- Email: [email protected]
A new method for calculation of the chlorine demand of natural and treated waters
ARTICLE IN PRESS WAT E R R E S E A R C H
40 (2006) 2877 – 2884
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A new method for calculation of the chlorine demand of natural and treated waters Ben Warton, Anna Heitz, Cynthia Joll, Robert Kagi Centre for Applied Organic Geochemistry, Curtin University of Technology, GPO Box U1987, Perth WA 6845, Australia
art i cle info
A B S T R A C T
Article history:
Conventional methods of calculating chlorine demand are dose dependent, making
Received 4 October 2004
intercomparison of samples difficult, especially in cases where the samples contain
Received in revised form
substantially different concentrations of dissolved organic carbon (DOC), or other chlorine-
11 February 2006
consuming species. Using the method presented here, the values obtained for chlorine
Accepted 20 May 2006
demand are normalised, allowing valid comparison of chlorine demand between samples,
Available online 10 July 2006
independent of the chlorine dose. Since the method is not dose dependent, samples with
Keywords:
substantially differing water quality characteristics can be reliably compared. In our
Chlorine demand
method, we dosed separate aliquots of a water sample with different chlorine concentra-
Water
tions, and periodically measured the residual chlorine concentrations in these subsamples.
Chlorine decay
The chlorine decay data obtained in this way were then fitted to first-order exponential
Chlorine requirement
decay functions, corresponding to short-term demand (0–4 h) and long-term demand (4–168 h). From the derived decay functions, the residual concentrations at a given time within the experimental time window were calculated and plotted against the corresponding initial chlorine concentrations, giving a linear relationship. From this linear function, it was then possible to determine the residual chlorine concentration for any initial concentration (i.e. dose). Thus, using this method, the initial chlorine dose required to give any residual chlorine concentration can be calculated for any time within the experimental time window, from a single set of experimental data. & 2006 Elsevier Ltd. All rights reserved.
1.
Introduction
Microbiological safety of drinking water is generally achieved by treating the water with a chemical disinfectant, usually a powerful oxidising agent such as chlorine, chlorine dioxide or ozone. Chlorine, added to water in the form of hypochlorous acid/hypochlorite ion or as chlorine gas, is still the most widely used and most cost-effective disinfectant worldwide (Hrudey and Hrudey, 2004), despite the well-known problems of disinfection by-product (DBP) formation. Chlorine is a non-selective oxidant, and therefore reacts with both organic and inorganic chemical species in water, as well as functioning as an antimicrobial agent. Generally, in Corresponding author. Tel.: +61 8 9266 3907; fax: +61 8 9266 2300.
E-mail address: [email protected] (B. Warton). 0043-1354/$ - see front matter & 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2006.05.020
natural waters, the species that are most reactive with chlorine are inorganic substances in a reduced valence state such as iron (II), manganese, sulfide, bromide and ammonia (Vasconcelos et al., 1997; Brezonik, 1994). Reactions with these species are relatively fast (Brezonik, 1994), occurring within minutes to hours. Chlorine also reacts with natural organic matter (NOM), and rates of these reactions can vary greatly, depending on the nature of the organic species present (Clark and Sivaganesan 2002; Gang et al., 2002; Brezonik, 1994). The variation in reactivity of chlorine with these different substances leads to complications in predicting the concentration of chlorine remaining after a given contact time in natural and treated waters. In water distribution systems, a
ARTICLE IN PRESS 2878
WAT E R R E S E A R C H
40 (2006) 2877– 2884
number of other factors also contribute to a decrease in the chlorine concentration, including reactions between chlorine and biofilms attached to distribution system pipe walls, consumption of chlorine in corrosion processes and mass transfer of chlorine from bulk water to pipe walls (Vasconcelos et al., 1997). Most of the models reported in the literature to represent chlorine decay in bulk water incorporate either first-order or second-order kinetics. The general first-order kinetic expression for the decrease in the concentration of chlorine in water is expressed as follows: Ct ¼ C0 expðktÞ,
(1)
where Ct is the chlorine concentration at time t, C0 is the initial chlorine concentration, and k is the first-order decay constant. First-order models are dependent only on the concentration of chlorine and do not take into account other species with which the chlorine is reacting. In bulk water, ignoring reactions with pipewall materials, and after initial rapid reactions of chlorine with inorganic species, first-order kinetics has been used to describe chlorine decay in several models (e.g., Gang et al., 2002; Vasconcelos et al., 1997; Rossman et al., 1994). Second-order models for bulk water chlorine decay taking into account both chlorine and a second general reactant referred to as ‘‘chlorine demand’’ have also been developed (e.g. Boccelli et al., 2003). A number of models make use of a sequence of different models (either first or second-order, or with different kinetic parameters) to characterise the different reactions occurring over the period of interest. Sung et al. (2001) divided the decay process into two time intervals: t ¼ 15–90 min and 490 min, and applied a different first-order decay model to each of these intervals. Jadas-He´cart et al. (1992) used t ¼ 4 h as the boundary between the two time intervals, however they used a first-order model for the first interval and a second-order model for the second interval. Clark and Sivaganesan (2002) used second-order models for both time intervals, and used t ¼ 1 h as the boundary. In an even more involved process, the USEPA Water Treatment Plant model (USEPA, 1992) used three models in sequence: zero-order from t ¼ 0–5 min, secondorder from t ¼ 5 min to 5 h and first-order for t45 h. In a model perhaps more representative of true chlorine demand, Haas and Karra (1984) developed a parallel firstorder comprising two kinetic terms, one representing rapid decay and the other representing slower decay, both occurring at the same time. The model uses two separate decay constants (k) and a coefficient, x, representing the fraction of chlorine reacting by each of the two mechanisms: Ct ¼ C0 ½x expðk1 tÞ þ ð1 xÞ expðk2 tÞ.
(2)
Models have also been developed to empirically relate the consumption of chlorine with various water quality parameters. Sohn et al. (2004), based on the earlier work of Amy et al. (1998) developed a two-stage model with a boundary at t ¼ 5 h in which the kinetic variable, k, in each stage can be evaluated using the values for C0, dissolved organic carbon (DOC) concentration and ammonia concentration (mg L1 as N). Conventional procedures to determine chlorine demand have a major disadvantage in that the application of the data
obtained is limited to the parameters under which the test was conducted. The procedure for chlorine demand/requirement described in ‘‘Standard Methods for the Examination of Water and Wastewater’’ (American Public Health Association [APHA], American Water Works Association [AWWA], Water Environment Federation [WEF]) requires the operator to dose aliquots of the sample water at several chlorine concentrations, such that the required residual will be attained at the required time (Clesceri et al., 1998). In this ‘‘APHA Method’’, the chlorine demand is calculated by subtracting the residual chlorine concentration from the concentration of chlorine dosed, after correction for blanks. The chlorine demand obtained in this way must be quoted for a given chlorine dose, contact time, temperature and sample pH. A further disadvantage of this method is that the chlorine demand is dependent on the chlorine dose (Clesceri et al., 1998), and is therefore only applicable at the particular chlorine dose used. Due to the variation in the reactivity of chlorine, depending on the dose, it is not possible to predict the concentration of chlorine that will be consumed under a given set of conditions, by monitoring chlorine decay at a single dose. Therefore, we have developed a new method to predict chlorine consumption, and using this method, the dose required to provide any residual concentration at any contact time (within the experimental timeframe) can be calculated. We have used the term ‘‘chlorine demand’’ throughout this study: this term has been defined as ‘‘the difference between the added oxidant dose and the residual oxidant concentration measured after a prescribed contact time at a given pH and temperature’’ (Clesceri et al., 1998), while the oxidant requirement was defined as ‘‘the oxidant dose required to achieve a given oxidant residual at a prescribed contact time, pH and temperature’’ (Clesceri et al., 1998). Our method can be used to calculate both the chlorine demand and the chlorine requirement. In the present study, we have used the term ‘‘chlorine demand’’ to represent the dose of free chlorine required to give a residual of zero after a given contact time, usually 168 h. It should be noted that the chlorine demand of water measured in the laboratory will differ from the actual chlorine demand in a distribution system, because factors such as reactions of chlorine with biofilms and with pipe walls are not incorporated into laboratory testing procedures. These factors typically increase the chlorine demand as they provide for additional reactants to account for chlorine consumption. The proportion of the chlorine demand in distribution systems due to reactions with pipes varied widely, and was correlated with pipe diameter and type of pipe material (Haas et al., 2002). Several models describing, or incorporating, chlorine demand in a distribution system have been developed elsewhere (e.g. Rossman et al., 1994).
