- Email: [email protected]
Scan of water treatment processes to achieve desirable chlorine stability in water supply systems
Accepted Manuscript Title: Scan of water treatment processes to achieve desirable chlorine stability in water supply systems Authors: G. Kastl, I. Fisher, A. Sathasivan PII: DOI: Reference:
S0957-5820(17)30185-4 http://dx.doi.org/doi:10.1016/j.psep.2017.06.005 PSEP 1085
To appear in:
Process Safety and Environment Protection
Received date: Revised date: Accepted date:
30-12-2016 26-5-2017 7-6-2017
Please cite this article as: Kastl, G., Fisher, I., Sathasivan, A., Scan of water treatment processes to achieve desirable chlorine stability in water supply systems.Process Safety and Environment Protection http://dx.doi.org/10.1016/j.psep.2017.06.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Scan of water treatment processes to achieve desirable chlorine stability in water supply systems Kastl G.a, Fisher I.b and Sathasivan A.a a
School of Computing, Engineering and Mathematics, University of Western Sydney, Australia, (email: [email protected] phone +61415 819 514), bWatervale Systems, Sydney, Australia
Graphical abstract
Highlights
NOM removal with a pre-ozonation and coagulation with ferric salts
Chlorine decay test provides a better indication of water quality than DOC
Chlorine decay model parameters can be used for comparison of treatments
The chlorine decay model can predict chlorine and THMs (trihalomethane) profile
Abstract Removal of natural organic matter (NOM) is increasingly important in production and distribution of drinking water complying with health and aesthetic guidelines. The impact of treatment is often investigated using the resulting dissolved organic carbon (DOC) concentration and trihalomethane formation potential. Water treatment processes of ferric and alum coagulation with and without pre-oxidation by ozone and permanganate were scanned for their ability to treat synthetic raw water from peat extract. To compare the efficiency of the treatment methods the dissolved organic carbon (DOC) concentration and chlorine decay and trihalomethane (THM) formation tests – the important parameters needed in practice were used. A high level of DOC removal (> 80%) was achieved with all the treatments. A significant difference was observed in chlorine decay rates and THM production. Pre-oxidation with 2 mg/L of O3 and coagulation with 45 mg/L of ferric sulphate produced the most chlorine stable water followed by water treated with 80 mg/L of ferric sulphate. Chlorine decay test was found to be a more sensitive indicator of a successful treatment than DOC. The optimisation of the ozone dose is likely to identify the best water quality and chemical cost combination which is not likely achievable by the coagulation alone. .
Keywords: NOM removal, DOC removal, coagulation, chlorine decay model, THMs formation, pre-ozonation
1 Introduction Natural Organic Matter (NOM) concentration in natural water resources is increasing globally (Worrall and Burt, 2009, Evans et al., 2005). The pressure on consistent disinfection and simultaneous control of disinfection by-products requires reduction of NOM by water treatment plants. Recently (Zhang et al., 2015) summarised a number of processes for NOM removal, including: 1. enhanced coagulation; 2. granulated Activated Carbon (GAC) and Powdered Activated Carbon (Kristiana et al., 2011); 3. ion exchange resins; 4. oxidation (Cl, O3, KMnO4, UV, H2O2 and ferrate) pre-reaction rather than removal; 5. separation of NOM by membrane; and 6. biological treatment. The most cost-effective method of NOM removal for a majority of cases is enhanced coagulation (USEPA, 2001). Enhanced coagulation refers to the use of a higher dose of coagulant than would be required for turbidity reduction alone. The advantage of enhanced coagulation is that it can be easily adopted in a classical water treatment plant using coagulation/flocculation, sedimentation and filtration. In most cases, it is more efficient for NOM reduction to use ferric rather than aluminium coagulants (Volk et al., 2000, Kastl et al., 2004) at low pH (5-6). After settling, the pH can be re-adjusted to higher values to oxidise and remove iron and manganese on media filters. However, the higher dose of coagulant needed for enhanced coagulation results in the production of more sludge, which has to be processed in an environmentally safe manner and contributes to increased salinity according to following equations. FeCl3 + 3H2O = Fe(OH)3 + 3HCl
Eq. 1
Subsequently, HCl has to be neutralised, usually by calcium hydroxide or sodium hydroxide or carbonate. As calcium is a cheaper source of alkali, it is usually used to increase pH. It is also considered healthier to supplement water with calcium rather than to increase its sodium load (Brown et al., 2009). Additionally, the increase in calcium carbonate saturation reduces metal and concrete corrosion. 2HCl + Ca(OH)2 = CaCl2 +2H2O
Eq. 2
HCl + Na2CO3 = NaCl+ NaHCO3
Eq. 3
