Fate of ammonia and implications for distribution system water quality at four ion exchange softening plants with elevated source water ammonia

Abstract

Hard water and elevated ammonia are problems for many United States groundwater drinking water utilities, and some utilities, particularly those in the Midwest, face both challenges. Ion (cation) exchange (IX) is a common treatment technique for hardness reduction (i.e., softening) and may be used to remove ammonia as well, but these constituents may compete in IX and impact overall treatment performance. Few data have been reported on the impact on ammonia concentrations when using IX for softening in full-scale systems. This study investigated four full-scale groundwater treatment plants in Illinois that practice IX for softening (raw water hardness > 220 mg/L as CaCO₃) and have elevated groundwater ammonia concentrations (> 2 mg N/L). Sampling throughout the year revealed consistent finished water hardness levels but variable ammonia concentrations. Ammonia removal varied and depended on how much water had been treated since the last regeneration. High ammonia removal (sometimes > 90%) occurred in the first half of the IX service cycle, while effluent ammonia concentrations increased compared to the influent (sometimes > 200%) towards the end of the IX cycle (total length 50,000–92,000 gallons [190–350 m³]). Ammonia removal efficiency varied among the plants, but the overall trends were similar. Because variable ammonia concentrations may make it difficult to produce a consistent total chlorine residual, they can negatively impact disinfection and water quality in the distribution system. Ammonia concentrations should be considered when designing softening systems to determine regeneration frequency, develop blending strategies, or include an alternative ammonia treatment process before IX softening to produce a more stable and consistent finished water.

1. Introduction

Many groundwater drinking water sources in the United States (U.S.) are very hard (> 180 mg/L as CaCO₃) and contain elevated free ammonia (the sum of ammonia (NH₃) and ammonium ion (NH₄⁺), hereinafter referred to as free NH₃), neither of which is regulated in drinking water by the U.S. Environmental Protection Agency (EPA). Hardness is the sum of dissolved multivalent cations in a water, with calcium (Ca²⁺) and magnesium (Mg²⁺) ions being the primary constituents. Hardness in groundwater is associated with naturally occurring mineralogical processes and depends on the local geology where limestone and sedimentary rock aquifers tend to have greater hardness. Elevated hardness can cause scaling and staining of plumbing materials (e.g., valves, faucets, and fixtures), which are nuisance aesthetic issues, and decrease the efficiency of drinking water delivery processes (e.g., water heaters). Softening treatment techniques include ion (cation) exchange (IX), nanofiltration, and lime softening.

There are no federal regulatory requirements to monitor NH₃ in source water so there is no way to precisely identify how many IX softening plants treat NH₃-containing waters. To get an indication as to the potential prevalence and geographic distribution of this treatment scenario, Fig. 1 shows community water systems (CWSs) in the conterminous U.S. that treat groundwater using IX softening along with groundwater free NH₃ concentrations from monitoring wells (i.e., the NH₃ data are not from the CWSs’ source water). In total, 1,167 CWSs were identified, and 353 of these systems had at least one monitoring well within a 10-mile radius. Of these 353, 155 CWSs were located near a well with free NH₃ > 0.5 mg N/L. Most of these CWSs (> 90%) were small systems serving fewer than 10,000 people. The states that were found to have the highest number of IX softening water treatment plants that might be treating groundwater with elevated NH₃ were Illinois, Minnesota, Iowa, Ohio, Florida, and Louisiana. This analysis illustrates that the treatment scenario investigated in the current study may exist in several areas throughout the conterminous U.S.

Fig. 1.

Community water systems that treat groundwater using ion exchange for softening (“X”) and groundwater free ammonia concentrations in the conterminous U.S.

Sources of free NH₃ in groundwater include agricultural and urban runoff, degradation of naturally occurring organic matter, septic systems, and discharges from municipal wastewater treatment plants and concentrated animal feeding operations. The EPA has set a health advisory for the lifetime intake of free NH₃ through drinking water of 30 mg N/L (USEPA 2018). At concentrations much less than the health advisory level (e.g., 1 mg N/L), free NH₃ may negatively impact some drinking water treatment processes and cause problems in the distribution system. Treatment processes that involve free chlorine (free Cl₂) addition (chlorination) can be negatively impacted because free NH₃ reacts with free Cl₂ to produce chloramines, which are weaker oxidants than free Cl₂. Chlorination often is used in arsenic and manganese removal processes, and the presence of free NH₃ may inhibit treatment efficacy (Chen et al. 2018, Lytle et al. 2007). In the distribution system, the biological oxidation of free NH₃ (nitrification) can lead to disinfectant residual loss, corrosion of distribution system materials, taste and odor complaints, elevated nitrite (NO₂⁻) and nitrate (NO₃⁻) concentrations, and excessive oxygen demand (Bremer et al. 2001, Edwards and Dudi 2004, Odell et al. 1996, Rittmann and Snoeyink 1984, Speitel et al. 2011, Suffet et al. 1996). Of particular concern is NO₂⁻ generation, which is regulated with a maximum contaminant level (MCL) of 1 mg as N/L entering the distribution system as an acute health risk (USEPA 2018). Incomplete nitrification resulting in NO₂⁻ accumulation can occur when insufficient dissolved oxygen is present (Park et al. 2010), and NO₂⁻ accumulation can approach 1 mg N/L in distribution systems experiencing nitrification (Odell et al. 1996). Although federally regulated at the entry point into the distribution system, some states (e.g., Iowa) choose to also enforce the MCL in the distribution system if nitrification is suspected.

