Professional Documents
Culture Documents
Ammar 2014
Ammar 2014
Desalination
journal homepage: www.elsevier.com/locate/desal
H I G H L I G H T S
• We propose a novel empirical mathematical model for chlorine dioxide decay prediction.
• Model covers decay at a wide range of both temperature and ClO2 concentrations.
• Model provides an unprecedented opportunity to properly design disinfection systems.
• Model verifications indicate good agreement between calculated and measured values.
a r t i c l e i n f o a b s t r a c t
Article history: Chlorine dioxide is one of the most promising disinfectants though usually used as a secondary disinfectant. Several
Received 2 July 2014 studies addressed the decay rate of chlorine while no study so far tackled the chlorine dioxide decay. This work pro-
Received in revised form 8 August 2014 vides a novel mathematical equation for chlorine dioxide decay prediction in desalinated water.
Accepted 11 August 2014
The study was performed at five different chlorine dioxide dosages; 1.4, 1.2, 1.0, 0.8 and 0.6 mg/L (to cover the
Available online xxxx
most frequently used dosages within both production facilities “as primary disinfectant” and/or the distribution
Keywords:
system “as secondary disinfectant”). Each dosage had been tested at four different temperatures; 20°, 27°, 35°
Desalinated drinking water and 45 °C.
Bulk decay To confirm the validity of the proposed decay rate model/equation, site verification was performed (real concen-
Chlorine dioxide tration vs. predicted concentration) and then t-test formula was used to indicate the similarity of both test
Distribution system results.
As an overall conclusion, the study’s proposed a novel model/equation that shows reasonable levels of robust-
ness. Furthermore, it covers the chlorine dioxide decay at a wide range of temperature profiles as well as a
wide range of chlorine dioxide initial concentration dosages.
The proposed model provides an unprecedented opportunity to more realistically plan and model distribution
systems to achieve disinfection; among other water quality; goals. Network modelling software such as
EPANET provides the platform where the proposed chlorine dioxide decay model can be implemented. Moreover,
the model can aid in the decision variables (design, placement, number of stations and operation control & oper-
ation optimization) for disinfection boosters in water distribution systems.
© 2014 Published by Elsevier B.V.
http://dx.doi.org/10.1016/j.desal.2014.08.010
0011-9164/© 2014 Published by Elsevier B.V.
46 T.A. Ammar et al. / Desalination 352 (2014) 45–51
dioxide reduction. The distribution of chlorite and chlorate is influenced solution was stored in a refrigerator at 4 °C (±0.5 °C). For standardiza-
by pH and sunlight. Approximately 50 to 70% of the chlorine dioxide tion, the Iodometric Method (4500 ClO2-B) was used.
consumed by oxidation reactions is converted to chlorite under condi-
tions typical in water treatment [6,7]. Ultraviolet light and even fluores- 2.3. Raw water source [21]
cent lights can lead to photolysis [8,9].
Numerous inorganic and biological materials found in raw water will Given that the raw water should contain zero levels of chlorine
react with chlorine dioxide [10]. The concentration and types of various dioxide but also should represent water in the distribution system,
chemical and biological constituents that exert a chlorine dioxide de- that being, re-mineralized desalinated water, before the addition of
mand, as well as the temperature variance, will both impact the decay any disinfectant; was selected.
coefficient. The decay coefficient for specific bulk water can be deter-
mined using bottle decay tests. The coefficient for pipe wall decay how- 2.4. Chlorine dioxide concentration and temperature testing profile
ever, must be determined in the field or with pipe segments taken from
the distribution system piping. Overall, the relative importance of decay Bulk decay may be isolated from wall decay by carrying out
at the pipe wall increases as the pipe diameter decreases because the chlorine decay experiments on the source water under controlled
ratio of water volume to pipe surface area decreases [11–20]. conditions in the laboratory. In order to investigate the effects of
Based on the order of the reaction, the mathematical form for alter- chlorine dioxide concentration and water temperature on the bulk
native decay models can be proposed as [11–20]: decay, experiments were carried out at five different chlorine diox-
ide concentrations; 1.4, 1.2, 1.0, 0.8 and 0.6 mg/L. Each concentration
Zero order reaction : ½At ¼ −kt þ ½A0 was tested at four different temperatures; 20°, 27°, 35° and 45 °C.