2.
Materials and methods
2.1.
Water samples
Water samples were taken from locations throughout the Wanneroo groundwater treatment plant (GWTP), Perth, Western Australia, and from a separate bore drawing from
ARTICLE IN PRESS WAT E R R E S E A R C H
2879
4 0 (200 6) 287 7 – 288 4
an artesian aquifer near this treatment plant. The treatment process at Wanneroo GWTP involves a combination of aeration, chlorination, MIEX, alum coagulation and sedimentation and filtration, described in detail elsewhere (Allpike et al., 2005). A biofiltration pilot plant also operates at this plant, which receives water from the outlet of the combined MIEXs-alum coagulation process. The major physico-chemical parameters of the water samples used are given in Table 1.
water samples to 5 mg L1 for high-demand samples. The aim was to produce residual concentrations 7 days after dosing that incremented by 0.5–1.0 mg L1.
2.2.
2.4.
Chlorine dosing
Water samples were dosed with sodium hypochlorite solution (1000 mg L1) to give various pre-determined chlorine concentrations. All waters were dosed at four different chlorine concentrations with the lowest selected to produce a residual chlorine concentration close to zero, 7 days after dosing. Incremental increases in the chlorine doses for the remaining three concentrations were selected based on the estimated chlorine demand, and varied from 1 mg L1 for low demand
2.3.
Chlorine measurement
Chlorine measurements were conducted using the DPD colorimetric method (Clesceri et al., 1998), which has an experimental error of 1–2% RSD (Harp, 2002).
Calculation of chlorine demand
Water samples were divided into four subsamples, and each subsample was dosed with a different concentration of chlorine. Chlorine residual concentrations were measured periodically, typically over a 7 days period. Plots of chlorine residual concentration versus time were constructed for two time intervals, 5 min to 4 h and 4–168 h, (e.g. as shown in Figs. 1 and 2, respectively). For each time interval, an exponential first-order decay curve was fitted to each of the four data sets
Table 1 – Major physico-chemical parameters of the water samples used in this study Water sample
Parameter DOC (mg L1)
UV254
pH
Alkalinity (mg L1)
1.89 1.80 4.15 1.90 1.60 2.15
0.005 0.058 0.285 0.056 0.018 0.032
6.53 7.97 7.14 7.29 6.94 6.42
50 95 90 95 70 45
Wanneroo GWTP after biofiltration Artesian bore W257 Raw inlet water after aeration MIEXs outlet MIEXs-C (clarifier) outlet Enhanced coagulation clarifier outlet
Residual chlorine concentration (mg L-1)
10 ∗t
9
Ct =6.83+0.970∗e-0.164
8 7
Ct=5.39+0.494∗e-0.550∗t
6 5
∗t
Ct =3.39+0.688∗e-0.395
4 3
∗t
Ct =1.55+0.602∗e-0.649
2 1 0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Time (hours) Fig. 1 – Residual free chlorine concentration (Ct) versus time for a water sample dosed with chlorine (C0 ¼ 4 (’), 6 (K), 8 (m) and 10 (.) mg L1) for time t ¼ 0–4 h (short-term chlorine demand).
ARTICLE IN PRESS 2880
WAT E R R E S E A R C H
40 (2006) 2877– 2884
Residual chlorine concentration (mg L-1)
8 7 6
∗t
Ct =4.75+2.72∗e-0.0156
5 ∗t
Ct =3.56+1.98∗e-0.0143
4 3
∗t
Ct =1.60+1.98∗e-0.0116 2 1
∗t
Ct =-0.00531+1.61∗e-0.0152
0 0
20
40
60
80
100
120
140
160
180
Time (hours) Fig. 2 – Residual free chlorine concentration (Ct) versus time for a water sample dosed with chlorine (C0 ¼ 4 (’), 6 (K), 8 (m) and 10 (.) mg L1) for time t ¼ 4–168 h (long-term chlorine demand).
corresponding to an initial chlorine concentration, according to the following equation: Ct ¼ a þ b expðdtÞ.
(3)
The parameters a, b and d ( ¼ k from Eq. (1)) were determined using curve-fitting software (OriginLab Origin Version 6.1). The interval t ¼ 5 min to 4 h was used instead of 0–4 h as this gave a better fit with the decay curves. This is presumably the reason that the USEPA Water treatment Plant model (USEPA, 1992) modelled the first 5 min separately from the following several hours. In Eq. (3), the coefficient ‘‘a’’ shifts the curves in the y-direction, and the coefficient ‘‘b’’ expands or contracts the curves in the y-direction. According to Eq. (1), it is assumed that the reactant concentration (in this case chlorine) proceeds towards zero as the reaction progresses, but in the system used here, chlorine is present in excess and the concentration proceeds towards a residual value, represented by the variable ‘‘a’’. The variable ‘‘b’’ is analogous to C0 in Eq. (1), except that in this case, as chlorine is present in excess, ‘‘b’’ represents the amount of the initial chlorine dose that reacts (i.e., the initial dose, C0, minus the residual ‘‘a’’). To confirm that the reduction in chlorine concentration could be represented by an exponential first-order decay function, the natural logarithm of the chlorine concentrations were plotted against time and the linearity of the data assessed. To determine the chlorine demand at any desired time (i.e. Ct), the value of time, t, was substituted into Eq. (3), together with the appropriate values of a, b and d for each of the initial chlorine concentrations used. The initial concentration C0 was then plotted against Ct for each of the initial concentrations, and a linear function was fitted to this data, according to the following equation: C0 ¼ e þ f Ct ,
(4)
where f is the slope of the linear function, and e is the y intercept. It should be noted that the values of e and f are
specific for a given value of time, t. The equation for this linear function was then used to determine the chlorine dose (i.e. C0) required to give a specific residual concentration (chlorine requirement) at the desired time, by substituting the specified residual concentration Ct into Eq. (4).
3.
Results and discussion
To demonstrate the procedure for calculation of chlorine demand, as described in Section 2, the results for a typical water sample are presented. A sample of groundwater from an artesian bore was dosed with chlorine at concentrations of 4, 6, 8 and 10 mg L1 (time t ¼ 0), and the residual chlorine concentrations were measured periodically over a period of 168 h. The residual concentrations for each initial chlorine concentration over the time interval 5 min to 4 h (short-term demand) are shown in Fig. 1, and for 4–168 h (long-term demand) are shown in Fig. 2. To confirm first-order exponential decay in the intervals 5 min to 4 h and 4–168 h, the natural logarithms of the residual chlorine concentrations were plotted against time for the four initial concentrations. In the interval 4–168 h, the linear functions fitted the data adequately, with correlation coefficients of R240.90 (data not shown). However, in the interval 5 min to 4 h it was found that the the correlation coefficients R2 were lower, in the range 0.63–0.79, due to the data deviating from linearity prior to 15 min (30 min in the case of the 4 mg L1 chlorine dose). Data within the interval 15 min to 4 h (30 min to 4 h in the case of the 4 mg L1 chlorine dose) had correlation coefficients R240.90, and for this reason, only this data was used in the present study. For the data in this new interval, and for all of the data in the interval 4–168 h, the first-order exponential decay function provided a good fit, with correlation coefficients of R240.95.
