3) displayed similar patterns: (1) the three system’s average EP total chlorine concentration was 3.68±0.21 mg-Cl₂/L, and the residuals at the MRT varied within the range of 0.19–2.14 mg-Cl₂/L, dropping twice (0.43, 0.19 mg-Cl₂/L) below the state’s minimum residual requirement of 0.5 mg-Cl₂/L; (2) the average monochloramine concentration was 3.14±0.22 mg-Cl₂/L at the EP and experienced a gradual decrease toward the MRT (0.25–1.26 mg-Cl₂/L, except one below detection case), with the sum of dichloramine and NDPI accounting for 19% to nearly 100% of total chlorine residuals; (3) free ammonia within 0.2 to 0.6 mg-N/L were detected; and (4) SW1 and SW3 had low levels of nitrite (0.002–0.135 mg-N/L), while SW2 maintained nitrite levels below 0.01 mg-N/L at all times except in June 2018 at the MRT (0.034 mg-N/L). In contrast, the treated groundwater (GW) entered the distribution system with greater total chlorine (5.44±0.95 mg-Cl₂/L) and monochloramine concentrations (4.42±0.63 mg-Cl₂/L) and maintained total chlorine residuals between 2.12 and 4.61 mg-Cl₂/L and monochloramine residuals between 1.48 and 3.86 mg-Cl₂/L at the MRT, with the sum of dichloramine and NDPI accounting for 16–56% of the total chlorine residuals. For the GW system, which has naturally occurring ammonia, the annual average free ammonia concentration throughout the system was 2.36 ± 0.24 times as great as those in SW1, SW2, or SW3, and nitrite levels were below 0.01 mg-N/L. As expected, disinfectant residuals followed a general decreasing trend from EP to MRT in all four systems. Throughout all systems, except in SW3 in June and September, no notable increase of nitrite and accompanying decrease of free ammonia were observed, possibly indicating the absence of undesired nitrification, which is discussed in detail later in the manuscript.

Fig. 2.

Quarterly analysis of water quality in four full-scale chloraminated drinking water distribution systems: (a) SW1; (b) SW2; (c) SW3; and (d) GW. Sampling sites were listed in the order of assumed increasing residence time: entry point (EP), average residence time (ART; ART2 for GW), Tank (or ART1 for GW), and maximum residence time (MRT). Legend in panel (a) applies to all panels. T = temperature. The distances (in km) noted parenthetically in the abscissa labels are the driving distances of sampling sites relative to the EP.

For a visual representation of the seasonal and spatial effects on monochloramine residual, the desired disinfectant (AWWA 2006; Ratnayaka et al. 2009), Fig. 3 presents the monochloramine data normalized by the corresponding EP monochloramine concentration in the assumed order of increasing RT. Monochloramine concentration generally decreased from EP to MRT, with a few unexpected increases at MRT relative to Tank [Figs. 2(c and d) and 3(c and d)] or Tank relative to ART [Figs. 2(a) and 3(a)] that potentially resulted from the same-day collection of water samples from all sampling locations. Such a sampling scheme should reflect the general trend in water quality throughout the distribution system, but exceptions may occur depending on the magnitude of the day-to-day fluctuations in system operating conditions and treated water quality. For GW, there were two additional reasons for unexpected trends. First, the ART1 and ART2 sample locations were not in line hydraulically with the MRT sample location. ART2 was on the opposite side, and ART1 was on the same side but on a separate branch of the distribution system than the MRT, respectively. Second, GW has naturally occurring source water ammonia and uses ion exchange (IX) for softening, leading to highly variable treated free ammonia concentrations prior to forming chloramines. The variability in free ammonia creates issues for the system to make a stable and consistent chloramine residual, as highlighted in a recent publication for similar CDWDSs using IX with naturally occurring ammonia (Keithley et al. 2021).

Fig. 3.