Treatment techniques for elevated groundwater free NH₃ include free Cl₂ addition to form chloramines or remove free NH₃ by breakpoint chlorination, biofiltration, and IX. Chloramine formation can be difficult to control, and chloramines can lead to nitrification in the distribution system (Wilczak et al. 1996). Breakpoint chlorination may require an excessive amount of chlorine, which can be costly and generate undesired disinfection byproducts. In biofiltration, nitrification occurs as a controlled process at the treatment plant, producing an effluent with free NH₃ concentrations below 0.1 mg N/L that is more biologically stable (i.e., less likely to promote changes in the microbial community and thus water quality within the distribution system) (de Vet et al. 2011, Lytle et al. 2013, Lytle et al. 2015).

IX processes can remove free NH₃ because free NH₃ is predominantly NH₄⁺ at typical pH values of most drinking waters (pK_a of NH₄⁺/NH₃ is 9.24 at 25°C; Smith et al. 1996). Some zeolite resins, particularly clinoptilolite, are selective towards NH₄⁺. A typical selectivity profile for clinoptilolite is: K⁺ > NH⁴⁺ > Na⁺ > Ca²⁺ > Mg²⁺ (Colella 1996). Nevertheless, other cations, including Ca²⁺ and Mg²⁺, can compete with NH₄⁺ and lower its removal efficiency (Beler-Baykal and Cinar-Engin 2007, Wang et al. 2007).

IX is also commonly used to remove hardness ions (i.e., softening). IX tends to be more attractive to small groundwater systems because it is simpler and has lower capital and operation and maintenance costs than other softening processes, such as lime softening (Logsdon et al. 1990). IX softeners typically use a sulfonated polystyrene resin loaded with sodium (Na⁺). These resins are selective towards cations with a higher valence and smaller hydrated ionic radius with a typical selectivity profile of: Ca²⁺ > Fe²⁺ > Mg²⁺ > K⁺ > NH₄⁺ > Na⁺ (Clifford et al. 1986). In contrast to zeolite resins, sulfonated polystyrene resins generally are only slightly more selective towards NH₄⁺ than Na⁺.

IX resin selectivity determines removal profiles over time when multiple competing cations are present. In general, the most preferred cations are removed consistently and are the last to breakthrough with a sharp breakthrough curve, reaching their influent concentration. Less preferred ions are removed early in the service cycle, concentrated within the bed, and then are released as the more preferred cations out-compete and replace them on IX sites (Clifford 1982). The displacement of less preferred ions results in “peaking” for these ions where their effluent concentration is greater than their influent concentration. Peaking has been observed in drinking water anion exchange systems treating arsenate and NO₃⁻ (Ghurye et al. 1999, Wang et al. 2002). In IX softeners where Ca²⁺ and Mg²⁺ are the preferred species, however, the impact of competition between hardness cations and NH₄⁺ has not been reported.

Variable effluent free NH₃ concentrations and peaking could negatively impact finished water quality by affecting the form and concentration of the desired disinfectant (i.e., chlorination or chloramination) at the treatment plant that may subsequently impact the distribution system (e.g., promoting nitrification). IX softener water plant operators have indicated that they often observed unexplained changing disinfectant levels despite maintaining a consistent free Cl₂ feed rate (personal communication). Despite the broad use of IX softening, its impact on free NH₃ has not been extensively reported, particularly in full-scale drinking water treatment applications. The objective of this study was to investigate IX used to soften hard water and its concurrent impacts on free NH₃ removal, finished free NH₃ levels through an operation cycle, and subsequent chlorination. Four full-scale groundwater treatment plants serving small systems in Illinois were investigated.

2. Materials and methods

2.1. Systems

Four small drinking water systems in Illinois treating groundwater with hardness greater than 300 mg/L as CaCO₃ and free NH₃ greater than 2 mg N/L were identified. The treatment plants were designed to remove iron, manganese and/or arsenic and hardness ions, and the treatment trains all included aeration, filtration, IX softening (with some bypass), and free Cl₂ addition (Table 1). Three of the plants added free Cl₂ or permanganate before the filters to oxidize and remove iron, manganese, and arsenic ahead of the softeners. The plants operated a portion of the day.

Table 1.

Treatment plant descriptions.