Removal of NOM by ferric coagulants results from charge neutralisation and surface adsorption (Kastl et al., 2004, Duan and Gregory, 2003). Therefore, NOM is concentrated on ferric sludge in its original form and removed with the ferric sludge by sedimentation and filtration. Loading of ferric hydroxide is typically low; 20-40 mg/L of FeCl3 is expected to remove 2-4 mg/L of dissolved organic carbon (DOC), assuming that the raw water has DOC in the range 4-10 mg/L. Understandably, hydrophobic and larger molecules are easier to remove than small molecules (Matilainen et al., 2010). The purpose of NOM removal is to achieve better stability of chlorine and thus longer lasting disinfection and reduction of disinfection by-products(Matilainen et al., 2011, Bond et al., 2014). A conversion from a laboratory study into the application of a water supply system requires multidisciplinary input(Kastl et al., 2016). System/modelling engineers need to determine the level of NOM that will be acceptable in a given distribution system (in terms of chlorine stability and THM formation) and process engineers have to decide which process or a combination of processes can achieve the lowest/acceptable cost and operational robustness. The first step in conversion to practical application in a water supply system is to evaluate the chlorine reactivity and THM formation characteristics (Fisher et al., 2011b). While DOC is a good surrogate for chlorine stability in many cases, for more accurate determination of chlorine reactivity, a chlorine decay test is the ultimate measure. Simple variants of chlorine decay testing and THMs formation is a Distribution system simulation test (72h) and THMs formation potential test THMFP(de la Rubia et al., 2008). The Three Days Distribution System Simulation test aims at finding an initial chlorine dose which will result in 0.2-0.5 mg/L of free chlorine after 3 days. TMHFP measures the maximum amount of THMs produced after a large chlorine dose and sufficient reaction time at an elevated temperature. These tests provide useful indicators of chlorine stability and tendency to form THMs, but only a combined chlorine and THMs model (Kastl and Fisher, 1997) has the capacity to quantitatively predict chlorine and THMs profile in a distribution system under specific conditions. Kast et al. (2015) argued that requirements for the concentration of DOC in treated water or chlorine stability and THM formation characteristics are determined by a number of factors, with the most important being:
Regulations governing concentration of chlorine disinfectant in drinking water (max and min)
Water age in the distribution system Water temperature Maximum allowed concentration of disinfection by-products (DBPs) such as THMs and HAAs
In Australia, Australian drinking water guidelines (NHMRC, 2011) limit THMs to 0.25 mg/L. The DOC concentration in many raw waters is often high and it would be technologically challenging to reduce it to such a level that THMs can be consistently controlled near US EPA or EU guidelines of 0.08 or 0.1 mg/L respectively. The control of the formation of THMs in treated natural waters can be achieved in principle by either the removal of organic compounds that react with chlorine or by pre-reaction of the compounds with an oxidant that does not form THMs. These two strategies can be combined and, for example, compounds which cannot be easily removed by coagulation can be pre-oxidised. The limits of DOC reduction by enhanced coagulation have been described by a mathematical model (Kastl et al., 2004). According to the model, by increasing the coagulant dose at pH around 4, the remaining DOC in water is approaching the non-adsorbable DOC in the source water. Pre-oxidation before coagulation impacts the efficiency of DOC removal and consumes oxidant-reactive sites(Crittenden et al., 2012). In addition, both chlorine and ozone are effective in improving DOC removal by coagulation (Ghadimkhani et al., 2006, Xie et al., 2016). Chlorine has the disadvantage of producing DBPs (THMs and HAAs) and for that reason was not considered as a suitable pre-oxidant. Additionally, a pre-oxidant affects not only DOC concentration after coagulation but also chlorine stability of the remaining DOC (Agbaba et al., 2014, Crittenden et al., 2012). To scan potentially suitable processes compatible with a classical water treatment plant to produce treated water with sufficient chlorine stability and less THM, the following objectives were formulated: 1. Compare DOC removal with ferric sulphate, aluminium sulphate and investigate the effect of pre-oxidation (ozonation and potassium permanganate) 2. Measure chlorine decay and THM formation of the produced waters. 3. Compare methods of water quality characterization of the treated water (DOC with chlorine decay and THM formation)
2
Experimental Methods and Materials
2.1 Experimental design Experimental design to test DOC removal and chlorine stability after various coagulation treatments and coagulation with pre-oxidation is shown in Figure 1.