ARTICLE IN PRESS WAT E R R E S E A R C H
4 0 (200 6) 287 7 – 288 4
It was apparent that the exponential decay functions utilised by the Origin software provided a better fit to the data than the linear functions of the natural logarithms. Using the Origin curve fitting software, it was possible to obtain excellent correlations using two data sets (15 min to 4 h and 4–168 h), but for the linearised data it was necessary to divide the data into at least three data sets to obtain similar correlations. All further data manipulations were therefore carried out using the exponential decay functions (as shown in Figs. 1 and 2), rather than via linearization of the raw data. The parameters a, b and d (Eq. (3)) for each of the four initial concentrations in the time interval 15 min to 4 h were calculated, and these are listed in Table 2, with the corresponding parameters for the interval 4–168 h listed in Table 3. To determine the chlorine demand at, for example, t ¼ 144 h, the parameters a, b and d in Table 3 were used. The four appropriate values (for a, b, d and t) were substituted into Eq. (3) for each of the four initial chlorine concentrations C0, and the corresponding four values of Ct were calculated. C0 was plotted against Ct (Fig. 3) for each of the four initial concentrations, and a linear function was fitted to the data (R2 ¼ 0.993). The parameters e and f from the linear function were 3.68 and 1.21, respectively. The initial chlorine dose (C0) that would have been required to give a specific residual chlorine concentration Ct at t ¼ 144 h was then determined by substituting the values e, f and Ct into Eq. (4). Similarly high correlations for the linear functions used to provide the
Table 2 – Values of parameters a, b and d as calculated for Eq. (3), for various initial chlorine concentrations (C0) in the time interval 15 min to 4 h Initial chlorine concentration, C0 (mg L1) 4 6 8 10
a
b
d
1.55 3.39 5.39 6.83
0.602 0.688 0.494 0.970
0.649 0.395 0.550 0.164
Table 3 – Values of parameters a, b and d as calculated for Eq. (3), for various initial chlorine concentrations (C0) in the time interval 4–168 h Initial chlorine concentration, C0 (mg L1) 4 6 8 10
a
b
d
0.00531 1.60 3.56 4.75
1.61 1.98 1.98 2.72
0.0152 0.0116 0.0143 0.0156
2881
parameters e and f were obtained when calculating Ct values in the time interval t ¼ 15 min to 4 h, using the parameters a, b and d given in Table 2. The errors that can occur with use of conventional, dosedependent methods that use only fixed contact time models are apparent when this method is used to determine chlorine demands at various doses. According to the APHA method of Clesceri et al. (1998), chlorine demand is defined as the initial chlorine dose minus the residual concentration, at a fixed contact time. If this method were to be used to determine the chlorine demand in the example discussed above, using an initial dose of 3.68 mg L1 in the linear function given in Fig. 3 would give a chlorine demand at t ¼ 144 h of 3.68 mg L1, with Ct ¼ 0 mg L1. To produce Ct ¼ 1 mg L1, the value for C0 was 4.88 mg L1, giving a higher chlorine demand of 3.88 mg L1, while to produce Ct ¼ 2 mg L1, C0 was 6.09 mg L1, giving an even higher chlorine demand of 4.09 mg L1. The dependence of the calculated chlorine demand value on the initial chlorine dose is apparent. The chlorine demand procedure had previously been carried out on a water sample from the same artesian bore on a separate occasion with approximately half the number of data points collected. On this previous occasion, the correlations of the exponential decay functions for all of the chlorine doses and both time intervals were similar to the present results, and the calculated chlorine demand at t ¼ 144 h was 3.46 mg L1, similar to the 3.68 mg L1 reported here (parameter e). Furthermore, the slope of the linear plot of chlorine dose versus residual concentration was 1.20, similar to the 1.21 reported above (parameter f). It should be noted that the two water samples were not collected at the same time and there is some variation in quality of the bore water with time, which may account for the small difference between the two calculated chlorine demands. In a further demonstration of the dose dependence of the APHA method, chlorine demand values were determined for the above sample, as well as for a sample of treated water from Wanneroo GWTP after biofiltration, using both the APHA method and the new method (Table 4). When calculated using the APHA method, the values for chlorine demand for the biofiltration sample varied by more than 1 mg L1 at a contact time of 48 h and by 0.77 mg L1 at a contact time of 1 h, depending on the chlorine dose. The highest chlorine dose (6 mg L1) was clearly too high for this water sample, but even at the more realistic doses of 2 and 4 mg L1, the values for a contact time of 48 h varied by 0.58 mg L1 (i.e. 25%). In contrast, calculation of chlorine demand using the new method was independent of dose. In the case of the sample of artesian groundwater, differences in chlorine demand at different chlorine doses were also observed, but the differences were not as great as those observed for the biofiltration sample (Table 4). The greatest difference for the artesian groundwater was observed at the longest contact time: 0.26 mg L1 at 168 h contact time (7%). Notably, the DOC concentrations in the two examples in Table 4 were very similar (1.72 and 1.80 mg L1, for the biofiltration and groundwater samples, respectively), but the chlorine demand values were considerably higher for the artesian groundwater sample. This demonstrates that DOC concentration is not necessarily useful for predicting chlorine demand, or more
ARTICLE IN PRESS 2882
WAT E R R E S E A R C H
40 (2006) 2877– 2884
Initial chlorine concentration (mg L-1)
12
10
8 C 0 =3.68+1.21∗Ct 6
4
2
0 0
1
2
3
4
5
6
Residual chlorine concentration (mg L-1) Fig. 3 – Initial free chlorine concentration (C0) versus residual free chlorine concentration Ct for a water sample dosed with chlorine (C0 ¼ 4, 6, 8 and 10 mg L1) at time t ¼ 144 h.
Table 4 – Chlorine demand calculated using both the APHA and the new methods for two waters: (a) Wanneroo GWTP after biofiltration (DOC ¼ 1.89 mg L1), and (b) artesian bore W257 (DOC ¼ 1.80 mg L1) Water sample
a a a a b b b
Contact time (h)
1 3.5 24 48 1 48 168
Chlorine demand calculated using the APHA methoda (mg L1) Dose ¼ 2 mg L1
Dose ¼ 4 mg L1
Dose ¼ 6 mg L1
1.43 1.58 1.83 1.92 — — —
1.68 2.05 2.36 2.50 2.16 3.16 3.88
2.20 2.40 2.85 3.09 2.13 3.18 4.14
Chlorine demand calculated using the new methoda (mg L1) 1.40 1.56 1.91 2.06 2.04 3.01 3.75
a
The chlorine demand is defined as the dose required to exactly satisfy the chlorine requirement for the given contact time (i.e., to provide zero residual after the given contact time).