	System A	System B	System C	System D
Connections	620	200	1,300	456
Population	1,300	500	3,600	1,280
Number of Wells	3	2	3	3
Plant Capacity	233 gpm 1,270 m3/d	100 gpm 550 m3/d	192 gpm 1,050 m3/d	190 gpm 1,050 m3/d
Treatment Process	Aeration Detention time Filters IX Softeners (n=3)1 Fluoride addition Chlorine addition Polyphosphate addition	Aeration Chlorine addition Detention time Filters IX Softener (n=1) Caustic soda addition Fluoride addition Chlorine addition	Aeration Detention time Chlorine or permanganate addition Microfiltration IX Softeners (n=4) Chlorine addition	Aeration Chlorine addition Filters IX Softeners (n=2) Fluoride addition Polyphosphate addition Chlorine addition
IX Softener Bypass (approx.)	15%	10%	33%	25%
IX Softener Service Cycle	50,000 gal 190 m3	45,000–55,000 gal 170–210 m3	92,000 gal 350 m3	62,000–65,000 gal 235–245 m3
IX Resin Volume per Vessel	42 ft3 1,190 L	65 ft3 1,840 L	113 ft3 3,200 L	84 ft3 2,380 L
Daily Operating Time (hours/day)	10–12	7–10	8–10	8–9

¹“n” refers to the number of IX softening units in the treatment process.

2.2. Sampling

Routine monthly water samples were collected over a 6-month period. Generally, samples were collected from four locations within the treatment each train: (i) raw water (RW), (ii) after filtration/before IX softening (BS), (iii) after IX softening (AS), and (iv) finished water (FW). The AS sample captured the effluent from a single IX softener. The FW sample included the combined IX softener effluent as well as filtered water (i.e., BS sample) that bypassed softening and was collected after chlorination. An AS sample was not collected at System B because a sample tap was not available at this location within the treatment train. Two AS samples were collected at System C. At System D, two BS and two AS samples were collected, one from each IX softener. The volume of water treated by each sampled IX softener since its last regeneration was recorded on the chain of custody form.

In addition to routine samplings, at least one set of AS samples was collected at regular time intervals within an IX softener service cycle (i. e., between brine regenerations) at each system to understand cation breakthrough patterns. The intention was to collect samples every 5,000–10,000 gallons (20–38 m³) of water treated over the course of one cycle; however, complete profiles were not captured at all facilities because samples could only be collected during normal operating hours.

Sampling taps were flushed for at least one minute at a moderate rate prior to collecting water samples. Samples were collected in 250-mL high-density polyethylene (HDPE) bottles and were filled to the top and tightly sealed. Two sample bottles were filled at each location. One sample bottle was preserved with 0.15% nitric acid (HNO₃) for metals analysis, and the other did not contain a preservative. Collected water samples were shipped overnight on ice to the EPA laboratories in Cincinnati, Ohio for analysis. Operators recorded the volume of water that had been treated by the IX softener(s) since the last regeneration when the samples were collected on the chain of custody form that accompanied the samples.

2.3. Water quality analyses

No water quality analyses were performed on-site, which is a limitation because some parameters (e.g., pH and total Cl₂) change over time, and it is preferred to measure them immediately. Upon receipt, water samples were immediately tested for pH, free Cl₂, and total Cl₂. The pH was measured using a benchtop pH/ISE meter (EC40, Hach, Loveland, CO) and combination pH electrode (Model 48600, Hach, Loveland, CO) with temperature corrections. The probe and meter were calibrated daily using pH 7 and 10 standard solutions (Whatman, Hillsboro, OR). Free Cl₂ and total Cl₂ were measured using the DPD method (4500-Cl, APHA 2005), and absorbance was measured using a spectrophotometer (DR/2000, Hach, Loveland, CO).

Total inorganic NH₃ (350.1, USEPA 1983) and NO₂⁻ and NO₃⁻ (353.2, USEPA 1983) were measured via automated colorimetric methods. Total NH₃ includes free NH₃ and NH₃ that is combined in chloramines. All NH₃ in the RW was assumed to be present as free NH₃. Free NH₃ in the FW samples was calculated using total Cl₂, free Cl₂, and total NH₃ data. The combined Cl₂ concentration was calculated by subtracting the free Cl₂ concentration from the total Cl₂ concentration. The amount of NH₃ combined in chloramines was calculated, assuming that all combined Cl₂ was present as monochloramine. Finally, the free NH₃ concentration was calculated by subtracting the calculated combined NH₃ concentration from the measured total NH₃ concentration. Organic nitrogen species were not considered. Alkalinity was measured via potentiometric titration (2320, APHA 2005). Metals and other elements (iron, manganese, calcium, magnesium, and potassium) were measured by inductively coupled plasma atomic emission spectroscopy (200.7, USEPA 1994; Thermo Jarrel Ash, Franklin, MA). Arsenic was measured by inductively coupled plasma mass spectrometry (200.8, USEPA 1994; Agilent, Santa Clara, CA). Hardness was calculated as the sum of the magnesium and calcium concentrations.