Figure 1 schematics of experimental design coagulation with pre-oxidation
The same synthetic water was used in all coagulation tests (Figure 1). Coagulation was performed with ferric sulphate and aluminium sulphate. The pH was adjusted with sodium
carbonate. The effect of pre-oxidation was tested on coagulation with ferric sulphate using ozone and potassium permanganate as pre-oxidants. The coagulant doses were selected on the basis of preliminary tests, to be sufficient to achieve formation of floc and high DOC removal.
2.1.1 Synthetic water model Synthetic model water was used to simulate consistent source water with moderately high DOC of approximately 7.5 mg/L. Water was prepared by extracting mountain fibre peat (Aqua exotic, Slovakia) in Milli-Q water. This extract contained humic substances (Hsu and Singer, 2010) commonly present in NOM, (Pivokonsky et al., 2015). The apparent molecular weight (MW) distribution of the humic substances, determined by high-performance size exclusion chromatography (HPSEC), shows that the majority of molecules fall in the range of 4.3-15.8 kDa with the highest UV signal for MW of 10.8 kDa. This MW is in a higher range than would be expected for natural waters (Zhan et al., 2012) and therefore higher removal of DOC by coagulation would be expected in such synthetic water. The DOC of the initial concentration of the HS extract was 18 mg/L. It was diluted with Milli-Q water to reach a DOC concentration of approximately 7.5 mg/L, which represents a DOC content common for surface waters (Fabris et al., 2008). Based on the distribution of MW it would be anticipated that most of the DOC would be removable by coagulation. 2.1.2 Chemicals used All chemicals used – ferric sulphate, aluminium sulphate, potassium permanganate and sodium carbonate – were of analytical grade and the sodium hypochlorite was of laboratory grade. 2.1.3 Pre-ozonation To test the effect of ozone pre-oxidation on DOC removal by coagulation, a 1L sample was dosed with a preselected dose of ozone. Pre-ozonation of synthetic raw water was carried out in a glass column of inner diameter 55 mm and height 800 mm. Ozone was generated from dry oxygen gas by LIFEPOOL 1.0/OXCW apparatus (manufacturer Lifetech, the Czech Republic). The generated ozone concentration was 1-10mg O3/L. The pre-ozonation was operated in a batch mode and the ozone dose was controlled by time of bubbling the ozone/oxygen mixture through the water column. Doses of ozone were set to 1, 2 and 4 mg/L.
2.1.4 Jar Test Coagulation was performed in a 2L round container with a variable speed mixer. The container was filled with 1L of synthetic water and dosed with the coagulant under rapid mixing conditions of 200 rpm for 60 seconds. Slow mixing was conducted at 20 rpm for 30 minutes. The pH was adjusted by dosing sodium carbonate solution in the early stages of slow mixing. Water was allowed to settle for 30 minutes. Settled water was clear (except in KMnO4 pre-oxidised samples) and contained minimal visible suspended particles. 2.1.5 Centrifugation Settled water pH was measured as pH determines DOC equilibrium on floc (Kastl et al., 2004), then the sample was centrifuged (at 3000rpm for 5 minutes) to simulate media
filtration. The main driver for centrifugation was to protect the DOC analyser from particulate matter. Additionally, it would be expected that the separation of particulate matter by centrifugation may (as would media filtration) marginally reduce the DOC concentration.