particularly, for predicting the required chlorine dose rates in chlorine demand determinations. The new procedure presented here takes into account the dose-dependence of chlorine demand determinations, allowing calculation of the chlorine dose required to give any desired residual from a single set of experimental data. The method enables valid intercomparison of chlorine demand values for different water samples, in effect giving a ‘‘normalised’’ chlorine demand, by allowing for a standard residual concentration (e.g. 0 mg L1) to be quoted for all measurements. Even though more determinations may be necessary than when using the APHA method, this is balanced by the increased flexibility offered by the new method. Once the sample has been analysed, chlorine demand can be determined for any dose, contact time or residual within the experimental constraints. According to the APHA method of Clesceri et al. (1998), it is necessary to ‘‘yincrease dosage in increments of 0.1 mg L1 for determining low demands/
requirementsy’’. Using this method, it may therefore be necessary to conduct experiments using a large number of chlorine doses to ensure that a sample contains close to the desired residual chlorine concentration at the desired time, but this is not critical when using the new method. The APHA method also only allows for the determination of chlorine demand (or residual chlorine concentration) for the time specified, whereas the procedure presented here permits these values to be calculated for any time within the experimental time window (up to 168 h in the examples presented), from the data collected. In order to demonstrate the applicability of the new method to a practical problem, we briefly describe its use in comparing the effectiveness of two different potable water treatment processes, a conventional ‘‘enhanced coagulation’’ (EC) process using alum, and a MIEXs process that is operated in conjunction with alum coagulation (MIEXs-C). At the Wanneroo GWTP, the enhanced alum coagulation process
ARTICLE IN PRESS WAT E R R E S E A R C H
2883
4 0 (200 6) 287 7 – 288 4
Table 5 – Chlorine demand, DOC concentration and specific chlorine demand (SCD) for water sampled from the Wanneroo GWTP, involving treatment by the MIEXs-C and enhanced alum coagulation treatment processes, on two occasions—summer and winter (chlorine demand was for a 7 days contact time, to provide a residual chlorine concentration of 0.0 mg L1) Sample Raw inlet water after aeration MIEXs outlet MIEXs-C (clarifier) outlet Enhanced coagulation clarifier outlet
7 days chlorine demand (mg L1)
DOC (mg L1)
SCD
11.9 6.1 3.7 4.8
4.15 1.90 1.60 2.15
2.8770.14 3.2170.16 2.3170.12 2.2370.11
train is operated in parallel with the MIEXs-C train, and the two processes can be compared simultaneously, using identical source waters. Since the present method enables the determination of normalised values for chlorine demand, calculation of the ‘‘specific chlorine demand’’, i.e. direct comparison of the relative contribution of DOC to chlorine demand, is possible. The specific chlorine demand (SCD) is defined as [chlorine demand/DOC]. Since SCD (or chlorine demand) can be determined for any chlorine residual, at any contact time, these variables must be included when quoting this parameter. In the present study, we have calculated, by our new method, chlorine demand for 7 days, to give a theoretical residual of 0.0 mg L1 (Table 5). Note that, since this chlorine demand value equates to the chlorine dose that is required to give a residual of exactly zero after a 7 days contact time, it cannot be determined using the APHA method. SCD can be useful in comparing the efficiency of treatment processes in removing fractions of DOC that cause chlorine decay, even when absolute DOC concentrations differ substantially. SCD increased slightly upon treatment with MIEXs, showing that this process preferentially removed DOC that had a lower reactivity with chlorine when compared with DOC that is typically removed via coagulation (Table 5). Clarification (i.e. alum coagulation) resulted in a decrease in the SCD values in both the MIEXs-C and the enhanced coagulation processes. These observations were not unexpected, since coagulation processes are well-known to remove aromatic and conjugated organic species, which are the functional groups within NOM that are the most reactive with hypochlorous acid (Volk et al., 2000). Conversely, the MIEXs process operates via an anion exchange mechanism, preferentially removing charged species, not all of which may be as reactive with hypochlorous (or hypobromous) acid as the aromatic species targeted by alum (although MIEXs is well known to also effectively remove DBP precursors (Smith et al., 2003)). Normalised values of chlorine demand and SCD, as calculated in the present study, can be used to evaluate and compare the reactivity to chlorine of NOM in natural and treated waters. Even though the DOC concentrations and the absolute values of chlorine demand varied considerably for each of the water samples tested (i.e. DOC concentrations ranged from 1.60 to 4.15 mg L1 through the Wanneroo GWTP, Table 5), the values for chlorine demand and for SCD are directly comparable and are not affected by errors due to
chlorine dose, as would have been the case using the APHA method.
4.
Conclusions
This method of chlorine demand calculation has a number of advantages over the widely used APHA method, and other similar methods:
Conventional methods of calculating chlorine demand are
dose dependent, making intercomparison of samples difficult, especially in cases where the samples contain substantially different concentrations of chlorine-consuming species (e.g. DOC). Using the new method, chlorine demand values are independent of the chlorine dose and samples of vastly different chlorine demand can be reliably compared; To determine chlorine demand, a particular sample does not have to be dosed with an amount of chlorine estimated to give a near-zero chlorine residual at the desired time; Each water sample is characterised by its unique set of decay curves, so the chlorine dose required to produce any given residual concentration can be calculated for any time within the experimental period; Calculations can be performed retrospectively from the data acquired, without the need for repeating analyses if information is required that was not measured in the initial experiment.
Acknowledgements We thank Water Corporation (Western Australia) and the Australian Research Council (ARC SPIRT Grant) for financial support. We also thank Water Corporation and Orica Australia Pty. Ltd., for provision of samples from the Wanneroo GWTP and for their valuable expertise on many aspects of water treatment operations. R E F E R E N C E S
Allpike, B.P., Heitz, A., Joll, C.A., Kagi, R.I., Abbt-Braun, G., Frimmel, F.H., Brinkmann, T., Her, N., Amy, G., 2005. Size
ARTICLE IN PRESS 2884
WAT E R R E S E A R C H
40 (2006) 2877– 2884
exclusion chromatography to characterise DOC removal in drinking water treatment. Environ. Sci. Technol. 39, 2334–2342. Amy, G., Siddiqui, M., Ozekin, K., Zhu, H.W., Wang, C., 1998. Empirically based models for predicting chlorination and ozonation by-products: haloacetic acids, chloral hydrate, and bromate, EPA Report CX 819579. Boccelli, D.L., Tryby, M.E., Uber, J.G., Summers, R.S., 2003. A reactive species model for chlorine decay and THM formation under rechlorination conditions. Water Res. 37, 2654–2666. Brezonik, P.L., 1994. Chemical Kinetics and Process Dynamics in Aquatic Systems. CRC Press Inc., Boca Raton. Clark, R.M., Sivaganesan, M., 2002. Predicting chlorine residuals in drinking water: second-order model. J. Water Resour. Plann. Manage. 128 (2), 152–161. Clesceri, L.S., Greenberg, A.E., Eaton, A.D. (Eds.), 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, Washington, DC. Gang, D.D., Segar Jr., R.L., Clevenger, T.E., Banerji, S.K., 2002. Using chlorine demand to predict TTHM and HAA9 formation. J. Am. Water Works Assoc. 94, 76–86. Haas, C.N., Karra, S.B., 1984. Kinetics of wastewater chlorine demand exertion. J. Water Pollut. Control Fed. 56, 170–173. Haas, C.N., Gupta, M., Chitluru, R., Burlingame, G., 2002. Chlorine demand in disinfecting water mains. J. Am. Water Works Assoc. 94, 97–102. Harp, D.J., 2002. Current Technology of Chlorine Analysis for Water and Wastewater. Hach Company, USA, 34pp.
Hrudey, S.E., Hrudey, E.J., 2004. Safe Drinking Water: Lessons from Recent Outbreaks in Affluent Nations. IWA Publishing, London. Jadas-He´cart, A., El Morer, A., Stitou, M., Bouillot, P., Legube, B., 1992. The chlorine demand of a treated water. Water Res. 26 (8), 1073–1084. Rossman, L.A., Clark, R.M., Grayman, W.M., 1994. Modeling chlorine residuals in drinking-water distribution systems. J. Environ. Eng. 120 (4), 803–820. Smith, P., O’Leary, B., Tattersall, J., Allpike B., 2003. The MIEXs Process: A year of operation. Proceedings of Water Convention, Perth, April 2003, on CD. Sohn, J., Amy, G., Cho, J., Lee, Y., Yoon, Y., 2004. Disinfectant decay and disinfection by-products formation model development: chlorination and ozonation by-products. Water Res. 38, 2461–2478. Sung, W., Levenson, J., Toolan, T., O’Day, D.K., 2001. Chlorine decay kinetics of a reservoir water. J. Am. Water Works Assoc. 93 (10), 101–110. Vasconcelos, J.J., Rossman, L.A., Grayman, W.M., Boulos, P.F., Clark, R.M., 1997. Kinetics of chlorine decay. J. Am. Water Works Assoc. 89, 54–65. Volk, C., Bell, K., Ibrahim, E., Verges, D., Amy, G., LeChevalier, M., 2000. Impact of enhanced and optimised coagulation on removal of organic matter and its biodegradable fraction in drinking water. Water Res. 34, 3247–3257.
