Bilal
Abada
,
Ariel J.
Atkinson
and
Eric C.
Wert
*
Southern Nevada Water Authority (SNWA), P.O. Box 99954, Las Vegas, NV 89193-9954, USA. E-mail: bilal.abada@snwa.com; ariel.atkinson@snwa.com; eric.wert@snwa.com
First published on 5th April 2024
Climate change and drought can lead to unprecedented changes in surface water temperature requiring utilities to examine their ozone system's disinfection capability while minimizing bromate production. This pilot-scale study investigated temperature (15–30 °C) as a single/isolated variable affecting ozone operating performance (demand, decay rate, exposure (CT)) and the ability to achieve a Cryptosporidium log reduction value (LRV) of 0.5–1.5logs, as defined by the United States Environmental Protection Agency (USEPA). When dosing 3.0 mg L−1 of ozone into a surface water with 2.5 mg L−1 of total organic carbon, an increase in temperature from 15 °C to 30 °C increased ozone demand in the dissolution zone from 1.0 mg L−1 to 1.6 mg L−1 (60%) and ozone decay rate from 0.07 min−1 to 0.27 min−1 (385%). Despite more rapid demand/decay, the required ozone dose to achieve an LRV of 1.5
logs remained at 2.4–2.8 mg L−1 due to the reduction in USEPA's CT requirement at higher temperatures (9.35 mg min L−1 at 15 °C vs. 2.31 mg min L−1 at 30 °C). Bromate formation exceeded the USEPA maximum contaminant level of 10 μg L−1 when ozone was dosed to achieve LRV > 0.5
log at all temperature conditions. Chlorine–ammonium pretreatment (0.5 mg L−1 Cl2, 0.1–0.5 mg L−1 NH4+-N) lowered bromate formation to <5 μg L−1 under ambient (80 μg L−1) and elevated (120 μg L−1) bromide concentrations at all temperatures. These results were applied to evaluate a full-scale ozone system designed to achieve an LRV of 1.5
logs if drought increases temperature from 13 °C to 26 °C. The study systematically examined the role of temperature on ozone system performance, which can assist utilities planning for future drought-driven changes.
Water impactDrought-driven water quality changes warrant evaluation by utilities. This work demonstrates the value of pilot-scale testing to evaluate changing temperature conditions on a full-scale ozone facility (i.e. demand, decay rate, CT) while maintaining regulatory compliance (i.e. Cryptosporidium, bromate). The findings can be applied by utilities to better prepare for future temperature fluctuations related to climate change. |
Regulations governing ozone disinfection credit and compliance are temperature dependent. The United States Environmental Protection Agency (USEPA) regulates Giardia lamblia and viruses under the Surface Water Treatment Rule (SWTR), by treatment technique, requiring log reduction values (LRVs) of 3 and 4logs, respectively.9Cryptosporidium is regulated under the USEPA Long Term 2 Enhanced Surface Water Treatment Rule (LT2ESWTR), with LRV requirements determined by Cryptosporidium raw water occurrence and associated bin classification.10 LRVs for Cryptosporidium (eqn (1)), Giardia (eqn (2)), and viruses (eqn (3)) are calculated as a function of water temperature (in °C) and ozone exposure (i.e., CT in mg min L−1) according to the USEPA LT2ESWTR Guidance Manual.11 The ozone CT requirements decrease as the temperature and efficiency of microbial inactivation increases.12
LRVCryptosporidium = 0.0397 × 1.09757Temp × CT; | (1) |
LRVGiardia = 1.0380 × 1.0741Temp × CT; | (2) |
LRVvirus = 2.1744 × 1.0726Temp × CT; | (3) |
Warmer water temperatures can also accelerate the production of disinfection byproducts (DBPs), which must be balanced with treatment goals.13 Bromate is an ozone DBP regulated under the USEPA Stage 2 Disinfectants/DBP Rule, with a maximum contaminant level (MCL) of 10 μg L−1.14 Increased temperature and/or ozone dosing requirements can enhance the rate of bromate production and bring its concentration above the MCL.13,15 Accordingly, bromate formation and control should be evaluated simultaneously as part of the ozone testing matrix. Ammonium addition, chlorine–ammonium (Cl2–NH4+–N) addition, and pH adjustment, upstream of ozonation have demonstrated efficacy to mitigate bromate.16 The Cl2–NH4+–N pretreatment approach has some advantages compared to ammonium-only and pH adjustment since it does not require the use of corrosive acid/base and achieves up to 94% bromate reduction depending on treatment conditions.16–18 During the process, Cl2–NH4+–N combine to form monochloramine (NH2Cl) ahead of the ozone contactor, which disrupts both ozone and hydroxyl radical (˙OH) pathways toward bromate formation.16,19 The efficacy of Cl2–NH4+–N pretreatment to minimize bromate at elevated temperature has not been well examined in the literature.