2.1.6 Analysis Samples of produced water were analysed for: 2.1.6.1 DOC The DOC was measured using a total organic carbon analyser (TOC-VCPH) (Shimadzu, Japan) with an estimated accuracy of 5%. The measurements were carried in triplicate and eventually repeated if the variation was too high. The NOM removal was determined as DOC before and after the treatment. The DOC was measured by converting all organic compounds in water to CO2 by combustion and by detecting the CO 2 produced. By subtracting inorganic carbon present in water in the form of carbonates from the total amount of carbon, the organic carbon was calculated. Water was centrifuged after coagulation (at 3000rpm for 5 minutes) to remove particles. Samples of centrifuged water were used for DOC determination and chlorine decay tests. 2.1.6.2 Chlorine decay and THM formation The treated water sample was dosed with nominal 3.5 mg/L of chlorine. A chlorine stock solution was prepared by dilution of nominal 15% NaClO (Sigma-Aldrich) in Milli-Q water. A sample of Milli-Q water was also dosed with an equal amount of chlorine and used as a blank for chlorine decay without DOC. The chlorine concentration in samples was determined using the DPD (N, N diethyl-paraphenylenediamine) colorimetric method (4500CI G. DPD Colorimetric Method (Eaton et al., 2008)). Chlorine concentration was monitored over time. At the end of chlorine decay tests, samples were analysed for THMs (using GCMS, headspace gas sample). 2.1.6.3 Derivation of chlorine decay and THM formation model parameters Reviews (Fisher et al., 2012, Fisher et al., 2011b, Fisher et al., 2011a) concluded that a tworeactant (2R) chlorine decay model is the simplest and sufficiently accurate description of chlorine reaction with organic compounds in natural waters. The formulation of chlorine decay and THM formation models is based on a description of the following reactions:
Cl + F = Cl- + THMs
Eq. 4
Cl + S = Cl- + THMs
Eq. 5
where: F is the fast reacting component S is the slow reacting component and THMs are trihalomethanes (CHCl3, CHBrCl2, CHBr2Cl and CHBr3).
The reactions described by Eq. 4 and Eq. 5 are of first order with respect to both chlorine and organic compounds so that the reaction rates can be described as follows:
dcCl k F cCl cF k S cCl cS k z dt
dcF k F cCl cF dt dcS kS cCl cS dt dcTHM dc yTHM (- Cl ) dt dt
Eq. 6
Eq. 7
Eq. 8
Eq. 9
where: kF and kS is the reaction rate constant for a given component [(mg/L) -1h-1] kz is the zero rate reaction rate constant for chlorine decomposition in pure water [mg/L/h] c is the concentration of a given component [mg/L[ yTHM is yield for THM [ mg THM/ mg Cl reacted] and t is time [h]. The concentrations of fast and slow reacting compounds can be interpreted as the maximum concentration of chlorine which can be consumed by organic compounds. Both compounds react with chlorine in parallel. Initially, the chlorine decay rate is dominated by the fast reacting compound. Once the fast reacting compound is consumed (usually within the first few hours) the chlorine decay rate is determined by reaction of the residual chlorine with the slow reacting compound (Jabari Kohpaei and Sathasivan, 2011). Water treatment can remove and, by pre-oxidation, modify various components of DOC. That explains the reduction of fast and slow compounds and also changes in selectivity of THMs (production of THMs per mg of reacted chlorine). As shown in Eq. 9, the formation of THMs is directly proportional to the amount of reacted chlorine. For this reason, it is not recommended to pre-chlorinate raw water. This practice is common, as pre-chlorination provides the benefit in the reduction of turbidity (Xie et al., 2016), but it invariably results in the production of THMs, which are often not removed by subsequent treatment processes and hence contribute to the final concentration of THMs in the distribution system. The modelling approach described above quantifies formation of THMs for any given dose of chlorine and position in the distribution system. Such an approach is superior to a THM
formation potential, which only determines the maximum concentration of THMs generated at an excessive chlorine dose and long contact time. It could also be noted that THM formation potential can be derived from the model (on the basis of the initial concentration of fast and slow reacting compounds and the THM yield). The described chlorine decay model was fitted to the derived data using AQUASIM software (Reichert et al., 1995, Fisher et al., 2011a). The software solves the differential equations Eq. 6 to Eq. 9 for an estimated value of the model parameters. It varies the estimated parameters until the best possible fit with the experimental data is achieved. Such a model can be used to simulate chlorine decay and THM formation tests for any initial chlorine dose and at any reaction time.