40 (2006) 2877 – 2884
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A new method for calculation of the chlorine demand of natural and treated waters Ben Warton, Anna Heitz, Cynthia Joll, Robert Kagi Centre for Applied Organic Geochemistry, Curtin University of Technology, GPO Box U1987, Perth WA 6845, Australia
art i cle info
A B S T R A C T
Article history:
Conventional methods of calculating chlorine demand are dose dependent, making
Received 4 October 2004
intercomparison of samples difficult, especially in cases where the samples contain
Received in revised form
substantially different concentrations of dissolved organic carbon (DOC), or other chlorine-
11 February 2006
consuming species. Using the method presented here, the values obtained for chlorine
Accepted 20 May 2006
demand are normalised, allowing valid comparison of chlorine demand between samples,
Available online 10 July 2006
independent of the chlorine dose. Since the method is not dose dependent, samples with
Keywords:
substantially differing water quality characteristics can be reliably compared. In our
Chlorine demand
method, we dosed separate aliquots of a water sample with different chlorine concentra-
Water
tions, and periodically measured the residual chlorine concentrations in these subsamples.
Chlorine decay
The chlorine decay data obtained in this way were then fitted to first-order exponential
Chlorine requirement
decay functions, corresponding to short-term demand (0–4 h) and long-term demand (4–168 h). From the derived decay functions, the residual concentrations at a given time within the experimental time window were calculated and plotted against the corresponding initial chlorine concentrations, giving a linear relationship. From this linear function, it was then possible to determine the residual chlorine concentration for any initial concentration (i.e. dose). Thus, using this method, the initial chlorine dose required to give any residual chlorine concentration can be calculated for any time within the experimental time window, from a single set of experimental data. & 2006 Elsevier Ltd. All rights reserved.
1.
Introduction
Microbiological safety of drinking water is generally achieved by treating the water with a chemical disinfectant, usually a powerful oxidising agent such as chlorine, chlorine dioxide or ozone. Chlorine, added to water in the form of hypochlorous acid/hypochlorite ion or as chlorine gas, is still the most widely used and most cost-effective disinfectant worldwide (Hrudey and Hrudey, 2004), despite the well-known problems of disinfection by-product (DBP) formation. Chlorine is a non-selective oxidant, and therefore reacts with both organic and inorganic chemical species in water, as well as functioning as an antimicrobial agent. Generally, in Corresponding author. Tel.: +61 8 9266 3907; fax: +61 8 9266 2300.
E-mail address: [email protected] (B. Warton). 0043-1354/$ - see front matter & 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2006.05.020
natural waters, the species that are most reactive with chlorine are inorganic substances in a reduced valence state such as iron (II), manganese, sulfide, bromide and ammonia (Vasconcelos et al., 1997; Brezonik, 1994). Reactions with these species are relatively fast (Brezonik, 1994), occurring within minutes to hours. Chlorine also reacts with natural organic matter (NOM), and rates of these reactions can vary greatly, depending on the nature of the organic species present (Clark and Sivaganesan 2002; Gang et al., 2002; Brezonik, 1994). The variation in reactivity of chlorine with these different substances leads to complications in predicting the concentration of chlorine remaining after a given contact time in natural and treated waters. In water distribution systems, a
ARTICLE IN PRESS 2878
WAT E R R E S E A R C H
40 (2006) 2877– 2884
number of other factors also contribute to a decrease in the chlorine concentration, including reactions between chlorine and biofilms attached to distribution system pipe walls, consumption of chlorine in corrosion processes and mass transfer of chlorine from bulk water to pipe walls (Vasconcelos et al., 1997). Most of the models reported in the literature to represent chlorine decay in bulk water incorporate either first-order or second-order kinetics. The general first-order kinetic expression for the decrease in the concentration of chlorine in water is expressed as follows: Ct ¼ C0 expðktÞ,
(1)
where Ct is the chlorine concentration at time t, C0 is the initial chlorine concentration, and k is the first-order decay constant. First-order models are dependent only on the concentration of chlorine and do not take into account other species with which the chlorine is reacting. In bulk water, ignoring reactions with pipewall materials, and after initial rapid reactions of chlorine with inorganic species, first-order kinetics has been used to describe chlorine decay in several models (e.g., Gang et al., 2002; Vasconcelos et al., 1997; Rossman et al., 1994). Second-order models for bulk water chlorine decay taking into account both chlorine and a second general reactant referred to as ‘‘chlorine demand’’ have also been developed (e.g. Boccelli et al., 2003). A number of models make use of a sequence of different models (either first or second-order, or with different kinetic parameters) to characterise the different reactions occurring over the period of interest. Sung et al. (2001) divided the decay process into two time intervals: t ¼ 15–90 min and 490 min, and applied a different first-order decay model to each of these intervals. Jadas-He´cart et al. (1992) used t ¼ 4 h as the boundary between the two time intervals, however they used a first-order model for the first interval and a second-order model for the second interval. Clark and Sivaganesan (2002) used second-order models for both time intervals, and used t ¼ 1 h as the boundary. In an even more involved process, the USEPA Water Treatment Plant model (USEPA, 1992) used three models in sequence: zero-order from t ¼ 0–5 min, secondorder from t ¼ 5 min to 5 h and first-order for t45 h. In a model perhaps more representative of true chlorine demand, Haas and Karra (1984) developed a parallel firstorder comprising two kinetic terms, one representing rapid decay and the other representing slower decay, both occurring at the same time. The model uses two separate decay constants (k) and a coefficient, x, representing the fraction of chlorine reacting by each of the two mechanisms: Ct ¼ C0 ½x expðk1 tÞ þ ð1 xÞ expðk2 tÞ.
(2)
Models have also been developed to empirically relate the consumption of chlorine with various water quality parameters. Sohn et al. (2004), based on the earlier work of Amy et al. (1998) developed a two-stage model with a boundary at t ¼ 5 h in which the kinetic variable, k, in each stage can be evaluated using the values for C0, dissolved organic carbon (DOC) concentration and ammonia concentration (mg L1 as N). Conventional procedures to determine chlorine demand have a major disadvantage in that the application of the data
obtained is limited to the parameters under which the test was conducted. The procedure for chlorine demand/requirement described in ‘‘Standard Methods for the Examination of Water and Wastewater’’ (American Public Health Association [APHA], American Water Works Association [AWWA], Water Environment Federation [WEF]) requires the operator to dose aliquots of the sample water at several chlorine concentrations, such that the required residual will be attained at the required time (Clesceri et al., 1998). In this ‘‘APHA Method’’, the chlorine demand is calculated by subtracting the residual chlorine concentration from the concentration of chlorine dosed, after correction for blanks. The chlorine demand obtained in this way must be quoted for a given chlorine dose, contact time, temperature and sample pH. A further disadvantage of this method is that the chlorine demand is dependent on the chlorine dose (Clesceri et al., 1998), and is therefore only applicable at the particular chlorine dose used. Due to the variation in the reactivity of chlorine, depending on the dose, it is not possible to predict the concentration of chlorine that will be consumed under a given set of conditions, by monitoring chlorine decay at a single dose. Therefore, we have developed a new method to predict chlorine consumption, and using this method, the dose required to provide any residual concentration at any contact time (within the experimental timeframe) can be calculated. We have used the term ‘‘chlorine demand’’ throughout this study: this term has been defined as ‘‘the difference between the added oxidant dose and the residual oxidant concentration measured after a prescribed contact time at a given pH and temperature’’ (Clesceri et al., 1998), while the oxidant requirement was defined as ‘‘the oxidant dose required to achieve a given oxidant residual at a prescribed contact time, pH and temperature’’ (Clesceri et al., 1998). Our method can be used to calculate both the chlorine demand and the chlorine requirement. In the present study, we have used the term ‘‘chlorine demand’’ to represent the dose of free chlorine required to give a residual of zero after a given contact time, usually 168 h. It should be noted that the chlorine demand of water measured in the laboratory will differ from the actual chlorine demand in a distribution system, because factors such as reactions of chlorine with biofilms and with pipe walls are not incorporated into laboratory testing procedures. These factors typically increase the chlorine demand as they provide for additional reactants to account for chlorine consumption. The proportion of the chlorine demand in distribution systems due to reactions with pipes varied widely, and was correlated with pipe diameter and type of pipe material (Haas et al., 2002). Several models describing, or incorporating, chlorine demand in a distribution system have been developed elsewhere (e.g. Rossman et al., 1994).