Higher water temperatures impact ozone system performance by accelerating ozone demand and decay rate resulting in reduced ozone exposure (∫[O3]dt).12,20,21 Temperature-driven ozone decomposition also accelerates production of hydroxyl radicals (˙OH), which may also contribute toward meeting water quality goals.20,22 However, the overall hydroxyl radical exposure (∫[˙OH]dt) may remain relatively unchanged, as was demonstrated when temperature was varied from 5–35 °C during a bench-scale study using Lake Zurich water.21 This was likely associated with using a consistent organic matter matrix comprised of initiators, promoters, and scavengers.23 As a result, increased temperature can be expected to have greater impact on meeting treatment goals related to ozone exposure versus hydroxyl radical exposure.
Systematic temperature evaluations on ozone process performance are limited in the literature, often being performed at either bench-scale or full-scale and confounded by other water quality and operational factors. Bench-scale studies often involve the addition of a chilled concentrated ozone stock solution (∼2 °C) into a sample at room temperature (20 °C). The scalability of the bench-scale ozone dissolution method, mixing, temperature dynamics, and dilution of the samples are not well evaluated.21 In addition, ozone demand is sometimes defined in bench-scale batch systems as the first 30 seconds of ozone decomposition (i.e., initial ozone demand (IOD)),22 which differs from pilot- and full-scale systems. In pilot- and full-scale systems, ozone demand in the dissolution zone (ODdiss) is defined as the difference between the transferred ozone dose and the initial ozone residual (Cin) at the beginning of the credited disinfection zone, which could involve minutes of contact time depending on hydraulic conditions.22,24 Full-scale studies investigating temperature effects often evaluate data from different seasons. These seasonal changes introduce additional water quality variability (e.g., total organic carbon (TOC) concentration, type of dissolved organic matter, pH, alkalinity) that can also influence ozone demand and decay.20,25 Pilot-scale studies have flexibility similar to bench-scale testing while simulating full-scale conditions in terms of ozone dissolution, design, and operation.
This pilot-scale study investigated temperature as a single water quality variable affecting ozone operation, which addresses experimental limitations associated with bench-scale and full-scale studies. The specific objectives of this study were to examine the effect of temperature (15–30 °C) and ozone dose (1–3 mg L−1) on (i) ozone demand, (ii) ozone decay rate, (iii) operating conditions required to meet a range of Cryptosporidium LRVs from 0.5–1.5logs following USEPA guidance;11 and (iv) bromate formation and control using Cl2–NH4+–N pretreatment with Cl2
:
N mass ratios between 1
:
1 and 5
:
1. Hydroxyl radical exposure was not examined as part of this study due to added complexity of dosing an ˙OH probe compound at pilot-scale and since ˙OH are typically considered insignificant contributors to disinfection compared to ozone.15 The pilot-scale results were applied to a full-scale ozone system design (Southern Nevada Water Authority (SNWA), Las Vegas, NV, USA) to evaluate Cryptosporidium LRV targets under anticipated warm water conditions (∼26 °C) due to prolonged drought conditions within the Colorado River watershed. While Lake Mead modeling has projected additional drought-driven water quality changes (i.e. TOC, pH) that can impact ozone performance, this study systematically evaluates the role of the anticipated temperature changes on ozone performance as a single variable.2,7,8
![]() | ||
Fig. 1 Schematic of pilot-scale testing system: raw water module, temperature adjustment module, and ozonation module. The figure depicts that after raw water module, the flow is split into two parallel trains to enable two temperature-independent tests to be completed simultaneously. Chemical application points are identified for chlorine (Cl2), ammonium (NH4+), bromide (Br−), and calcium thiosulfate (CTS). Online analyzer locations are identified for turbidity, pH, conductivity, chlorine (Cl2), dissolved oxygen (DO2), and dissolved ozone (DO3). Grab sample locations are identified for DO3 and BrO3−. A photo of the pilot-scale water heaters is shown in Fig. S1.† |