2.4. Mapping

Records of CWSs that treat groundwater using IX for softening were obtained from the 3^rd quarter 2020 of EPA’s Safe Drinking Water Information System (SDWIS) Federal Data Warehouse (USEPA 2020). The system addresses were geocoded using the U.S. Census Bureau’s Public AR Current reference database (U.S. Census Bureau 2020). Groundwater data from January 1, 2015 to November 1, 2020 were obtained from the Water Quality Portal (Read et al. 2017). Reported free NH₃ concentrations were averaged if a well was sampled multiple times during that period. If more than one well was located within 10 miles of a PWS, the highest free NH₃ concentration was used to determine whether the CWS might have elevated free NH₃ in its source water. Geospatial calculations and mapping were performed in R using the sf and tidyverse packages (Pebesma 2018, R Core Team 2020, Wickham et al. 2019).

3. Results

3.1. Treatment performance

Table 2 summarizes the RW and FW quality observed at each system during this study. The four systems treated very hard groundwater, with hardness levels usually exceeding 300 mg/L as CaCO₃. They also had high iron levels, with average RW iron concentrations ranging from 2.11–3.84 mg/L, which are much greater than the secondary MCL (SMCL) of 0.3 mg/L. Manganese was less prevalent, and average RW manganese concentrations were below the SMCL of 0.05 mg/L. These groundwaters contained elevated levels of free NH₃, ranging from 1.99–7.37 mg N/L.

Table 2.

Summary of raw and finished water quality [average (minimum − maximum)].

System		n	pH	Alkalinity (mg/L CaCO3)	Hardness (mg/L CaCO3)	Ca (mg/L)	Mg (mg/L)	Fe (mg/L)	Mn (mg/L)	As (μg/L)	K (mg/L)	Total NH3 (mg N/L)1	Free NH3 (mg N/L)2
A	RW3	7	7.97 (7.88–8.28)	465 (444–472)	322 (305–339)	78.5 (73.9–83.5)	30.6 (29.3–32.0)	2.11 (1.58–2.60)	0.044 (0.030–0.051)	0.4 (<0.4–0.7)	17.3 (1.8–110.3)	2.09 (1.99–2.14)	2.09 (1.99–2.14)
	FW4	7	7.82 (7.62–8.06)	467 (460–471)	72 (33–119)	12.9 (7.9–24.1)	9.6 (3.1–16.9)	0.02 (0.01–0.03)	0.011 (0.004–0.019)	<0.35	7.9 (0.6–47.4)	0.46 (0.02–1.10)	0.08 (0.01–0.29)
B	RW	6	7.52 (7.16–8.16)	430 (406–451)	220 (36–341)	49.8 (8.2–78.3)	23.4 (3.7–36.0)	3.15 (0.09–7.61)	0.028 (0.005–0.043)	18.3 (2.0–37.9)	2.2 (0.4–5.9)	5.20 (0.93–7.37)	5.20 (0.93–7.37)
	FW	5	8.13 (7.71–9.27)	436 (407–470)	71 (16–309)	15.9 (3.9–68.5)	7.5 (1.5–33.5)	0.10 (0.01–0.57)	0.010 (0.004–0.040)	3.5 (1.4–9.3)	1.0 (0.4–2.2)	4.45 (0.90–8.76)	4.45 (0.38–8.75)
C	RW	7	7.56 (7.25–7.85)	470 (462–481)	367 (361–380)	87.2 (85.3–91.3)	36.4 (35.6–37.9)	3.84 (3.14–4.86)	0.094 (0.080–0.105)	26.2 (24.1–31.0)	2.1 (1.9–2.5)	6.27 (5.95–6.52)	6.27 (5.95–6.52)
	FW	6	7.45 (7.25–7.60)	447 (439–456)	162 (127–196)	33.3 (27.1–38.8)	19.1 (14.1–24.2)	0.05 (0.02–0.19)	0.044 (0.033–0.055)	8.0 (6.8–9.4)	2.9 (2.1–3.6)	5.22 (3.05–7.45)	5.20 (2.87–7.03)
D	RW	7	7.35 (7.11–7.98)	458 (448–467)	314 (301–325)	73.1 (70.4–76.0)	32.0 (30.5–33.0)	2.80 (2.28–3.42)	0.029 (0.024–0.042)	14.1 (12.4–15.1)	2.5 (2.3–2.8)	2.93 (2.73–3.50)	2.93 (2.73–3.50)
	FW	6	7.59 (7.45–7.76)	446 (438–453)	124 (93–168)	28.0 (21.4–36.3)	13.2 (9.6–18.9)	0.27 (0.08–0.40)	0.012 (0.006–0.019)	4.2 (2.9–5.7)	1.8 (1.1–3.3)	2.13 (0.65–4.02)	1.73 (0.24–4.02)

¹Total NH₃ includes both free NH₃ and NH₃ combined in chloramines.