2.
Materials and methods
2.1.
Water samples
Water samples were taken from locations throughout the Wanneroo groundwater treatment plant (GWTP), Perth, Western Australia, and from a separate bore drawing from
ARTICLE IN PRESS WAT E R R E S E A R C H
2879
4 0 (200 6) 287 7 – 288 4
an artesian aquifer near this treatment plant. The treatment process at Wanneroo GWTP involves a combination of aeration, chlorination, MIEX, alum coagulation and sedimentation and filtration, described in detail elsewhere (Allpike et al., 2005). A biofiltration pilot plant also operates at this plant, which receives water from the outlet of the combined MIEXs-alum coagulation process. The major physico-chemical parameters of the water samples used are given in Table 1.
water samples to 5 mg L1 for high-demand samples. The aim was to produce residual concentrations 7 days after dosing that incremented by 0.5–1.0 mg L1.
2.2.
2.4.
Chlorine dosing
Water samples were dosed with sodium hypochlorite solution (1000 mg L1) to give various pre-determined chlorine concentrations. All waters were dosed at four different chlorine concentrations with the lowest selected to produce a residual chlorine concentration close to zero, 7 days after dosing. Incremental increases in the chlorine doses for the remaining three concentrations were selected based on the estimated chlorine demand, and varied from 1 mg L1 for low demand
2.3.
Chlorine measurement
Chlorine measurements were conducted using the DPD colorimetric method (Clesceri et al., 1998), which has an experimental error of 1–2% RSD (Harp, 2002).
Calculation of chlorine demand
Water samples were divided into four subsamples, and each subsample was dosed with a different concentration of chlorine. Chlorine residual concentrations were measured periodically, typically over a 7 days period. Plots of chlorine residual concentration versus time were constructed for two time intervals, 5 min to 4 h and 4–168 h, (e.g. as shown in Figs. 1 and 2, respectively). For each time interval, an exponential first-order decay curve was fitted to each of the four data sets
Table 1 – Major physico-chemical parameters of the water samples used in this study Water sample
Parameter DOC (mg L1)
UV254
pH
Alkalinity (mg L1)
1.89 1.80 4.15 1.90 1.60 2.15
0.005 0.058 0.285 0.056 0.018 0.032
6.53 7.97 7.14 7.29 6.94 6.42
50 95 90 95 70 45
Wanneroo GWTP after biofiltration Artesian bore W257 Raw inlet water after aeration MIEXs outlet MIEXs-C (clarifier) outlet Enhanced coagulation clarifier outlet
Residual chlorine concentration (mg L-1)
10 ∗t
9
Ct =6.83+0.970∗e-0.164
8 7
Ct=5.39+0.494∗e-0.550∗t
6 5
∗t
Ct =3.39+0.688∗e-0.395
4 3
∗t
Ct =1.55+0.602∗e-0.649
2 1 0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Time (hours) Fig. 1 – Residual free chlorine concentration (Ct) versus time for a water sample dosed with chlorine (C0 ¼ 4 (’), 6 (K), 8 (m) and 10 (.) mg L1) for time t ¼ 0–4 h (short-term chlorine demand).
ARTICLE IN PRESS 2880
WAT E R R E S E A R C H
40 (2006) 2877– 2884
Residual chlorine concentration (mg L-1)
8 7 6
∗t
Ct =4.75+2.72∗e-0.0156
5 ∗t
Ct =3.56+1.98∗e-0.0143
4 3
∗t
Ct =1.60+1.98∗e-0.0116 2 1
∗t
Ct =-0.00531+1.61∗e-0.0152
0 0
20
40
60
80
100
120
140
160
180
Time (hours) Fig. 2 – Residual free chlorine concentration (Ct) versus time for a water sample dosed with chlorine (C0 ¼ 4 (’), 6 (K), 8 (m) and 10 (.) mg L1) for time t ¼ 4–168 h (long-term chlorine demand).
corresponding to an initial chlorine concentration, according to the following equation: Ct ¼ a þ b expðdtÞ.
(3)
The parameters a, b and d ( ¼ k from Eq. (1)) were determined using curve-fitting software (OriginLab Origin Version 6.1). The interval t ¼ 5 min to 4 h was used instead of 0–4 h as this gave a better fit with the decay curves. This is presumably the reason that the USEPA Water treatment Plant model (USEPA, 1992) modelled the first 5 min separately from the following several hours. In Eq. (3), the coefficient ‘‘a’’ shifts the curves in the y-direction, and the coefficient ‘‘b’’ expands or contracts the curves in the y-direction. According to Eq. (1), it is assumed that the reactant concentration (in this case chlorine) proceeds towards zero as the reaction progresses, but in the system used here, chlorine is present in excess and the concentration proceeds towards a residual value, represented by the variable ‘‘a’’. The variable ‘‘b’’ is analogous to C0 in Eq. (1), except that in this case, as chlorine is present in excess, ‘‘b’’ represents the amount of the initial chlorine dose that reacts (i.e., the initial dose, C0, minus the residual ‘‘a’’). To confirm that the reduction in chlorine concentration could be represented by an exponential first-order decay function, the natural logarithm of the chlorine concentrations were plotted against time and the linearity of the data assessed. To determine the chlorine demand at any desired time (i.e. Ct), the value of time, t, was substituted into Eq. (3), together with the appropriate values of a, b and d for each of the initial chlorine concentrations used. The initial concentration C0 was then plotted against Ct for each of the initial concentrations, and a linear function was fitted to this data, according to the following equation: C0 ¼ e þ f Ct ,
(4)
where f is the slope of the linear function, and e is the y intercept. It should be noted that the values of e and f are
specific for a given value of time, t. The equation for this linear function was then used to determine the chlorine dose (i.e. C0) required to give a specific residual concentration (chlorine requirement) at the desired time, by substituting the specified residual concentration Ct into Eq. (4).
3.
Results and discussion
To demonstrate the procedure for calculation of chlorine demand, as described in Section 2, the results for a typical water sample are presented. A sample of groundwater from an artesian bore was dosed with chlorine at concentrations of 4, 6, 8 and 10 mg L1 (time t ¼ 0), and the residual chlorine concentrations were measured periodically over a period of 168 h. The residual concentrations for each initial chlorine concentration over the time interval 5 min to 4 h (short-term demand) are shown in Fig. 1, and for 4–168 h (long-term demand) are shown in Fig. 2. To confirm first-order exponential decay in the intervals 5 min to 4 h and 4–168 h, the natural logarithms of the residual chlorine concentrations were plotted against time for the four initial concentrations. In the interval 4–168 h, the linear functions fitted the data adequately, with correlation coefficients of R240.90 (data not shown). However, in the interval 5 min to 4 h it was found that the the correlation coefficients R2 were lower, in the range 0.63–0.79, due to the data deviating from linearity prior to 15 min (30 min in the case of the 4 mg L1 chlorine dose). Data within the interval 15 min to 4 h (30 min to 4 h in the case of the 4 mg L1 chlorine dose) had correlation coefficients R240.90, and for this reason, only this data was used in the present study. For the data in this new interval, and for all of the data in the interval 4–168 h, the first-order exponential decay function provided a good fit, with correlation coefficients of R240.95.