Entry point normalized monochloramine concentrations from full-scale chloraminated drinking water distribution system samples by season: (a) SW1; (b) SW2; (c) SW3; and (d) GW. Sampling sites were listed in the assumed order of increasing residence time: entry point (EP), average residence time (ART; ART2 for GW), Tank (ART1 for GW), and maximum residence time (MRT). Legend in panel (a) applies to all panels. The distances (in km) noted parenthetically in the abscissa labels are the driving distances of sampling sites relative to the EP.

SW3 displayed faster monochloramine loss in June and September compared to March and December. This may be attributed to the near 100% increase in temperature from December and March (15.2°C average) to June and September (29.2°C average), potentially accelerating reactions and enhancing microbial activities (Arevalo 2007; Kiéné et al. 1998; Powell et al. 2000; Sathasivan et al. 2009). In contrast, the other three distribution systems (SW1, SW2, and GW) that experienced only 3–8°C increases in temperature from December/March to June/September showed some counterintuitive results—a slower monochloramine loss in June and September (only June for SW1) than in December (SW2, GW) and March (SW2).

Chloramine stability is complex and is affected by many factors such as pipe material and diameter, system hydraulics, and pH (Arevalo 2007; Vikesland and Valentine 2000, 2002a, b). Besides, various additional water constituents are important, including natural organic matter (NOM) (Duirk 2003; Duirk et al. 2005 2006), nitrite (Margerum et al. 1994; Vikesland et al. 2001; Wahman and Speitel 2012), microorganisms (Herath et al. 2015; Maestre et al. 2013), and chloramine-decaying soluble microbial products (Herath et al. 2018; Krishna et al. 2012; Wahman et al. 2016). Therefore, apart from contributing to the spatial changes in chloramine from EP to MRT in the distribution system, when seasonal temperature changes were subtle, the changing and complex effects of NOM and other water constituents may have affected the seasonal patterns in SW1, SW2, and GW.

THM4 and HAA9

NOM-chloramine reactions decrease chloramine stability and lead to regulated DBP (i.e., THM4 and HAA5) formation, although at lower levels than from NOM-chlorine reactions (Goslan et al. 2009; Hua and Reckhow 2007; Lu et al. 2009; Sakai et al. 2016). THM4 concentrations in samples collected from the four CDWDSs ranged from 6.7 μg/L to 127 μg/L, with the locational running annual averages (LRAA) staying below the USEPA maximum contaminant level (MCL) of 80 μg/L, except at the ART (86 μg/L) and the Tank (83 μg/L) of SW1. As for HAA9, in all four systems, MBAA, MCAA, and TBAA were nondetectable. HAA9 concentrations in SW1, SW2, and GW were in the range of 1.5–69 μg/L, and HAA5 was in the range of 1.5–56.7 μg/L. The HAA5 LRAAs at all sampling locations in all four systems were below the MCL of 60 μg/L. Seasonally, THMs showed increased formation during warmer seasons across all four CDWDSs, while HAAs showed varying seasonal patterns. Spatially, in terms of hydraulic RT, THMs and HAAs initially increased and then stabilized or slightly decreased. Individual THM and HAA concentrations (Figs. S2 and S3) and a more detailed discussion on seasonal and spatial changes in DBPs are provided in the Supplemental Materials (THM4 and HAA9).