²All RW NH₃ is assumed to be free NH₃. FW free NH₃ is calculated using free Cl₂ and total Cl₂ data, assuming all combined Cl₂ is present as monochloramine.

³RW: raw water

⁴FW: finished water

All systems achieved good hardness reduction (55–80% on average), with average FW hardness ranging from 71–162 mg/L as CaCO₃ after blending the IX softener effluent with unsoftened bypass stream. Average iron removal was >88% at all systems, although the FW iron concentration sometimes was greater than the non-enforceable SMCL at Systems B and D.

In contrast, total NH₃ removal was inconsistent and varied among and within the four systems. At System C, for example, the FW total NH₃ concentration ranged from 3.05–7.45 mg N/L despite a consistent RW free NH₃ concentration of 6.3 ± 0.21 mg N/L (average ± standard deviation). The range in FW total NH₃ concentrations was even wider at System B where it spanned from <1 mg N/L to 8.76 mg N/L. The observed wide range in FW total NH₃ levels could substantially impact the ability to maintain a consistent disinfectant residual and the water quality in the distribution system.

3.2. Nitrogen speciation through treatment

To better understand the fate of nitrogen through the treatment plants, individual inorganic nitrogen species were measured through the treatment processes (Fig. 2), allowing calculation of total inorganic nitrogen concentrations (sum of total NH₃, NO₂⁻, and NO₃⁻, hereinafter TOTIN). Free NH₃ was the dominant constituent of TOTIN in the RW samples at all four plants, and the greatest change in TOTIN concentration occurred across the IX softeners. System A was the only plant that was observed to consistently achieve total NH₃ removal within the plant during the routine sampling events. At the other three facilities, the FW total NH₃ concentration sometimes exceeded the RW total NH₃ concentration. It is important to note that the IX softeners were at different stages in their service cycle (i.e., they had treated different volumes of water since regeneration) between the sampling dates and sometimes between multiple IX softeners at one system on the same sampling date. This difference likely contributed to the range in observed FW NH₃ concentrations.

Fig. 2.

Nitrogen balances through Systems (A) A, (B) B, (C) C, and (D) D. RW: raw water, BS: before IX softener, AS: after IX softener, FW: finished water. Multiple BS and AS sample locations at one plant are indexed as 1 and 2. * indicates no data are available. Note that the ordinate scale is different for each system.

NH₃ removal occurred through multiple processes at System A. Nitrification appeared to occur before the IX softeners. Between the RW and BS samples, approximately 0.5–1 mg N/L total NH₃ was converted to NO₃⁻. System A was the only plant where evidence of nitrification (i.e., NO₂⁻ and/or NO₃⁻ formation) was observed, and importantly, it also was the only system that did not practice prechlorination, thereby providing a conducive environment for biological activity. The IX softeners were very effective at removing total NH₃. The IX softeners reduced total NH₃ levels during all six routine sampling events by 1–1.5 mg N/L so that the total NH₃ concentration in the AS sample was <0.05 mg N/L. Notably, the resulting total NH₃ concentration after nitrification was 1.34 ± 0.17 mg N/L, which was the lowest BS total NH₃ concentration of all systems, and the lower IX softener influent concentration likely allowed for better total NH₃ removal through the IX softeners. The total NH₃ concentration in the FW was greater than that in the IX softener effluent and had greater variability, which was attributed to blending with unsoftened water. NO₃⁻, a monovalent anion, levels were not impacted by IX softening, as expected.

The total NH₃ trends at Systems B, C, and D were similar to each other and differed substantially from the trend observed at System A. At these three facilities, total NH₃ was the dominant TOTIN species, and the TOTIN speciation did not change within the plant. The total NH₃ concentration generally was consistent between the RW and BS samples but varied widely across the IX softeners and in the FW samples over the routine sampling events.

At System B, the total NH₃ concentration of RW and BS samples were typically similar (Figure 2B). It is unclear why the total NH₃ concentration was greater in the BS sample than the RW sample in June 2012. The total NH₃ concentration in the FW sample fluctuated to be both greater than and less than that in the RW sample, with the difference ranging from −2.87 mg N/L to +1.86 mg N/L. An AS sample, separate from the FW sample, was not collected at System B. Approximately 10% of the plant flow bypassed the IX softener, so blending may have contributed to increased total NH₃ in the FW sample. Nevertheless, the IX softener demonstrated the ability to both remove (up to 77%) and increase (up to 33%) total NH₃.

The variability in IX softened total NH₃ levels was largest at System C. Compared to BS total NH₃, sometimes total NH₃ removal was greater than 95% while other times up to a 200% increase in total NH₃ occurred (Fig. 2C). This variability resulted in AS total NH₃ concentrations that ranged from 0.05–19.3 mg N/L. The total NH₃ concentration often was very different between the two collected AS samples. Sometimes the FW sample appeared to be a blend of the two IX softener effluents, with a total NH₃ concentration somewhere between them. However, the total NH₃ concentration in the FW sample sometimes was greater or less than both AS samples.