ARTICLE IN PRESS WAT E R R E S E A R C H
4 0 (200 6) 287 7 – 288 4
It was apparent that the exponential decay functions utilised by the Origin software provided a better fit to the data than the linear functions of the natural logarithms. Using the Origin curve fitting software, it was possible to obtain excellent correlations using two data sets (15 min to 4 h and 4–168 h), but for the linearised data it was necessary to divide the data into at least three data sets to obtain similar correlations. All further data manipulations were therefore carried out using the exponential decay functions (as shown in Figs. 1 and 2), rather than via linearization of the raw data. The parameters a, b and d (Eq. (3)) for each of the four initial concentrations in the time interval 15 min to 4 h were calculated, and these are listed in Table 2, with the corresponding parameters for the interval 4–168 h listed in Table 3. To determine the chlorine demand at, for example, t ¼ 144 h, the parameters a, b and d in Table 3 were used. The four appropriate values (for a, b, d and t) were substituted into Eq. (3) for each of the four initial chlorine concentrations C0, and the corresponding four values of Ct were calculated. C0 was plotted against Ct (Fig. 3) for each of the four initial concentrations, and a linear function was fitted to the data (R2 ¼ 0.993). The parameters e and f from the linear function were 3.68 and 1.21, respectively. The initial chlorine dose (C0) that would have been required to give a specific residual chlorine concentration Ct at t ¼ 144 h was then determined by substituting the values e, f and Ct into Eq. (4). Similarly high correlations for the linear functions used to provide the
Table 2 – Values of parameters a, b and d as calculated for Eq. (3), for various initial chlorine concentrations (C0) in the time interval 15 min to 4 h Initial chlorine concentration, C0 (mg L1) 4 6 8 10
a
b
d
1.55 3.39 5.39 6.83
0.602 0.688 0.494 0.970
0.649 0.395 0.550 0.164
Table 3 – Values of parameters a, b and d as calculated for Eq. (3), for various initial chlorine concentrations (C0) in the time interval 4–168 h Initial chlorine concentration, C0 (mg L1) 4 6 8 10
a
b
d
0.00531 1.60 3.56 4.75
1.61 1.98 1.98 2.72
0.0152 0.0116 0.0143 0.0156
2881
parameters e and f were obtained when calculating Ct values in the time interval t ¼ 15 min to 4 h, using the parameters a, b and d given in Table 2. The errors that can occur with use of conventional, dosedependent methods that use only fixed contact time models are apparent when this method is used to determine chlorine demands at various doses. According to the APHA method of Clesceri et al. (1998), chlorine demand is defined as the initial chlorine dose minus the residual concentration, at a fixed contact time. If this method were to be used to determine the chlorine demand in the example discussed above, using an initial dose of 3.68 mg L1 in the linear function given in Fig. 3 would give a chlorine demand at t ¼ 144 h of 3.68 mg L1, with Ct ¼ 0 mg L1. To produce Ct ¼ 1 mg L1, the value for C0 was 4.88 mg L1, giving a higher chlorine demand of 3.88 mg L1, while to produce Ct ¼ 2 mg L1, C0 was 6.09 mg L1, giving an even higher chlorine demand of 4.09 mg L1. The dependence of the calculated chlorine demand value on the initial chlorine dose is apparent. The chlorine demand procedure had previously been carried out on a water sample from the same artesian bore on a separate occasion with approximately half the number of data points collected. On this previous occasion, the correlations of the exponential decay functions for all of the chlorine doses and both time intervals were similar to the present results, and the calculated chlorine demand at t ¼ 144 h was 3.46 mg L1, similar to the 3.68 mg L1 reported here (parameter e). Furthermore, the slope of the linear plot of chlorine dose versus residual concentration was 1.20, similar to the 1.21 reported above (parameter f). It should be noted that the two water samples were not collected at the same time and there is some variation in quality of the bore water with time, which may account for the small difference between the two calculated chlorine demands. In a further demonstration of the dose dependence of the APHA method, chlorine demand values were determined for the above sample, as well as for a sample of treated water from Wanneroo GWTP after biofiltration, using both the APHA method and the new method (Table 4). When calculated using the APHA method, the values for chlorine demand for the biofiltration sample varied by more than 1 mg L1 at a contact time of 48 h and by 0.77 mg L1 at a contact time of 1 h, depending on the chlorine dose. The highest chlorine dose (6 mg L1) was clearly too high for this water sample, but even at the more realistic doses of 2 and 4 mg L1, the values for a contact time of 48 h varied by 0.58 mg L1 (i.e. 25%). In contrast, calculation of chlorine demand using the new method was independent of dose. In the case of the sample of artesian groundwater, differences in chlorine demand at different chlorine doses were also observed, but the differences were not as great as those observed for the biofiltration sample (Table 4). The greatest difference for the artesian groundwater was observed at the longest contact time: 0.26 mg L1 at 168 h contact time (7%). Notably, the DOC concentrations in the two examples in Table 4 were very similar (1.72 and 1.80 mg L1, for the biofiltration and groundwater samples, respectively), but the chlorine demand values were considerably higher for the artesian groundwater sample. This demonstrates that DOC concentration is not necessarily useful for predicting chlorine demand, or more
ARTICLE IN PRESS 2882
WAT E R R E S E A R C H
40 (2006) 2877– 2884
Initial chlorine concentration (mg L-1)
12
10
8 C 0 =3.68+1.21∗Ct 6
4
2
0 0
1
2
3
4
5
6
Residual chlorine concentration (mg L-1) Fig. 3 – Initial free chlorine concentration (C0) versus residual free chlorine concentration Ct for a water sample dosed with chlorine (C0 ¼ 4, 6, 8 and 10 mg L1) at time t ¼ 144 h.
Table 4 – Chlorine demand calculated using both the APHA and the new methods for two waters: (a) Wanneroo GWTP after biofiltration (DOC ¼ 1.89 mg L1), and (b) artesian bore W257 (DOC ¼ 1.80 mg L1) Water sample
a a a a b b b
Contact time (h)
1 3.5 24 48 1 48 168
Chlorine demand calculated using the APHA methoda (mg L1) Dose ¼ 2 mg L1
Dose ¼ 4 mg L1
Dose ¼ 6 mg L1
1.43 1.58 1.83 1.92 — — —
1.68 2.05 2.36 2.50 2.16 3.16 3.88
2.20 2.40 2.85 3.09 2.13 3.18 4.14
Chlorine demand calculated using the new methoda (mg L1) 1.40 1.56 1.91 2.06 2.04 3.01 3.75
a
The chlorine demand is defined as the dose required to exactly satisfy the chlorine requirement for the given contact time (i.e., to provide zero residual after the given contact time).