Time Dependent Changes in Water Quality in Laboratory-Scale Hold Studies

Inorganic Chloramines

In the laboratory-scale hold studies, WBA simulations were generally representative of the actual monochloramine loss patterns in the EP-HSS, but monochloramine in the MRT-HSS decreased substantially faster than in the WBA simulations. Fig. 4 shows the experimental data and WBA simulations for SW2 hold studies. SW2 was chosen as the representative system because: (1) it provided complete year-round data for both EP-HSS and MRT-HSS; and (2) the IX-related issues in GW and the turbidity increase in GW samples during the pH cycling (see Supplemental Materials, pH cycling method) interfered with the assessment of chloramine species. Data for the remaining three systems are provided in the Supplemental Materials, and the remaining three systems displayed similar results (Figs. S4-S6). The EP reflects the quality of finished water leaving the WTP, which contains minimum concentrations of organics, nitrite, microorganisms, and other undesired contaminants, aligning with the WBA simulation conditions that only include intrinsic inorganic chloramine reactions. Therefore, the WBA provided a reasonable simulation of inorganic chloramine concentrations for the EP-HSS. In contrast, the MRT contained lesser chloramine residuals and potentially varied water constituents from transportation through the distribution system. Hence, chloramine demand reactions in MRT-HSS are expected to be of greater importance, leading to faster monochloramine loss compared to the WBA simulation, as reflected in Figs. 4(e-h). Researchers have quantified monochloramine loss in bulk water using simplified pseudo-first-order reaction kinetics (Arevalo 2007; Sathasivan et al. 2005). In the current study, we determined monochloramine loss rates using a similar pseudo-first-order kinetics approach, where the effects of various factors [i.e., NOM (TOC), inorganic ions, nitrifiers and other microbes from pipe scales, biofilm, and sediment that can directly react or excrete organics that can react with monochloramine (Arevalo 2007; Duirk 2003; Herath 2014; Herath et al. 2018; Maestre et al. 2013; Wahman et al. 2016; Wahman and Speitel 2012)] are all combined into one overall reaction rate constant. Calculated pseudo-first-order decay rate constants (Table S6) agreed with a previous study (Sathasivan et al. 2005) and provided a quantitative comparison of the different monochloramine loss patterns observed in Fig. 4. Detailed discussion on the pseudo-first-order decay rate constants is provided in the Supplemental Materials Pseudo-first-order decay rate and Table S6.

Fig. 4.

(a–d) Chloramine concentrations and model simulations for entry point hold study samples (EP-HSS); and (e–h) maximum residence time hold study samples (MRT-HSS). Hold studies initiated from samples originating from SW2. Legend in panel (a) applies to all panels. WBA = web-based application.

Fig. 4 also includes the dichloramine concentrations estimated using the pH cycling method. Overall, dichloramine loss patterns were comparable to WBA simulations. Most chloramination practices use pH values between 7.5 and 8.5, where monochloramine is stable and predominant (Ratnayaka et al. 2009). The four CDWDSs maintained pH values on the lower end of this range (i.e., between pH 6.5–7.9), which produced measurable initial dichloramine concentrations in EP-HSS. Dichloramine concentrations accounted for 23%–41% of total chlorine in the HSS. As explained in the pH cycling method description, due to the potential loss of some NDPI from organic chloramine loss during pH cycling, dichloramine concentrations may have been overestimated, yielding a conservative (i.e., potentially low biased) estimation of NDPI, but because WBA simulated and experimentally measured dichloramine concentrations in the EP-HSS and MRT-HSS coincided, the impact appears to be minor in these samples.

Nitrification

Nitrite concentrations were monitored during the EP and MRT hold studies to investigate the onset and occurrence of nitrification in EP-HSS and MRT-HSS [Figs. 5(a and e)]. As expected, nitrite concentrations were low in EP-HSS (mostly around 0.01 mg-N/L) but were measurable in some MRT-HSS where they continued to increase temporally. Increasing nitrite concentrations in MRT-HSS in December and June corresponded to lower monochloramine residuals in samples from the same month [Figs. 4(e and g)], indicating a possible effect of nitrification. Similar patterns were observed in samples from the remaining three distribution systems (Figs. S7-S9).

Fig. 5.

(a–d) Nitrogen mass balance in entry point hold study samples (EP-HSS); and (e–h) maximum residence time hold study samples (MRT-HSS). Legend in panel (a) applies to all panels. Hold studies initiated from samples originating from SW2.