Ozone process operation used the online instrumentation identified previously along with dissolved ozone grab sample analysis to determine several operational parameters: transferred ozone dose (DO3,transferred) (eqn (4)), first-order ozone decay rate constant (k*) (eqn (5)), initial dissolved ozone residual (Cin) at the entry point of the disinfection zone (eqn (6)), and ozone demand in the dissolution zone (ODdiss) calculated as the difference between DO3,transferred and Cin. Ozone CT was calculated using grab sample data according to the extended integrated T10 method adapted from the LT2ESWTR Toolbox Guidance Manual as shown in eqn (7).11,27 The online dissolved ozone analyzer data was used for process verification and not included in the ozone CT calculation. LRV calculations were determined according to LT2ESWTR guidance according to eqn (1)–(3). A visual depiction of these parameters is included in Fig. S5.†
![]() | (4) |
![]() | (5) |
Cin = C1 × ek*×HDT1 | (6) |
![]() | (7) |
During the period of testing, the raw water had the following water quality characteristics (i) based on online instrumentation: temperature (14.6–15.4 °C), pH (7.5–7.8), turbidity (0.6–0.9 NTU), and conductivity (850–980 μS cm−1), and (ii) based on grab samples: total alkalinity (140–144 mg L−1 as CaCO3), TOC (2.4–2.6 mg L−1), and bromide (81–82 μg L−1).
![]() | ||
Fig. 2 Pilot-scale data depicting correlations between ODdiss as a function of the transferred ozone doses at (a) 15 °C, (b) 20 °C, (c) 26 °C, and (d) 30 °C. Pilot-scale results illustrating the different Cl2–NH4+–N dosing scenarios are presented in Fig. S5.† |
DOM transformation has been evaluated as the primary source of ozone decomposition for decades and documented through second-order rate constants (kO3) for the ozone reactions with specific functional groups and transformation of DOM from high molecular weight compounds to low molecular weight compounds.38–41 The heterogenous mixture of functional groups within DOM can initiate ozone decomposition through reactions spanning several orders of magnitude with aromatic compounds (kO3 < 0.1 to 109 M−1 s−1), olefins (kO3 = 10–106 M−1 s−1), heterocyclic compounds (kO3 < 0.1–108 M−1 s−1), aliphatic amines (kO3 = 103–108 M−1 s−1) and aliphatic nitrogen-containing compounds (kO3 < 1–106 M−1 s−1).35 These DOM reactivities and their relevance to ODdiss may be better contextualized through micropollutant groupings with a similar range of reaction rate constants.42,43 Micropollutants (and corresponding DOM components) with high second-order rate constants (group I: kO3 > 105 M−1 s−1) can be expected to react rapidly with ozone (>90%) and included in the ODdiss calculation. Whereas micropollutants (and corresponding DOM components) with moderate reactivity (Group IIA: 103 < kO3 < 105 M−1 s−1; Group IIB: 10 < kO3 < 103 M−1 s−1) require greater ozone exposure to achieve similar extents of abatement. As the ozone dose increases, a greater proportion of group I/IIA DOM components are included in the ODdiss calculation. As temperature increases from 15 °C to 30 °C, the reactivity of group IIA DOM components are higher leading to an increased ODdiss.39
As discussed previously and demonstrated in other studies, the ozone demand phase exhibits an enhanced formation of ˙OH similar to advanced oxidation process (AOP) characteristics (∫[˙OH]dt) via reactions with different organic and inorganic water quality constituents.21,22,36,44 As the ODdiss increases with ozone dose, the ∫[˙OH]dt increases during this initial phase, which also contributes to additional ozone decomposition via the hydroxyl radical chain reaction (kO3 ∼ 108–109 M−1 s−1).6,20 While DOM is responsible for initiating ozone decomposition and the corresponding production of ˙OH, some DOM components along with carbonate may also serve as inhibitors, which reduces ˙OH availability and terminates the chain reaction responsible for ozone decomposition.34 For a complete ozone depletion in Lake Zurich water, the overall ∫[˙OH]dt was unchanged across a temperature range of 5–35 °C, as demonstrated in a previous study.21 These results further demonstrate a finite number of types of reactive sites in DOM (i.e., initiators, promotors, and inhibitors) leading to an increase in Rct (∫[˙OH]dt/∫[O3]dt) as ∫[˙OH]dt is unchanged and ∫[O3]dt decreases with increased temperature.