Those specific temperatures were maintained by incubating the
1.0 l amber glass bottles. The five different chlorine dioxide concen-
trations were selected to cover the most frequently used dosages
First order reaction : ½At ¼ ½A0 expð−ktÞ:
within both production facilities “as primary disinfectant” and/or
the distribution system “as secondary disinfectant” while the four
Where: temperature regimes were selected in order to ascertain the relationship
between the bulk decay coefficient and water's temperature so that
[A]t Concentration of disinfectant in mg/L at time t appropriate values for the winter and summer condition simulations
[A]0 Initial concentration of disinfectant in mg/L at t = 0 are derived.
t Time in days The chlorine dioxide concentrations were measured on duplicate
k Decay coefficient day−1 sub-samples using ChlorodioXense® instrument with disposable am-
n Order of reaction. perometric sensors (the US EPA's approved instruments). The accuracy
of the instrument was checked regularly against the standards set pro-
Given that several studies addressed the decay rate of chlorine while vided by the manufacturer and were always found to be within the ac-
no study so far tackled the chlorine dioxide decay, the objective of this ceptable tolerance level.
work was to determine an empirical predictive novel model for chlorine
dioxide bulk decay prediction in desalinated water over both a wide 2.5. Experiment
range of temperatures and a wide range of chlorine dioxide
concentrations. One bulk raw water sample (50 L) was collected and the following
procedure was performed at the five different temperatures:
2. Materials and methods Five blank samples were prepared and then dosed with the stan-
dardized chlorine dioxide (prepared from the on-site generator) to at-
2.1. Sampling bottle selection, preparation and cleaning tain 1.4, 1.2, 1.0, 0.8 and 0.6 mg/L in the five bottles respectively. The
chlorine dioxide concentration promptly measured at each bottle and
1.0 l amber glass bottles were used. As laboratory glassware can the time was registered as the zero time.
potentially exhibit a demand for chlorine dioxide which could bias the
analytical results and lead to poor accuracy and/or precision, all bottles 2.6. Bulk decay coefficient estimation & model verification
were thoroughly cleaned with soap and water and rinsed several times
with reagent water. Vials (without caps) were dried in an oven at 140 °C. For each temperature (20°, 27°, 35° and 45 °C), five graphs were
plotted illustrating the chlorine dioxide concentration (mg/L) vari-
2.2. Standard chlorine dioxide solution [21] ance “until non-detectable levels” vs. time (hrs) at each dosage.
The natural logarithmic for chlorine dioxide concentrations was
Since chlorine dioxide is a relatively unstable gas, it cannot be com- calculated to estimate the bulk decay coefficient (Kb). Several types
pressed and liquefied, and must, therefore, be generated “on-site” and of trend line were examined to choose the type of trend line most
dissolved in water. appropriate to the results. The best fit trend line type was chosen
Today's technology utilizes chlorite and hydrochloric acid reac- based on the highest R2 (Pearson Coefficient of Determination).
tions to produce chlorine dioxide through a system called: “Chlorine Then, model's precision was verified vs. samples collected (within
dioxide under water generating system” (manufactured by ISIA®).
As this special generator works “under water”, the formation of
chlorine dioxide takes place in water only, limiting the possibility Table 1
Site parameter results for the raw water.
for chlorine dioxide to be released from the water. For that reason
it is often times considered 100% safe fulfilling the most diversified Parameter Unit Result
requirements in terms of usage, safety, reliability, yield, purity of Temperature °C 39
ClO2 solution and ease of conduction. This system was used to get a pH – 7.9
chlorine dioxide solution. Electrical conductivity (EC) μS/cm 164
The generated chlorine dioxide solution was used for preparing the Turbidity NTU 0.1
Chlorine dioxide mg/L ND
temporary chlorine dioxide standards to the desired strengths and the
T.A. Ammar et al. / Desalination 352 (2014) 45–51 47
Table 2
Chlorine dioxide concentration vs. time at 20 °C for all dosages: 1.4, 1.2, 1.0, 0.8 and
0.6 mg/L along with the calculated bulk decay coefficients at each dosage.
Fig. 3. Ln chlorine dioxide concentration vs. time at 20 °C. Chlorine dioxide initial dosage
was 1.2 and 1.0 mg/L. Fig. 4. Illustrates ClO2 concentration vs. time for the five different dosages at 27 °C.
48 T.A. Ammar et al. / Desalination 352 (2014) 45–51
Fig. 7. Illustrates ClO2 concentration vs. time for the five different dosages at 35 °C.