3 Results The results of DOC removal and the residual of metal coagulant after settling and centrifugation are shown in Table 1. The residual metal concentration suggests that most of the metal had been removed by settling and centrifugation. The implications of DOC results are discussed below. DOC measurements with a zero coagulant dose represent synthetic water without any treatment. The average DOC of all four prepared water is 7.78 mg/L and it is within ±2% of for all samples. This represents good reproducibility of preparation of water sample and DOC measurement. Table 1 Residual DOC and metal concentrations after coagulation
3.1 DOC results From Table 1 it is apparent that practically all coagulation conditions (except 35 mg/L dose of ferric sulphate and a sample with permanganate pre-oxidation) achieved exceptionally high DOC removal. DOC was approximately reduced from 7.8 mg/L to 1 mg/L which represents a removal of 87%. The high DOC removal is consistently reproducible which suggests that the used model water contained NOM fractions which were easier to remove than a typical natural source water where the highest reported removal is around 80%(Volk et al., 2000). High DOC removal in the tested water can be explained by the greater proportion of high molecular weight NOM (>4.3 kDa) fraction (Pivokonsky et al., 2015) than commonly encountered in surface waters (Zhan et al., 2012, Nissinen et al., 2001). DOC removal in all jar tests is generally high, and based on DOC results only, it is impossible to exactly rank the produced water quality as DOC is subjected to the experimental error and some remaining DOC may have been pre-reacted with an oxidant. A positive finding is that the chlorine decay results are sensitive enough to clearly differentiate the produced water quality. Pre-ozonated coagulation produced the lowest DOC concentration, using only 45 mg/L of ferric sulphate as a coagulant. The lowest DOC (0.64 mg/L) was also achieved with the lowest ozone dose (1 mg/L). It is likely that there is a local optimum in DOC removal for ozone pre-oxidation dose. This optimal dose can be between 0 and 2 mg/L of ozone.
Experiments with permanganate pre-oxidation did not fulfil expectations. Coagulation of permanganate treated water produced fine, poorly settling floc with unacceptably high concentrations of soluble manganese. Usually a low (<1 mg/L) dose of permanganate is applied in water treatment plants to oxidise manganese, but here the permanganate dose was much higher (> 3.0 mg/L), possibly the reason for the poor result.
3.2 Chlorine decay results The results of chlorine decay for samples coagulated with increasing doses of ferric sulphate are shown in Figure 2. It can be seen that coagulation with ferric sulphate dramatically reduces the chlorine decay with the increasing coagulant dose at pH around 5.5. Chlorination of raw water (DOC=7.8 mg/L) resulted in rapid chlorine decay, with the concentration dropping to near zero within 24h. The chlorine dose was repeated (2.7 mg/L), but even the second dose did not lead to a lasting chlorine concentration. Apparently, the chlorine demand of raw water was far too high for comparison with the same chlorine decay procedure used for coagulated samples. In other samples, a dramatic change in chlorine decay rates of raw and coagulated samples is consistent with high DOC removal efficiencies. The chlorine decay curves are more consistent with what could be expected from coagulation with increasing coagulant dose. While the residual DOC fluctuates with the increasing dose of coagulant (55 mg/L of ferric sulphate resulted in DOC of 0.95 mg/L and 80 mg/L dose of ferric sulphate produced DOC of 1.12 mg/L), the data for chlorine decay clearly indicate a consistent improvement in chlorine stability with the increasing coagulant dose. These results are against the traditional use of DOC as a surrogate for chlorine stability; i.e., the DOC values did not rank the water samples in the same way as the chlorine decay profiles (Table 3). Given the quick response of DOC measurements, DOC could be a good qualitative surrogate for chlorine decay if DOC is removed by physical means. If oxidants are used, they may change the nature of the remaining DOC and a direct comparison is no longer valid. DOC indicates the relative status in the treatment, but it is difficult to assess how such water would perform in a distribution system or, in other words, translate DOC to a chlorine decay curve.
Figure 2 Chlorine decay in raw water (and after re-chlorination) and in water treated with various doses of ferric sulphate FS (35, 55 and 80 mg/L at pH 5.5). Lines connect consecutive experimental points.
Chlorine decay curves from Figure 2 can be used to derive parameters of a chlorine decay model. The model parameters characterise a given sample of treated water and can be used to simulate chlorine decay; for example, chlorine concentration profiles in a distribution system can be predicted for various initial chlorine doses and consequently the adequacy of different water treatments(Fisher et al., 2016). The derivation of chlorine decay model parameters has the additional advantage of quantifying decay so that differently treated waters can be compared on a basis directly related to the initial “fast” and longer-term “slow” chlorine decay. As the rate of chlorine decay cannot be calculated based on water composition parameters such as DOC, this approach is the only feasible way to characterise chlorine decay in a specific water sample.