Continuous biological nitrification may occur at a baseline level (nitrite concentration <0.01 mg-N/L) in CDWDSs without causing a water quality concern (Wilczak 2011). However, nitrification episodes may occur when the environment allows the growth of ammonia-oxidizing microorganisms (AOM) to exceed their inactivation from monochloramine or toxicity from biologically degrading THMs (Gomez-Alvarez et al. 2013; Speitel et al. 2011; Wahman et al. 2005, 2006a, b). Apart from some areas closest to the WTP, AOM are pervasive in distribution systems (Wilczak 2011). As stated previously, EP-HSS are expected to contain minimum levels of bacteria, low nitrite, and greater monochloramine stability, which was observed throughout the hold studies. A slight increase in free ammonia concentrations over time in EP-HSS [Fig. 5(b)] further confirms low AOM activity. Whereas, in the MRT-HSS, low chloramine residuals and sufficient levels of free ammonia promoted the growth of existing AOM and associated nitrification. Fig. 5(f) depicts the consumption (oxidization) of free ammonia by AOM in MRT-HSS. Nitrite produced from free ammonia oxidation can exert a chloramine demand and release more free ammonia (Margerum et al. 1994). In addition to oxidizing free ammonia, AOM and mixed culture nitrifiers may also biologically remove monochloramine via cometabolism under drinking water conditions and produce microbial products that exert a chloramine demand (Maestre et al. 2013, 2016; Wahman et al. 2016). Thus, nitrification chemically and biologically accelerates monochloramine loss.

Nitrification occurs frequently in CDWDSs (Kirmeyer et al. 2004; Wilczak et al. 1996), including both incomplete nitrification (nitrite formation) and complete nitrification where nitrite-oxidizing bacteria further oxidize nitrite into nitrate. Some utilities detected increased nitrate-nitrogen (nitrate-N) concentrations without observing any changes in nitrite-N concentrations (Wilczak et al. 1996), possibly due to prolonged nitrification (all nitrite oxidized to nitrate). Wilczak et al. (1996) suggested nitrite/nitrate simultaneous monitoring, a precise nitrogen mass balance, and/or specific biomonitoring of nitrifiers as several recommended ways to detect nitrification. Therefore, a nitrogen mass balance was conducted to investigate the possible production of nitrate in HSS. Fig. 5 presents the changes in the concentration of nitrite, free ammonia, available ammonia-N (sum of inorganic chloramine-N and free ammonia-N), and available ammonia-N + nitrite-N in samples from SW2 (see Figs. S7-S9 for the remaining three systems). To help explain what the changes in the different groups of nitrogen-containing species in Fig. 5 imply, Table 1 describes three simplified scenarios. Without nitrification, the inherent instability of monochloramine (neglecting small amounts of dichloramine for simplification) causes changes in the form of nitrogen. A previous study (Vikesland et al. 1998) experimentally measured monochloramine and its decomposition products and reported that most monochloramine-N decays to free ammonia-N (47%) and nitrogen gas (N₂, 43%) with low amounts of nitrate-N (1.5%). In the presence of NOM, the decomposition produced slightly greater amounts of free ammonia-N (53%) and nitrate-N (7%) and lower amounts of N₂ (11%) and some unknown organic species. Therefore, in scenario A (Table 1, abiotic decomposition), the free ammonia concentration should increase because of monochloramine decomposition, whereas available ammonia-N and available ammonia-N + nitrite-N should decline due to the loss of nitrogen as N₂ and/or as some unknown organic species. In scenario B (Table 1, incomplete nitrification), oxidization (and hence loss) of free ammonia should occur with an increase in nitrite concentration. Furthermore, while available ammonia-N is expected to decrease due to the loss of free ammonia and monochloramine, available ammonia-N + nitrite-N should stay relatively consistent. Finally, scenario C (Table 1, complete nitrification) is similar to scenario B but with a decrease (or no increase) in nitrite concentration due to further oxidization to nitrate and hence a decrease in available ammonia-N + nitrite-N. In Fig. 5, changes in different forms of nitrogen in EP-HSS resembled the description of scenario A, whereas the MRT-HSS were closer to scenario B. Therefore, this nitrogen mass balance not only corresponded well with our previous analysis of monochloramine loss, where EP-HSS closely followed WBA simulations, but it also provided further insight into the extent of nitrification in MRT-HSS—that is, most likely, incomplete nitrification.

Table 1.