At System D, the IX softeners achieved greater than 90% total NH₃ removal at times, but at other times, the total NH₃ concentrations increased more than 100% from the BS total NH₃ (Fig. 2D). The FW sample total NH₃ concentration sometimes was between the concentrations in the two IX softener effluents, helping to dampen IX softener effluent concentrations, but the FW sample total NH₃ concentrations were also, at times, greater or less than either of the AS samples.

During the routine sampling events at three of the four facilities (Systems B, C, and D), the IX softeners both removed and increased total NH₃ such that the total NH₃ concentration in the FW sample was sometimes greater than that in the RW sample. The IX softener at System A appeared to consistently remove total NH₃ during the routine samplings, and importantly it was sampled early in the service cycle and had the lowest average BS sample total NH₃ concentration (1.34 mg N/L vs. 4.90 mg N/L, 6.27 mg N/L, and 2.82 mg N/L at Systems B-D, respectively). Overall, the total NH₃ plant profiles demonstrate that IX softeners can have a strong but inconsistent impact on the total NH₃ concentration.

3.3. Ammonia removal throughout IX softener service cycles

The FW sample total NH₃ concentrations observed during the routine sampling varied widely despite consistent total NH₃ concentrations in the RW samples (Table 2), likely because the FW samples were collected at different stages within the IX softener service cycle (i.e., between regenerations). Fig. 3 shows the total NH₃ concentration ratio between the AS and BS samples as a function of the stage in the service cycle based on (A) the volume of water treated and (B) the equivalents of hardness loaded per unit volume of IX resin. Occasionally, an IX softener was operated beyond its stated treated water volume regeneration setpoint. At all systems, total NH₃ tended to be removed (i.e., C/C₀ < 1) when the normalized hardness loading was less than approximately 0.5 eq/L, which typically occurred during the first half of the service cycle. The initial removal efficiency varied from 40% up to approximately 100%. The total NH₃ removal efficiency tended to decrease as the service cycle continued, with total NH₃ release (i.e., C/C₀ > 1) occurring when the normalized hardness loading exceeded 0.5 eq/L resin. In one instance, the total NH₃ concentration in the AS sample was more than twice as large as the concentration in the BS sample. Competition for exchange sites on the resin likely increased as the service cycle continued, causing the more preferred cations (i.e., Ca²⁺ and Mg²⁺) to displace the previously adsorbed NH₄⁺. Thus, the IX softeners sometimes functioned as a source of NH₄⁺ within the treatment plant. IX softeners are operated and regenerated based on hardness removal goals, whereas if NH₄⁺ removal was the goal, the units would need to be regenerated at approximately 50% of the current service cycle length to prevent NH₄⁺ breakthrough and release. Consequently, brine waste production would approximately double at an increased financial cost.

Fig. 3.

Total ammonia IX softener effluent and influent concentration ratios observed during routine sampling events as a function of the stage in the service cycle based on (A) the relative volume of water treated and (B) the amount of hardness loaded per unit volume of resin.

To better understand the fate of multiple cations as a function of the volume of water treated by IX softening, a series of water samples were collected from the AS sample location throughout a service cycle at each system (Fig. 4). Two profiles were collected at System C on different dates. The profile at System D was split between the two AS sampling locations such that samples were collected from AS_1 during the first half of the profile and from AS_2 during the second half. It was not always possible to collect samples over the entire service cycle due to personnel and operational constraints, so the end of the service cycle was prioritized when possible.

Fig. 4.

IX softener effluent ammonia concentrations at each system over one service cycle based on (A) the relative volume of water treated and (B) the amount of hardness loaded per unit volume of resin.

Fig. 4 shows the AS sample total NH₃ concentration at each system over the course of one service cycle. The RW sample total NH₃ concentration is shown before the service cycle for reference. At each system, the total NH₃ concentration increased at some point in the second half of the service cycle, and it exceeded the RW concentration at Systems A, C, and D. Peaking above the RW total NH₃ concentration was followed by a decrease at Systems A and D. The behavior at System C could not be determined because sampling did not continue to the end of the cycle. The normalized loaded hardness at which peaking was observed differed among the systems, ranging from 0.45–0.7 eq/L resin.

The changes observed over one service cycle help explain the variability shown in Fig. 2, and these profiles indicate that large changes happened over a short period. For instance, the AS sample total NH₃ concentration changed from 6.5 mg N/L to 13.8 mg N/L to 3.9 mg N/L over a period of 3 hours (treated water volume of 10,000 gal [38 m³]) in one of the IX softeners at System D.

Fig. 5 shows the concentration ratio of common cations in the IX softener effluent as compared to the RW samples at each system. At most of the systems, the IX softeners continued to perform well by removing Ca²⁺ and Mg²⁺, the most dominant hardness cations, throughout the service cycle, as expected. However, breakthrough of Mg²⁺ was observed at Systems A and D in the last 80–90% of the service cycle, at which point the normalized loaded hardness were 0.8 eq/L resin and 0.6 eq/L resin, respectively.