Fig. 5. Ln chlorine dioxide concentration vs. time at 27 °C. Chlorine dioxide initial dosage
was 1.2, 0.8 and 0.6 mg/L.
3.4.2. Estimation of bulk decay coefficient (Kb) at 35 °C
Figs. 8 and 9 illustrate the Ln(x) vs. time plots as well as both trend
lines and R2.
The ln results in all dosages show first order trend lines with R2 rang-
ing between 0.9949 and 0.9421. Hence, at least 94.2% of the results fit
the first order trend line.
Using the trend lines (representing each dosage), the bulk decay co-
efficient (Kb) was calculated as well as the decay rate (day−1). To have
one unique bulk decay coefficient (Kb) representing all the dosages at
35 °C, the average for all calculated coefficients was considered.
Fig. 6. Ln chlorine dioxide concentration vs. time at 27 °C. Chlorine dioxide initial dosage
was 1.4 and 1.0 mg/L.
Using the trend lines (representing each dosage), the bulk decay co-
efficient (Kb) was calculated as well as the decay rate (day−1). To have
one unique bulk decay coefficient (Kb) representing all the dosages at
27 °C, the average for all calculated coefficients was considered.
Table (3) shows all the dosages results at 27 °C including the calculated
bulk decay coefficients and the estimated decay rates.
Fig. 8. Ln chlorine dioxide concentration vs. time at 35 °C. Chlorine dioxide initial dosage
was 1.4, 1.2 and 0.8 mg/L.
3.4. Chlorine dioxide decay rate at 35 °C
Table 3
Chlorine dioxide concentration vs. time at 27 °C for all dosages: 1.4, 1.2, 1.0, 0.8 and
0.6 mg/L along with the calculated bulk decay coefficients at each dosage.
Table 4
Chlorine dioxide concentration vs. time at 35 °C for all dosages: 1.4, 1.2, 1.0, 0.8 and
0.6 mg/L along with the calculated bulk decay coefficients at each dosage.
Table (4) shows all the dosages results at 35 °C including the calculated
bulk decay coefficients and the estimated decay rates.
3.5. Chlorine dioxide decay rate at 45 °C Fig. 12. Ln chlorine dioxide concentration vs. time at 45 °C. Chlorine dioxide initial dosage
was 1.0 and 0.8 mg/L.
3.5.1. Chlorine dioxide concentration vs. time at 45 °C
ClO2 concentration vs. time for the five different dosages at 45 °C is one unique bulk decay coefficient (Kb) representing all the dosages at
illustrated in Fig. 10. 45 °C, the average for all calculated coefficients was considered.
Table (5) shows all the dosages results at 45 °C including the calculated
3.5.2. Estimation of bulk decay coefficient (Kb) at 45 °C bulk decay coefficients and the estimated decay rates.
Figs. 11 and 12 illustrate the Ln(x) vs. time plots as well as both trend
lines and R2. 3.6. Relationship between bulk decay coefficient (Kb) and water
The ln results in all dosages show first order trend lines with R2 rang- temperature (°C)
ing between 0.9896 and 0.9624. Hence, at least 96.2% of the results fit
the first order trend line. To illustrate the relation between bulk decay coefficient (−Kb) and
Using the trend lines (representing each dosage), the bulk decay co- water temperature (°C), the average (−Kb) for each temperature was
efficient (Kb) was calculated as well as the decay rate (day−1). To have plotted vs. the corresponding temperature. Table (6) summarizes the
average (− Kb) for each temperature and Fig. 13 illustrates this
relationship.
The exponential relationship between the bulk decay coefficient and
the water temperature was calculated. The trend has an R2 of 0.9868.
Hence, the bulk decay coefficient (kb) could be calculated utilizing
the following Eq. (1):
Where:
Table 5
Fig. 10. Illustrates ClO2 concentration vs. time for the five different dosages at 45 °C. Chlorine dioxide concentration vs. time at 45 °C for all dosages: 1.4, 1.2, 1.0, 0.8 and
0.6 mg/L along with the calculated bulk decay coefficients at each dosage.
Table 6
Summary for the average calculated Kb and the corresponding temperature in °C.
20 −0.0763 1.8307
27 −0.1029 2.4686
35 −0.1566 3.7581
Fig. 11. Ln chlorine dioxide concentration vs. time at 45 °C. Chlorine dioxide initial dosage
45 −0.2110 5.0644
was 1.4, 1.2 and 0.6 mg/L.