Figure 3 Chlorine decay experimental data (markers) and chlorine model (lines), FS ferric sulphate, AS aluminium sulphate, Oz ozone and PP potassium permanganate, number indicates the dose (details for each experimental condition are shown in Table 1)
It can be seen (Figure 3) that the chlorine decay model (Fisher et al., 2011a) describes chlorine profiles with an acceptable accuracy, from the initial fast decay in the first few hours to the slow decay over hundreds of hours. The most chlorine-stable water is produced with a pre-ozone dose of 2 mg/L and 45 mg/L ferric sulphate coagulation dose. This is understandable as this treatment option resulted in a lower DOC concentration than in other compared curves and ozone pre-oxidised chlorine-reactive sites on DOC molecules. The second most chlorine-stable water was produced by coagulation with 80 mg/L of ferric sulphate. Ferric salts at low pH (5-6) in higher doses are recognised as suitable for deep removal of DOC and excellent stabilisation of chlorine. The use of permanganate preoxidation followed by coagulation with 45 mg/L of ferric sulphate partially compensates for the reduction in ferric dose from 80 mg/L. Alum-treated water was the least stable even though DOC was not significantly higher. The derived chlorine decay parameters for water treated under different conditions are listed in Table 2.
Table 2 Parameters of chlorine decay model (Fast and Slow initial concentration of chlorine reactive compounds) for waters coagulated under various conditions
Concentrations of fast and slow reacting compounds for treated waters are shown in Figure 4. It can be seen that the variation in chlorine reactive compounds is much more discerning than values of DOC (Table 2 and Table 3). The THM yield is low on average. This fact may be attributed to the tested synthetic water which had a low bromide concentration (<0.020mg/L). Most variations in the THM yield can be explained by the changing DOC composition by coagulation. Increasing the dose of ferric sulphate increases THMs yield (FS 25, FS55 and FS80) consistently. This could be possible if organic compounds with predominantly chlorine reducing properties are removed preferentially to compounds which form chlorinated products. Amongst pre-ozonated samples, the highest yield was recorded in a sample with the highest chlorine stability. Interestingly, all THM yields of ozonated waters were low. A possibility may be that the maximum reduction of chlorine reactive compounds does not coincide with the proportional removal of THMs producing compounds. In principle, each treatment method can have different affinity for reduction of sites for oxidation and for halogenation.
Figure 4 Initial concentrations of Fast and Slow chlorine decay components (FS – Ferric sulphate, O3 – ozone; each test followed by coagulant dose in mg/L)
Figure 4 indicates that there is a significant difference in total amount of chlorine-reactive compounds in differently treated water samples. This comparison presents a straightforward and obvious indicator of chlorine stability. The best chlorine stability based on Figure 4 was for O3 2 FS45 mg/L, which is the water treated with 2 mg/L of pre-ozonation and coagulation with 45 mg/L of ferric sulphate. A comparison of DOC and chlorine-reactive compounds is shown in Table 3. Table 3 Comparison of concentration of Chlorine reactive compounds and DOC
Table 3 shows that DOC and chlorine-reactive compounds do not provide parallel conclusions about chlorine stability of the produced water. Excluding the lowest quality water (FS 35), DOC ranges from 0.64 to 1.27 mg/L – a ratio of approximately 2, while chlorinereactive compounds range from 3.79 to 17.27 mg/L – a ratio of 4.5. This suggests that for a similar relative error of the measurement, the conclusion from chlorine-reactive compounds is likely to be more reliable than DOC measurements.