Fig. 5.

Concentration ratios of common cations in the IX softener effluent as compared to the raw water through one service cycle at Systems (A) A, (B) B, (C) C in November 2012, (D) C in March 2013, and (E) D. Note that the volume of water treated in one service cycle is system-specific based on plant conditions and hardness removal goals.

Total NH₃ removal ranged from 28–98% in all systems early in the service cycle; then, removal tended to decrease as the cycle progressed. The concentration ratio values and trends were similar to those observed during the routine sampling events (Fig. 3). Release of NH₄⁺ (i.e., C/C₀ > 1) was observed at three of the systems at some point during the last 30% of the service cycle (Fig. 5A, 5D, and 5E). Towards the end of the cycle, Ca²⁺ and Mg²⁺ appeared to out-compete NH⁴⁺ and K⁺ for exchange sites on the IX resin such that Ca²⁺ and Mg²⁺ continued to be removed while NH⁴⁺ and K⁺ were released. Effluent peaking of less preferred ions is a known phenomenon in IX (Clifford et al. 1986), and these data demonstrate that effluent peaking of NH₄⁺ can occur in IX softeners.

3.4. Disinfection chemistry implications

Fluctuating total NH₃ concentrations associated with IX softening operations as illustrated in this work can have important, possibly dramatic, implications on Cl₂ disinfection by complicating the ability to produce a consistent disinfectant residual (type and concentration). As noted previously, operators at the four systems included in this study have observed changing disinfectant levels in the FW despite maintaining a consistent Cl₂ feed rate. Fig. 6 shows free and total Cl₂ concentrations in FW samples collected during the routine sampling events. Although there was the limitation that concentrations were measured when the samples were received at the laboratory (i.e., within 24 hours after collection), they may still reflect early points in the distribution system. Both free Cl₂ and total Cl₂ were detected in samples from System A, indicating that the disinfectant residual could be present as free Cl₂ and/or chloramines, whereas no free Cl₂ was detected in Systems B-D and their disinfectant residual was present exclusively as chloramines. The total Cl₂ concentration varied widely at each system and was ≤ 0.1 mg Cl₂/L at least once at each plant and was non-detect in one sample. For groundwater systems that use chemical disinfection, EPA requires a detectable Cl₂ residual throughout the distribution system, and many states require a residual above a specified limit. Illinois EPA previously required a minimum free Cl₂ residual of 0.2 mg/L or a minimum total Cl₂ residual of 0.5 mg/L throughout the distribution system for all public water systems, and in 2019, increased the minimum free Cl₂ and total Cl₂ residuals to 0.5 mg/L and 1.0 mg/L, respectively (Ill. Admin. Code 2019). Maintaining a greater total Cl₂ residual likely is necessary in chloraminated systems, such as the systems presented here, to achieve the intended effects of a detectable residual (AWWA 2013, Great Lakes-Upper Mississippi River Board of State and Provincial Public Health and Environmental Managers 2012). These systems sometimes had difficulty achieving a total Cl₂ residual in the FW sample, which could negatively impact water quality in the distribution system.

Fig. 6.

Free and total chlorine concentrations in routine finished water (FW) samples from Systems (A) A, (B) B, (C) C, and (D) D. n.d.: non-detect.

System A had the widest range in total Cl₂ residual, ranging from 0.06–4.1 mg Cl₂/L. System A had the lowest and most consistent AS sample total NH₃ (0.02 ± 0.01 mg N/L), but the FW sample total NH₃ ranged from 0.02–1.1 mg N/L (Fig. 2A). The FW sample total NH₃ is influenced by the AS concentration, the bypass stream, and chlorination. The lowest total Cl₂ residuals occurred in the March 2012 and July 2012 samples (Fig. 6A), and they coincided with the lowest FW sample total NH₃ (Fig. 2A). It is possible that the plant was operating very near the breakpoint during these two sampling events, meaning that the applied Cl₂ dose was very close to the amount needed to oxidize the free NH₃ present, so both the total NH₃ and Cl₂ in the FW sample were minimal. This process is discussed in more detail below. For Systems B-D, there was no apparent trend between the total Cl₂ residual and total NH₃ concentration in the FW samples. Other factors, such as temperature and the presence of other oxidant demands like natural organic matter, may have contributed to the observed range in total Cl₂ residuals.