50 T.A. Ammar et al. / Desalination 352 (2014) 45–51
Fig. 13. Relationship between bulk decay coefficient (Kb) and water temperature (°C). Fig. 14. Predicted chlorine dioxide concentrations (utilizing the proposed model) and the
observed concentration for 5 h interval:.
3.7. Mathematical form for chlorine dioxide decay model 3.9. Implementation in drinking water distribution systems
As illustrated in ln figures for all five dosages (1.4, 1.2, 1.0, 0.8 and 3.9.1. Distribution system design and modeling
0.6 mg/L) over the tested temperature ranges (20°, 27°,35°, 45 °C), all For network designers, the proposed novel model provides an un-
the trend lines fit a first order equation. The R2 for all the trend lines precedented opportunity to more realistically plan and model distribu-
was ideal and ranged from 0.9421 to 0.9949. Therefore, a first order ki- tion systems to achieve disinfection; among other water quality; goals.
netic model could be proposed for the estimation of the decay rate uti- Moreover, the proposed novel model can confidently assist chlorine di-
lizing the following formula (Eq. (2)): oxide disinfection planning/management in drinking water distribution
systems.
Network modelling software such as EPANET provides the platform
Ct ¼ C0 expð−kb tÞ: ð2Þ where the proposed chlorine dioxide decay model can be implemented.
This enables system planners/managers to find efficiently the proper
chlorine dioxide disinfection dosages that allow residual targets at sys-
Where: tem extremities to be met.
Table 7
Predicted chlorine dioxide concentration vs. observed concentration.
C0 a Time interval (hr) Time (hr) Time (days) Tb Kb (day−1)c Ctd Observed ClO2 (mg/L) Difference (Ct-observed)
variables include; but not limited to; boosters design, placement, number • There is good agreement between calculated and measured chlorine
of stations and operation control & operation optimization. dioxide values as indicated through the model verification.
3.9.3. Operation control for disinfection systems Therefore, the proposed novel model is appropriate and accurate for
Operational control and operation optimization strategies for chlo- predicting chlorine dioxide bulk decay in drinking water systems over
rine dioxide disinfection systems can be easily adopted utilizing the pro- both a wide range of temperatures (from 20° to 45 °C) and a wide
posed model. Thus, water treatment plant managers can easily plan/ range of chlorine dioxide concentrations (0.6, 0.8, 1.0, 1.2 and 1.4 mg/L).
alter chlorine dioxide dosages vs. seasonal temperature profiles/changes. Moreover, the proposed novel model can confidently assist chlorine
Those planned dosages may be fine-tuned by careful monitoring for dioxide disinfection planning/management in drinking water distribu-
chlorine dioxide residual at the far ends of the network. Optimum levels tion systems. For network designers, the proposed model provides an
will take account the chlorine dioxide decay in the distribution system unprecedented opportunity to more realistically plan and model distri-
under normal and drain-down conditions. bution systems to achieve disinfection; among other water quality;
goals. Network modelling software such as EPANET provides the
3.10. Wall coefficient and decay modeling improvement platform where the proposed chlorine dioxide decay model can be
implemented.
Given that the proposed model provides the chlorine dioxide bulk This novel model can aid in the decision variables (design, place-
decay only, a deeper understanding of the mechanisms underlying chlo- ment, number of stations and operation control & operation optimiza-
rine dioxide consumption in piped systems, namely the effect of pipe tion strategies) for primary and booster chlorine dioxide disinfection
material and surface conditions, including those associated with biofilm, systems in water distribution systems.
would be recommended for future research. Such researches are still Given that the proposed model provides the chlorine dioxide bulk
needed for a more accurate estimation of kw and chlorine dioxide decay only, a deeper understanding of the mechanisms underlying chlo-
decay modeling improvement. rine dioxide consumption in piped systems would be recommended for
future research.
4. Conclusion
In order to ensure delivery of high quality drinking water from treat- References
ment plant to customers, maintaining adequate levels of disinfectant is
[1] World Health Organization, Guidelines for Drinking-water Quality, Fourth edition
very important, so it's necessary to study and model disinfectant decay WHO Press, Geneva, 2011.
rate. [2] Mark W. LeChevallier, The case for maintaining a disinfectant residual, J. AWWA 91
Several studies addressed the decay rate for chlorine while no study (1) (1999) 86–94.