Changes in the forms of nitrogen under different scenarios

Scenario	Included processes	Nitrite	Free ammonia	Available ammonia-N (chloramine-N + free ammonia-N)	Available ammonia- N + nitrite-N
A	Abiotic monochloramine decomposition only	Consistently below 0.01 mg-N/L	Increase due to production from monochloramine decomposition	Decrease due to loss as nitrogen gas	Decrease due to loss as nitrogen gas
B	Incomplete nitrification only—nitrite production	Increase from the conversion of free ammonia	Decrease due to consumption by ammonia-oxidizing microorganisms	Decrease due to conversion to nitrite	Consistent
C	Complete nitrification only—nitrate production	Temporary increase, and then decrease; or negligible nitrite production	Decrease due to consumption by ammonia-oxidizing microorganisms	Decrease due to conversion to nitrite and then nitrate	Decrease due to conversion to nitrate

NDPI

The estimated NDPI in the HSS likely reflects a transient concentration, resulting from both formation and loss of organic chloramines impacting the measured NDPI. Throughout the hold studies, NDPI decreased slowly over time (Fig. S10). Most of the EP-HSS initially had NDPI concentrations between 0.15 and 0.30 mg-Cl₂/L, which decreased by 17%–62% through day 5 and then more slowly thereafter. Lee and Westerhoff (2009) reported comparable observations when 16 NOM solutions were prepared using NOM isolates collected from 14 surface waters and two chloraminated wastewater treatment plant effluents. When monochloramine was added at an initial concentration of 3 mg — Cl₂/L, which is close to the monochloramine concentrations for the EP samples of SW1, SW2, and SW3 (Fig. 2), organic chloramine (presumably the major component of NDPI in the current work) concentrations reached a mean value of 0.16 mg-Cl₂/L after 24 h (with 25th and 75th percentiles being approximately 0.05 and 0.28 mg-Cl₂/L organic chloramines, respectively). After 5 days, the concentrations decreased by 50% on average. Another study (Zhang et al. 2016) chloraminated algal NOM with 5 mg — Cl₂/L monochloramine, and the organic chloramine percentage of total chlorine reached its greatest level (22%, about 0.5 mg-Cl₂/L organic chloramine) after 24 h. Initial NDPI concentrations in MRT-HSS were generally in the range of 0.05–0.15 mg-Cl₂/L, close to the day-5 concentration range of NDPI in EP-HSS, which may roughly be the RT to reach the MRT locations in the participating CDWDSs.

Despite the aforementioned decline in NDPI concentrations, the fraction of total chlorine associated with NDPI notably increased as the total chlorine concentration decreased. Fig. 6 presents the NDPI fractions in EP-HSS and MRT-HSS at representative high, medium, and low total chlorine residual levels that were measured in all three SW EP-HSS and MRT-HSS. The NDPI fraction of total chlorine likely increased because inorganic chloramines were decaying faster than the net loss of NDPI (Fig. 4 versus Fig. S10). Therefore, as total chlorine decreased, NDPI became a greater proportion of the total chlorine measurement. NDPI, which contains organic chloramines that are ineffective for disinfection (Amiri et al. 2010; Donnermair and Blatchley 2003; Scully et al. 1996), accounted for about 14%–32% of the 0.5 mg-Cl₂/L total chlorine concentrations in the EP-HSS [Fig. 6(a)]; were within 18%–38% in MRT-HSS that maintained 0.3 mg-Cl₂/L total chlorine residuals [Fig. 6(b)]; exceeded 50% of the 0.1 mg-Cl₂/L total chlorine concentration in the MRT-HSS [Fig. 6(b)]; and represented as great as 100% of total chlorine when the total chlorine concentration decreased to 0.05 mg-Cl₂/L in the MRT-HSS [Fig. 6(b)].

Fig. 6.

Hold study changes in the nondisinfectant positive interference (NDPI) fraction of total chlorine binned by total chlorine concentration for: (a) entry point hold study samples (EP-HSS); and (b) maximum residence time hold study samples (MRT-HSS). Hold studies initiated from samples originating from three full-scale chloraminated drinking water distribution systems (SW1, SW2, and SW3).