The chloramine residual concentration following each of the three Cl2–NH4+–N dosing scenarios was 0.45–0.56 mg L−1 (as total chlorine) in the ozone influent (Table S2†). As a result, the three Cl2–NH4+–N dosing scenarios are combined in all figures for improved clarity and differentiated in the ESI.† ODdiss was not impacted by Cl2–NH4+–N pretreatment at all temperature conditions of 15, 20, 26, and 30 °C. Chloramine (kO3 = 26 M−1 s−1) and bromamine (kO3 = 40 M−1 s−1), and ammonium (kO3 = no reaction) react slowly with ozone (Group IIB) and are not expected to result in additional initial phase ozone demand.45–48 However, chloramine is a weak ˙OH scavenger (k˙OH = 5.2–5.7 × 108 M−1 s−1)49,50 and can partially disrupt the chain reaction of ozone decomposition by minimizing the available ˙OH. Since chloramine is only a weak ˙OH scavenger, reactive sites within the NOM likely outcompete NH2Cl for ˙OH which minimizes the effect of chloramine during the initial phase (ODdiss). In wastewater, minimal effect on ODdiss was also observed when using preformed chloramine, while ODdiss increased as a function of ozone dose.51 If longer chlorine contact times (>10 min) are applied before of NH4+ addition, a further decrease in ODdiss may be expected due to chlorine oxidation of DOM components (groups I and II) that contribute to ozone decomposition.17,18
![]() | ||
Fig. 3 Pilot-scale data depicting the correlations between the first-order ozone decay rate constants as a function of the transferred ozone dose at (a) 15 °C, (b) 20 °C, (c) 26 °C, and (d) 30 °C. Pilot-scale results illustrating the different Cl2–NH4+–N dosing scenarios are presented in Fig. S6.† Another representation of the data in terms of the effect of transferred ozone doses on ozone half-life for all tested temperatures is shown in Fig. S7.† |
A temperature increase from 15 °C to 30 °C accelerated ozone depletion which is demonstrated by higher first-order rate constants (k*). At a dose of 1 mg L−1, k* increased from 0.19 min−1 at 15 °C to 0.31 min−1 at 30 °C. The increase in k* indicates that the reactivities of the group IIA and IIB DOM components were enhanced by the temperature increase leading to a faster ozone depletion in the second phase. The role of temperature on ozone depletion kinetics has been examined for decades in model and natural waters.20,21,52,53 The Arrhenius plot (eqn (8)) allows for the determination of the activation energies (Ea) for these processes.12,21
![]() | (8) |
When applying Cl2–NH4+–N pretreatment, the k* for a transferred ozone dose of 1.5 mg L−1 decreased from 0.21 to 0.15 at 15 °C and from 0.48 to 0.35 at 30 °C. The chloramine residual concentration following the three Cl2–NH4+–N dosing scenarios was 0.45–0.56 mg L−1 as total chlorine in the ozone influent and 0.07–0.24 mg L−1 in the ozone contactor effluent (Table S2†). The decrease in chloramine residual across the ozone process can be attributed to decomposition from the water matrix (i.e. temperature, TOC) in addition to the reaction with ozone, which was described earlier (group IIB). If fully attributed to the reaction with ozone, the measured chlorine decrease (0.2–0.3 mg L−1) can result in up to 0.22 mg L−1 of ozone consumption. In addition to the direct reaction with ozone, scavenging of ˙OH by chloramine also inhibits the chain reaction of ozone decomposition resulting in a corresponding decrease in k*.