particularly, for predicting the required chlorine dose rates in chlorine demand determinations. The new procedure presented here takes into account the dose-dependence of chlorine demand determinations, allowing calculation of the chlorine dose required to give any desired residual from a single set of experimental data. The method enables valid intercomparison of chlorine demand values for different water samples, in effect giving a ‘‘normalised’’ chlorine demand, by allowing for a standard residual concentration (e.g. 0 mg L1) to be quoted for all measurements. Even though more determinations may be necessary than when using the APHA method, this is balanced by the increased flexibility offered by the new method. Once the sample has been analysed, chlorine demand can be determined for any dose, contact time or residual within the experimental constraints. According to the APHA method of Clesceri et al. (1998), it is necessary to ‘‘yincrease dosage in increments of 0.1 mg L1 for determining low demands/
requirementsy’’. Using this method, it may therefore be necessary to conduct experiments using a large number of chlorine doses to ensure that a sample contains close to the desired residual chlorine concentration at the desired time, but this is not critical when using the new method. The APHA method also only allows for the determination of chlorine demand (or residual chlorine concentration) for the time specified, whereas the procedure presented here permits these values to be calculated for any time within the experimental time window (up to 168 h in the examples presented), from the data collected. In order to demonstrate the applicability of the new method to a practical problem, we briefly describe its use in comparing the effectiveness of two different potable water treatment processes, a conventional ‘‘enhanced coagulation’’ (EC) process using alum, and a MIEXs process that is operated in conjunction with alum coagulation (MIEXs-C). At the Wanneroo GWTP, the enhanced alum coagulation process
ARTICLE IN PRESS WAT E R R E S E A R C H
2883
4 0 (200 6) 287 7 – 288 4
Table 5 – Chlorine demand, DOC concentration and specific chlorine demand (SCD) for water sampled from the Wanneroo GWTP, involving treatment by the MIEXs-C and enhanced alum coagulation treatment processes, on two occasions—summer and winter (chlorine demand was for a 7 days contact time, to provide a residual chlorine concentration of 0.0 mg L1) Sample Raw inlet water after aeration MIEXs outlet MIEXs-C (clarifier) outlet Enhanced coagulation clarifier outlet
7 days chlorine demand (mg L1)
DOC (mg L1)
SCD
11.9 6.1 3.7 4.8
4.15 1.90 1.60 2.15
2.8770.14 3.2170.16 2.3170.12 2.2370.11
train is operated in parallel with the MIEXs-C train, and the two processes can be compared simultaneously, using identical source waters. Since the present method enables the determination of normalised values for chlorine demand, calculation of the ‘‘specific chlorine demand’’, i.e. direct comparison of the relative contribution of DOC to chlorine demand, is possible. The specific chlorine demand (SCD) is defined as [chlorine demand/DOC]. Since SCD (or chlorine demand) can be determined for any chlorine residual, at any contact time, these variables must be included when quoting this parameter. In the present study, we have calculated, by our new method, chlorine demand for 7 days, to give a theoretical residual of 0.0 mg L1 (Table 5). Note that, since this chlorine demand value equates to the chlorine dose that is required to give a residual of exactly zero after a 7 days contact time, it cannot be determined using the APHA method. SCD can be useful in comparing the efficiency of treatment processes in removing fractions of DOC that cause chlorine decay, even when absolute DOC concentrations differ substantially. SCD increased slightly upon treatment with MIEXs, showing that this process preferentially removed DOC that had a lower reactivity with chlorine when compared with DOC that is typically removed via coagulation (Table 5). Clarification (i.e. alum coagulation) resulted in a decrease in the SCD values in both the MIEXs-C and the enhanced coagulation processes. These observations were not unexpected, since coagulation processes are well-known to remove aromatic and conjugated organic species, which are the functional groups within NOM that are the most reactive with hypochlorous acid (Volk et al., 2000). Conversely, the MIEXs process operates via an anion exchange mechanism, preferentially removing charged species, not all of which may be as reactive with hypochlorous (or hypobromous) acid as the aromatic species targeted by alum (although MIEXs is well known to also effectively remove DBP precursors (Smith et al., 2003)). Normalised values of chlorine demand and SCD, as calculated in the present study, can be used to evaluate and compare the reactivity to chlorine of NOM in natural and treated waters. Even though the DOC concentrations and the absolute values of chlorine demand varied considerably for each of the water samples tested (i.e. DOC concentrations ranged from 1.60 to 4.15 mg L1 through the Wanneroo GWTP, Table 5), the values for chlorine demand and for SCD are directly comparable and are not affected by errors due to
chlorine dose, as would have been the case using the APHA method.
4.
Conclusions
This method of chlorine demand calculation has a number of advantages over the widely used APHA method, and other similar methods:
Conventional methods of calculating chlorine demand are
dose dependent, making intercomparison of samples difficult, especially in cases where the samples contain substantially different concentrations of chlorine-consuming species (e.g. DOC). Using the new method, chlorine demand values are independent of the chlorine dose and samples of vastly different chlorine demand can be reliably compared; To determine chlorine demand, a particular sample does not have to be dosed with an amount of chlorine estimated to give a near-zero chlorine residual at the desired time; Each water sample is characterised by its unique set of decay curves, so the chlorine dose required to produce any given residual concentration can be calculated for any time within the experimental period; Calculations can be performed retrospectively from the data acquired, without the need for repeating analyses if information is required that was not measured in the initial experiment.
Acknowledgements We thank Water Corporation (Western Australia) and the Australian Research Council (ARC SPIRT Grant) for financial support. We also thank Water Corporation and Orica Australia Pty. Ltd., for provision of samples from the Wanneroo GWTP and for their valuable expertise on many aspects of water treatment operations. R E F E R E N C E S
Allpike, B.P., Heitz, A., Joll, C.A., Kagi, R.I., Abbt-Braun, G., Frimmel, F.H., Brinkmann, T., Her, N., Amy, G., 2005. Size
ARTICLE IN PRESS 2884
WAT E R R E S E A R C H
40 (2006) 2877– 2884
exclusion chromatography to characterise DOC removal in drinking water treatment. Environ. Sci. Technol. 39, 2334–2342. Amy, G., Siddiqui, M., Ozekin, K., Zhu, H.W., Wang, C., 1998. Empirically based models for predicting chlorination and ozonation by-products: haloacetic acids, chloral hydrate, and bromate, EPA Report CX 819579. Boccelli, D.L., Tryby, M.E., Uber, J.G., Summers, R.S., 2003. A reactive species model for chlorine decay and THM formation under rechlorination conditions. Water Res. 37, 2654–2666. Brezonik, P.L., 1994. Chemical Kinetics and Process Dynamics in Aquatic Systems. CRC Press Inc., Boca Raton. Clark, R.M., Sivaganesan, M., 2002. Predicting chlorine residuals in drinking water: second-order model. J. Water Resour. Plann. Manage. 128 (2), 152–161. Clesceri, L.S., Greenberg, A.E., Eaton, A.D. (Eds.), 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, Washington, DC. Gang, D.D., Segar Jr., R.L., Clevenger, T.E., Banerji, S.K., 2002. Using chlorine demand to predict TTHM and HAA9 formation. J. Am. Water Works Assoc. 94, 76–86. Haas, C.N., Karra, S.B., 1984. Kinetics of wastewater chlorine demand exertion. J. Water Pollut. Control Fed. 56, 170–173. Haas, C.N., Gupta, M., Chitluru, R., Burlingame, G., 2002. Chlorine demand in disinfecting water mains. J. Am. Water Works Assoc. 94, 97–102. Harp, D.J., 2002. Current Technology of Chlorine Analysis for Water and Wastewater. Hach Company, USA, 34pp.
Hrudey, S.E., Hrudey, E.J., 2004. Safe Drinking Water: Lessons from Recent Outbreaks in Affluent Nations. IWA Publishing, London. Jadas-He´cart, A., El Morer, A., Stitou, M., Bouillot, P., Legube, B., 1992. The chlorine demand of a treated water. Water Res. 26 (8), 1073–1084. Rossman, L.A., Clark, R.M., Grayman, W.M., 1994. Modeling chlorine residuals in drinking-water distribution systems. J. Environ. Eng. 120 (4), 803–820. Smith, P., O’Leary, B., Tattersall, J., Allpike B., 2003. The MIEXs Process: A year of operation. Proceedings of Water Convention, Perth, April 2003, on CD. Sohn, J., Amy, G., Cho, J., Lee, Y., Yoon, Y., 2004. Disinfectant decay and disinfection by-products formation model development: chlorination and ozonation by-products. Water Res. 38, 2461–2478. Sung, W., Levenson, J., Toolan, T., O’Day, D.K., 2001. Chlorine decay kinetics of a reservoir water. J. Am. Water Works Assoc. 93 (10), 101–110. Vasconcelos, J.J., Rossman, L.A., Grayman, W.M., Boulos, P.F., Clark, R.M., 1997. Kinetics of chlorine decay. J. Am. Water Works Assoc. 89, 54–65. Volk, C., Bell, K., Ibrahim, E., Verges, D., Amy, G., LeChevalier, M., 2000. Impact of enhanced and optimised coagulation on removal of organic matter and its biodegradable fraction in drinking water. Water Res. 34, 3247–3257.