4 Discussion DOC removals are very high considering that the commonly accepted range of DOC removal in natural water sources is up to 80%. High DOC removal may be due to the nature of DOC which may be easier to remove, as indicated by high molecular mass 4-15.8 kD of NOM derived from this source (Pivokonsky et al., 2015). In modelling DOC removal using a three-component model of nonpolar, polar and inert non-adsorbable compounds, Kastl et al. (2004) found that the inert non-adsorbable fraction in all tested waters was higher than 0.2, suggesting that the maximum DOC removal should approach 80%. In the case of humic substance, Humat Star80-Almedsa Agrochimic-Algeria both alum and ferric coagulant achieved practically complete removal of humic substances from water (Harfouchi et al., 2016). When it is not possible to directly predict an impact of DOC concentration on chlorine stability and disinfection by-products (DBP) formation, an empirical rule of the thumb is used to determine how far to increase the coagulant dose. For example, one such a rule requires that an increase of the coagulant dose by 10 mg/L results in the additional DOC removal by a minimum of 0.3 mg/L (Volk et al., 2000) to identify the point of diminishing return. US EPA uses prescriptive removal rates based on the raw water DOC concentration and alkalinity. The highest required DOC removal of 50% is specified for low alkalinity water with DOC >8 mg/L. A discussion of the application of these rules is presented by Kastl et al. (2016). In the absence of chlorine decay tests, the chemical oxygen demand of the treated water may be considered a better indicator of water quality than DOC. A comparison of DOC and chemical oxygen demand KMnO4 (CODKMnO4) removal with enhanced coagulation indicates that CODKMnO4 is removed more efficiently than DOC (Pivokonska et al., 2008). It was possible to remove 58% of DOC and 65% CODKMnO4 in one water and 41% DOC and 71% COD in another water sample. These results indicate that CODKMnO4, which is expected to be
closely related to the chlorine stability, is not removed at the same rate by coagulation as DOC is. Additionally, it may be argued that, while CODKMnO4 may be a better indicator of chlorine stability than DOC, it is still not a true representation of the chlorine stability which is the ultimate arbiter in the determination of process efficiency. CODKMnO4 has been widely used as a surrogate for chlorine stability before the introduction of automated DOC instrumentation. The reliability of DOC concentration as a chlorine decay predictor is seldom evaluated; therefore, this work also compares the use of DOC and chlorine decay and THM formation measurements. To improve the chlorine stability, pre-oxidation is an additional option to consume chlorinereactive molecules. It may be applied before, or as frequently is after flocculation/ sedimentation. It is often applied even after filtration when filtered water is, for example, ozonated and passed via biological activated carbon (BAC) to stabilise water to control biological regrowth in the distribution system. If pre-ozonation is followed by coagulation, it would be expected that pre-ozonation would oxidise functional groups amenable to chlorine oxidation. This action would also be expected to make organic molecules more hydrophilic and therefore less amenable to adsorption on ferric hydroxide or activated carbon surface. A typical application of ozonation after coagulation /sedimentation/ filtration would be consistent with this concept. The coagulation /sedimentation/ filtration would remove most of the DOC which can be removed by charge neutralisation and adsorption on metal hydroxide such as Fe(OH)3. The following ozonation could only react with those organics which are difficult to remove by coagulation and are susceptible to oxidation. This simple concept would suggest that post-ozonation would produce more chlorine-stable water using the same ferric and ozone dose. On the other hand, the experiments often show that pre-ozonation initially increases DOC removal in consequent coagulation and that higher pre-ozonation doses may lead to the reduction of DOC removal. Higher doses of ozone make DOC removal somewhat unpredictable (Agbaba et al., 2014). The explanation that ozone attacks double bonds and makes DOC molecules more hydrophilic sounds plausible, but the inference that this leads to better DOC removal by enhanced coagulation is not sufficiently supported, as more studies report better removal of hydrophobic DOC (Xie et al., 2016). The impact of preozonation on DOC removal usually shows an optimum at a relatively low ozone dose. An explanation of the improved DOC removal by enhanced coagulation in pre-ozonated water samples may be that forming additional organic acid groups on the surface of colloidal particles enhances the impact of charge neutralisation mechanisms which is more efficient in DOC removal than absorption on the ferric hydroxide surface (Xie et al., 2016). This hypothesis is consistent with the observation (Table 1) that, at higher pre-ozonation doses, DOC removal declines as larger DOC molecules are broken down to smaller and more hydrophilic molecules, which are less susceptible to both charge neutralisation by coagulants and adsorption on ferric hydroxide surface. Pre-ozonation before coagulation results in a higher reduction of THMs formation potential than coagulation alone (Chiang et al., 2009). This effect can be explained by increased DOC removal and pre-oxidation of chlorine-reactive compounds. An average THM yield (Table 2) is 0.020mg THMs/mg Cl with the minimum and maximum measured the yield of 0.012 and 0.031 mg/mg respectively. Under the chlorination test, all samples generated THMs below 0.06 mg/L, which is well below USEPA guideline limit of 0.08mg/L. Based on these results
the selection of a treatment process for this type of water would be driven by the required chlorine stability and operation cost.