![]() | ||
Fig. 4 Pilot-scale data depicting correlations between ozone CT (and corresponding Cryptosporidium LRV credit on the right y-axis) and the transferred ozone dose at (a) 15 °C, (b) 20 °C, (c) 26 °C, and (d) 30 °C. Fitting equations are associated with the left y-axis (ozone CT). Pilot-scale results illustrating the different Cl2–NH4+–N dosing scenarios and demonstrating ODCT calculations are presented in Fig. S9.† |
In Fig. 4, the linear regression lines intercept at the x-axis represent another potential method to determine ozone demand based on CT calculations (ODCT). Based on this interpretation, a single value is calculated for ODCT representing the transferred ozone dose that must be exceeded to create measurable ozone CT. The single value of ODCT contrasts with ODdiss, which varies based on the transferred ozone dose and Cin. In this study, ODCT was determined to be 1.03 mg L−1 at 15 °C or 1.49 mg L−1 at 30 °C. When comparing ODCT to ODdiss in Fig. 2, ODdiss appears to reach a maximum value around the ODCT. Additional research is needed to determine whether ODCT could be considered a maximum value for ODdiss. However, the ODCT approach should be viewed with caution as second phase NOM components (groups IIA and IIB) are being used to predict initial phase ODdiss, which is largely based on group I NOM components.
Across all temperature conditions, Cl2–NH4+–N pretreatment provided a slight increase in ozone CT at lower ozone dose conditions as expected from k* results (Fig. 3). For greater ozone dosages (2.5–3.0 mg L−1), there was little difference in the ozone CT achieved with or without Cl2–NH4+–N pretreatment. The linear regression line intercept at the x-axis for Cl2–NH4+–N pretreatment shows an approximate 35–45% decrease in the ODCT compared to the absence of Cl2–NH4+–N. Since Cl2–NH4+–N pretreatment had minimal impact on ODdiss (Fig. 2), these results show that ODCT is not a good indicator of the maximum ODdiss in the presence of NH2Cl.
Chloramine acts as a ˙OH scavenger and thereby reduces the oxidation of bromide and HOBr/OBr leading to bromate formation.16,18 In addition, it has been demonstrated that bromine radicals which are formed from the reaction of ˙OH with bromide can also be quenched by chloramine.19 Bromate results demonstrated the efficacy of Cl2–NH4+–N (i.e., NH2Cl) pretreatment as a bromate control strategy by maintaining concentrations below 10 μg L−1 for all ozone dose and temperature conditions. Increasing the NH4+–N from 0.1 to 0.5 mg L−1 results in some excess NH4+–N (i.e., lower Cl2:
N mass ratio) which partially quenches HOBr to form bromamine and further minimizes bromate production.16 However, the results showed that the excess NH4+–N (mass basis) provided only a marginal improvement (∼1–2 μg L−1) for bromate control (Fig. 5). At higher temperatures, Cl2–NH4+–N (i.e. NH2Cl) pretreatment provided more effective bromate control by scavenging ˙OH and the formed Br as RCT is expected to increase with temperature. During experiments in which bromide concentrations were increased from 80 μg L−1 (ambient) to 120 μg L−1 (spiked), Cl2–NH4+–N (i.e., NH2Cl) pretreatment again provided effective bromate control with concentrations below the MCL, magenta symbols in Fig. 5.
Both treatment facilities operate as direct filtration plants (i.e., coagulation + flocculation + granular media filtration) with pre-ozonation targeting a non-regulatory, internal goal of Cryptosporidium LRV of 0.5log. Currently, both plants are classified as Bin 1 for Cryptosporidium (<0.075 oocysts per L) under the LT2ESWTR, which requires no additional treatment beyond the credits received for direct filtration, assuming turbidity targets are achieved.11 However, if drought conditions increase Cryptosporidium concentrations in the raw surface water (e.g., ≥0.075 and <1.0 oocysts per L), the treatment facilities may be reclassified to Bin 2, which requires Cryptosporidium LRV of 1.5
logs beyond the baseline direct filtration credit. The impacts of potential reclassification and temperature were considered as SNWA proceeds with oxygen–ozone system refurbishment after 20 years of service.