Previous research reported an increase in organic chloramine fraction of total chlorine with increasing detention time in pipe loop studies (Baribeau et al. 2005), reaching close to 100% when the total chlorine concentration dropped below 0.2 mg-Cl₂/L. Because of the issues related to organic chloramine formation and persistence, Wahman and Pressman (2015) expressed concern over maintaining effective disinfectant residuals in CDWDSs. At very low concentrations (e.g., less than 0.2 mg Cl₂/L total chlorine), the total chlorine measurements in CDWDSs are likely composed of NDPI (i.e., organic chloramines) and thus lack an effective disinfectant (i.e., monochloramine), potentially rendering the disinfection barrier of the multibarrier approach (USEPA 2002) weak and leading to issues such as nitrification and microbial regrowth that may also impact corrosion control and metal-release (Zhang et al. 2008, 2009a, b). This underscores the importance of ensuring the maintenance of effective disinfectant residuals (i.e., monochloramine) in CDWDSs. Using the indophenol method to measure inorganic monochloramine (Table S2) would help with the proper maintenance of an effective disinfectant residual. Recently, USEPA Method 127 (USEPA 2021) has been published for measuring monochloramine in drinking water using the indophenol method. If only total chlorine is measured, then maintaining greater total chlorine residuals that are below the maximum residual disinfectant level may be necessary.

Conclusion

With respect to the study objectives, the effects of seasonal and spatial changes in inorganic chloramines (mainly monochloramine) and DBPs were investigated. As required by their state regulations, all participating full-scale CDWDSs maintained a minimum numeric total chlorine residual, except in two cases (SW1 and SW3 in September), and maintained regulated DBP compliance, except in two cases (ART and Tank) in SW1 for THM4. In general, both monochloramine loss and the formation and presence of DBPs did not show clear seasonal patterns. Spatially or in terms of hydraulic RT in the distribution system, monochloramine concentrations decreased from the EP to the MRT, while THMs and HAAs initially increased and then stabilized or slightly decreased.

Hold studies enabled further investigation of disinfectant residual stability, chloramine concentrations, nitrite production, and nitrogen species evolution over time in a controlled laboratory environment. The temporal measured concentrations in the EP-HSS generally followed WBA simulations, indicating that the finished water presumably contained minimum levels of contaminants that imposed additional chloramine demand. This result does not mean that NOM demand reactions were not important for the EP-HSS, but NOM demand reactions did not impact residual stability greater than what is expected from inherent inorganic chloramine chemistry. Overall, relative NOM demand reaction importance will depend on source water characteristics and subsequent treatment (Alexander et al. 2020). In contrast, MRT-HSS exhibited faster monochloramine residual loss than WBA simulations, indicating that increased water age led to a less stable chloramine residual from traversing in the distribution system than inherent inorganic chemistry would predict. EP-HSS maintained very low nitrite concentrations throughout the hold studies. MRT-HSS, however, displayed an increase in nitrite concentration occurring with the fast depletion of monochloramine and free ammonia concentrations, indicating nitrification occurrence.

During the hold studies, a simple and widely accessible technique was used to quantify the NDPI impact on total chlorine measurement. Results revealed an increase in the NDPI fraction of total chlorine as the total chlorine concentration declined. The NDPI fraction exceeded 50% when the total chlorine residual was at 0.1 mg-Cl₂/L and reached as great as 100% when the total chlorine residual was at 0.05 mg-Cl₂/L, indicating the lack of an effective disinfectant at low, yet detectable, total chlorine concentrations. An understanding of the changes in the NDPI (i.e., organic chloramine) fraction of total chlorine and the ability to distinguish and monitor effective disinfectants using chloramine-specific methods will help utilities to maintain sufficient disinfectant residuals to enhance the disinfection barrier and reduce the occurrence of nitrification episodes.

Data Availability Statement

Some or all data, models, or code that support the findings of this study are available from the corresponding author upon reasonable request (all data contributing to figures/tables). Some or all data, models, or code generated or used during the study are proprietary or confidential in nature and may only be provided with restrictions (water utility identities are confidential). Some or all data, models, or code generated or used during the study are available in a repository online in accordance with funder data retention policies (https://doi.org/10.23719/1520714) (all data contributing to figures/tables).