[3] Gates J. Donald, The Chlorine Dioxide Handbook (Water Disinfection Series),
so far tackled the chlorine dioxide decay. This research provides a AWWA Publishing, 1998.
unique insight into decay of chlorine dioxide disinfectant in desalinated [4] C.J. Volk, et al., Implementation of chlorine dioxide disinfection: effects of the treat-
water systems and to propose a novel model for chlorine dioxide decay ment change on drinking water quality in a full-scale distribution system, J. Environ.
Eng. Sci. 1 (5) (2002) 323–330.
prediction. Such model can aid not only in disinfection planning/man- [5] USEPA, Alternative Disinfectants and Oxidants Guidance Manual, US Environmental
agement but also in the design and optimization of booster disinfection Protection Agency, 1999.
stations. [6] Karen S. Werdehoff, Philip C. Singer, Chlorine dioxide effects on THMFP, TOXFP and
the formation of inorganic by-products, J. AWWA 79 (9) (1987) 107–113.
The research work indicated the following:
[7] Ch. Rav-Acha, et al., Disinfection of drinking water rich in bromide with chlorine and
chlorine dioxide, while minimizing the formation of undesirable byproducts, Water
• The ln results for all the dosages (0.6, 0.8, 1.0, 1.2 and 1.4 mg/L) at all
Sci. Technol. J. 17 (4–5) (1984) 611–621.
the testing temperature (20°, 27°, 35° and 45 °C) follow a first order [8] M.H. Griese, et al., Combining methods for reduction of oxychlorine compounds in
mathematical form for chlorine dioxide decay model with strong con- drinking water, J. AWWA 84 (1992) 69–77.
fidence (almost all the trend lines fit the first order with R2 “Pearson [9] R.G. Zika, et al., Sunlight-induced photodecomposition of chlorine dioxide, water
chlorination: chemistry, Environ. Impact Health Eff. J. 5 (1985) 1041–1053.
Coefficient of Determination” ranging between 0.9949 and 0.9421). [10] M.G. Naack, R.L. Doerr, Reactions of Chlorine, ClO2 and Mixtures Thereof With
Hence, the decay rate will be calculated using the first order mathe- Humic Acid, in: Jolley, et al., (Eds.), Water Chlorination, Environmental Impact
matical form: and Health, vol. 2, Ann Arbor Science publishers, 1981.
[11] J.J. Vasconcelos, et al., Characterization and Modeling of Chlorine Decay in Distribu-
tion Systems, AWWA, New York, 1996.
Ct ¼ C0 expð−kb tÞ: [12] C.F. Foster, et al., Factors which control bulk chlorine decay rates, Water Res. J. 34 (1)
(2000) 117–126.
Where: [13] B.H. Nicholas, et al., A suitable model of combined effects of temperature and initial
condition on chlorine bulk decay in water distribution systems, Water Res. J. 46 (10)
(2012) 3293–3303.
Ct Chlorine dioxide concentration at any time. [14] R.S. Powell, et al., Calibration and comparison of chlorine decay models for a test
C0 Initial chlorine dioxide concentration. water distribution system, Water Res. J. 34 (8) (2000) 2301–2309.
Kb Bulk decay coefficient (day−1) [15] H.R. Maier, et al., Forecasting chlorine residuals in a water distribution system using
a general regression neural network, Math. Comput. Model. J. 44 (5–6) (2006)
t Time (days). 469–484.
[16] I. Fisher, et al., Evaluation of suitable chlorine bulk-decay models for water distribu-
tion systems, Water Res. J. 45 (16) (2011) 4896–4908.
• The bulk decay coefficient (Kb) could be calculated based on the fol- [17] R.A. Falconer, et al., Appraisal of chlorine contact tank modeling practices, Water
lowing exponential formula: Res. J. 46 (18) (2012) 5834–5847.
[18] A. Sathasivanet, A.J. Kohpaei, Chlorine decay prediction in bulk water using the par-
Kb ¼ 0:8147 expð0:0415TÞ: allel second order model: an analytical solution development, Chem. Eng. J. 171 (1)
(2011) 232–241.
[19] H. Fang, et al., Modeling of chlorine decay in municipal water supplies, Water Res. J.
Where: 33 (12) (1999) 2735.
[20] J.R. West, et al., Performance of various kinetic models for chlorine decay, Water
Kb Bulk decay coefficient (day−1). Resour. Plan. Manag. J. 126 (1) (2000) 13–20.
[21] American Public Health Association, Standard Methods for the Examinations of
T Water temperature (°C). Water and Waste Water, 22nd ed. APHA AWWA-WEF, Washington, DC, 2012.