A water treatment plant (e.g., Systems A-D) that feeds a constant free Cl₂ dose but experiences large, temporal free NH₃ concentration fluctuations may experience a variable Cl₂ residual type and concentration. As an illustration, Fig. 7 shows a theoretical breakpoint curve for System A, which has pH 7.8 and a 465 mg/L as CaCO₃ alkalinity, assuming that a constant free Cl₂ dose of 4 mg Cl₂/L is applied while the free NH₃ concentration changes from 0.25–4 mg N/L at 1 hour after free Cl₂ addition (Wahman 2018). For an initial free NH₃ concentration between 0.3 mg N/L to 1 mg N/L, the simulated total Cl₂ residual varies from 0.3 mg Cl₂/L to 4 mg Cl₂/L and switches from being present as free Cl₂ to monochloramine even as the free Cl₂ dose remains constant at 4 mg Cl₂/L. AS free NH₃ concentrations at representative points in the service cycle are indicated on the total Cl₂ line in Fig. 7. The actual initial free NH₃ concentration simulated in the breakpoint curve is unknown because the AS sample does not include bypass water, and the FW sample was collected after chlorination, at which point total NH₃ may have been destroyed. Nevertheless, System A may operate in all regions of the breakpoint curve, including at the breakpoint, over the course of one day. The theoretical breakpoint curve helps explain the variability in total Cl₂ residuals observed in Fig. 6A and illustrates the challenges of maintaining a target Cl₂ residual when the initial free NH₃ concentration fluctuates.

Fig. 7.

Theoretical breakpoint curve 1 hour after free chlorine addition for System A: pH 7.8, 465 mg/L as CaCO₃ alkalinity, and 4 mg Cl₂/L free chlorine dose. Representative points in the service cycle are indicated by X. Ideal plug flow and chemical mixing conditions are assumed.

Similar to System A, the AS sample total NH₃ concentrations in Systems B-D that were measured during both the routine sampling events (Fig. 2) and the service cycle profiles (Fig. 4) span 0.3–1 mg N/L, over which the Cl₂ residual concentration and speciation can change. The FW total NH₃ concentrations tended to be greater in Systems B-D as compared to System A (Fig. 2), but it is possible that they sometimes operated near the breakpoint and at different points along the breakpoint curve.

Operating near the breakpoint is a concern for multiple reasons. First, adequate disinfection might not be achieved. Second, higher concentrations of regulated and unregulated disinfection byproducts may be formed near and past the breakpoint (Schreiber and Mitch 2007, Stefán et al. 2019). Third, the system can switch from operating with a monochloramine residual to a free Cl₂ residual, which creates different oxidation reduction potentials and can cause lead release (Copeland and Lytle 2014, Edwards and Dudi 2004). The preferred operating region for chloramine systems is at a Cl₂:N mass ratio between 4.5:1 and 5:1 (i.e., before the breakpoint) so that monochloramine is the predominant species and the resulting free NH₃ concentration is low to minimize the potential for nitrification (AWWA 2013).

Different concerns arise when the FW free NH₃ concentration is high. In the four systems presented here, IX softeners sometimes functioned to increase total NH₃ within the plant, resulting in FW sample total NH₃ concentrations greater than in the RW sample. As shown in Fig. 7, the free NH₃ concentration remaining in solution after chloramination increases as the initial free NH₃ concentration increases. The FW free NH₃ concentration at Systems B-D often were > 1 mg N/L (Table 2). Free NH₃ serves as a growth substrate for biomass to persist in biofilm on pipe walls or in sediment, leading to recurring nitrification episodes (Liu et al. 2019, Pressman et al. 2012). Nitrification in distribution systems can create a host of problems, including loss of total Cl₂ residual, regrowth of microorganisms, corrosion, and elevated levels of NO₂⁻ (AWWA 2013, Pressman et al. 2012, Wilczak et al. 1996, Zhang et al. 2009).

Thus, operating with a fluctuating free NH₃ concentration could be disruptive to the distribution system if not adequately addressed (e.g., by adjusting the Cl₂ dose), which in practice is very challenging. A better practice would be to operate the plant to achieve a more consistent FW free NH₃ concentration or remove the free NH₃ prior to IX softening, for instance through biological treatment.

4. Conclusions

The current study of four full-scale water treatment plants demonstrates that effluent peaking of total NH₃ can occur from IX softeners, resulting in a wide range of FW total NH₃ concentrations, including concentrations greater than the RW total NH₃. The variable total NH₃ concentration complicates chlorination and sometimes causes very low total Cl₂ residuals. Additionally, it can lead to conditions that may promote increased nitrification in the distribution system. IX softeners are commonly used in groundwater systems in the U.S., but when the groundwater has elevated free NH₃ levels, the IX softener can both remove and subsequently release free NH₃. This change in treatment performance can occur over short time scales on the order of hours, making it difficult to achieve consistent FW quality. Although free NH₃ is not a health concern itself, it impacts disinfectant form and concentration and water quality in the distribution system. Thus, it should be considered in the design and operation of IX softening systems. Management strategies to improve FW quality and increase stability in the distribution system include (beginning with the easiest to implement and progressing towards a more preferred permanent solution): monitoring free NH₃, regenerating the IX softeners more frequently based on the effluent free NH₃ concentration, implementing staggering and blending strategies to maintain a consistent FW free NH₃ concentration, or including an alternative free NH₃ removal process, such as biological treatment, prior to IX softening.