As part of the design review, the existing oxygen–ozone systems were evaluated for their ability to achieve a Cryptosporidium LRV of 0.5–1.5logs under the projected temperature conditions. Historical full-scale data (13–15 °C) were supplemented with previous pilot-scale data sets with limited information at temperatures greater than 15 °C and used to develop the initial full-scale design criteria shown in Fig. 6. Given the cost and scale of the ozone system refurbishment, additional pilot-scale data were requested to better inform the full-scale design over the projected temperature range associated with drought conditions (i.e., up to 26 °C), hence the basis for the current study.
Full-scale ozone capacity is designed based on the transferred ozone dose required to achieve temperature-dependent CT values according to the LT2ESWTR (Fig. S11†). For example, an LRV of 1.5logs requires a CT of 9.35 mg min L−1 at 15 °C and 2.31 mg min L−1 at 30 °C. By defining the target CT required, pilot-scale data can be utilized to calculate the required transferred ozone dose and verify k*, and ODdiss (Fig. 6). The results from this pilot-scale study confirmed that the design criteria were conservative with respect to the transferred ozone dose required (Fig. 6a and b). Results for k* (Fig. 6c and d) and ODdiss (Fig. 6e and f) showed these parameters are not constant as originally defined, based on the limited historical data; however, the differences were not significant enough to change the design ozone production capacity of 13
800 lb O3 per day (∼6260 kg day−1) for the refurbished full-scale system. The pilot results from this work provided the consultant and utility with greater confidence in the design assumptions.
• As temperature increased from 15 °C to 30 °C, ozone demand in the dissolution zone (ODdiss) increased from 0.3 to 0.7 mg L−1 (at 1.0 mg L−1 transferred ozone dose) and from 1.0 to 1.6 mg L−1 (at 3.0 mg L−1 transferred ozone dose), respectively.
• As temperature increased from 15 °C to 30 °C, the pseudo first-order ozone decay rate constant k* increased (e.g., from 0.19 min−1 to 0.31 min−1 for a transferred ozone dose of 1.0 mg L−1). The activation energies (Ea) derived from pseudo first-order rate constants for ozone decomposition across the different temperature conditions varied between 44–78 kJ mol−1, which aligned with other natural waters.
• Temperature-driven changes in ODdiss and k* can be explained by an enhanced ozone reaction with DOM sites spanning about 10 orders of magnitude (kO3 < 0.1 to 109 M−1 s−1). Fast reacting NOM moieties (group I; kO3 > 105 M−1 s−1) achieve near-complete oxidation (>90%) within the dissolution zone (ODdiss). DOM moieties with lower ozone reactivities (groups IIA/IIB; 10 < kO3 < 105 M−1 s−1) contribute to the second phase of ozone consumption (k*).
• For the waters tested in this study and across the range of water temperatures tested (15–30 °C), decreasing CT requirements (i.e., more rapid disinfection kinetics) offset the increases in ODdiss and k*, resulting in similar ozone dosing requirements to meet Cryptosporidium LRV goals according to the USEPA LT2ESWTR.
• At all temperature conditions (15–30 °C), bromate formation exceeded the MCL of 10 μg L−1 for target Cryptosporidium LRVs > 0.5log. Chlorine–ammonium (0.5 mg L−1 as Cl2 and 0.1–0.5 mg L−1 as NH4+–N, with Cl2
:
N molar ratios of 1
:
1 to 1
:
5) application upstream of ozone effectively minimized bromate to below the MCL at all evaluated ozone doses (1–3 mg L−1).
• While this study focused on temperature variability within a consistent water matrix, drought conditions can change other water quality parameters (i.e. pH, alkalinity, bromide concentration, TOC concentration) as well that can influence ozone demand, decay rate, and CT performance along with bromate formation and mitigation. Furthermore, the potential for greater wastewater influence can also change DOM composition and ozone reaction kinetics depending on the mixture of functional groups (groups I–V). Finally, there is a need for future studies to develop a comprehensive and systematic comparison between findings from bench-, pilot-, and full-scale studies and summarize the key differences and gaps between each system.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ew00042k |
This journal is © The Royal Society of Chemistry 2024 |