4.1 Practical considerations Based on the conducted experiments and evaluations, pre-ozonation with 2 mg/L of ozone followed by coagulation with ferric sulphate (45mg/L) resulted in the most chlorine stable water. This treatment generated lower DOC and better chlorine stability than water treated with 80 mg/L of ferric sulphate. Assuming cost (US$ in 2016) of ozone at $10/kg and ferric sulphate cost at $0.8 /kg, then the chemical cost for the above two treatments would be: For 1 ML pre-ozonated water
45kg/ML*$0.8/kg + 2 kg/ML*$10/kg=36 + 20= $56/ML
For 1 ML 80mg/L ferric sulphate
80kg/ML*$0.8/kg = $64/ML
As shown by the chemical cost calculation, the use of ozone not only reduces the dose of the coagulant, and therefore the amount of sludge for disposal, but also results in lower cost of the chemicals and better chlorine stability in this case. Pre-ozonated water would benefit from a biologically active filter, either in the form of dual media or even more efficiently including GAC, as ozonation is known to generate assimilable organic carbon (AOC) (Świetlik et al., 2004, Ramseier et al., 2011). In this instance, the pre-ozonation dose is quite low (2 mg/L), but it is feasible that AOC formed by ozonation can support biofilm growth in the distribution system when chlorine disinfectant residual becomes low. A potential limitation of pre-ozonation may be bromate concentration generated by oxidation of bromide. Bromate has a limit of 0.01 mg/L WHO guideline or 0.02 mg/L (NHMRC, 2011). If not limited by bromate, the option with pre-ozonation appears attractive due to low coagulant dose, chemical cost, sludge production and production of the most chlorine-stable water. The implementation of the derived chlorine decay and THM formation models in a network model containing the EPANET MSX – Multi-Species eXtension – module (Shang et al., 2007) can predict chlorine and THM profiles in a specific distribution system for given hydraulic conditions. The ability to predict chlorine and THM profiles in a hypothetical distribution network for a treated water sample is a superior indicator of the treated water quality than any other, including DOC.
5 Conclusion This study searched for an efficient method of NOM removal by enhanced coagulation with ferric and aluminium salts, in combination with pre-oxidation (with ozone) or permanganate to achieve better chlorine stability and lower THM formation. The produced water quality was characterised by the measurements of DOC, chlorine decay and THM formation. The chlorine decay data were fitted to two-reactant (2R)(Fisher et al., 2012) model to estimate initial fast and slow reacting agents and their respective reaction rates. The treatment that included pre-ozonation with 2 mg O3/L and coagulation with 45 mg/L ferric sulphate resulted in the best chlorine stability and THMs below 0.06 mg/L. Coagulation with 80 mg/L ferric sulphate produced the second-best results. Aluminium sulphate treatment did not achieve chlorine stability of ferric sulphate treatment even at a higher dose (100 mg/L) and when the DOC concentrations were comparable.
DOC was confirmed to be a good qualitative indicator of the treatment process. However, the ranking of the treatment processes on the basis of DOC concentration does not identify the water with the highest chlorine stability. The discrepancy in DOC and chlorine decay results may be due to small differences in DOC relative to the measurement error and better differentiation of chlorine decay curves. The results obtained strongly suggest that in coagulation of medium-high DOC raw water (612 mg/L) with a ferric coagulant, better chlorine stability of the treated water can be achieved with a lower coagulant dose and an optimised pre-ozonation dose. The approach demonstrated in this paper is an efficient way to scan a broad range of potentially suitable coagulation-based water treatment processes for waters with a moderately high concentration of DOC.
6 Funding and potential conflict of interest This research was supported by The Australian Postgraduate Award scheme and did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors are not aware of any potential conflict of interest.
References AGBABA, J., TUBIĆ, A., DALMACIJA, B., WATSON, M., MOLNAR, J., RONČEVIĆ, S. & MALETIĆ, S. 2014. Investigation of the impact of ozone pretreatment and powdered activated carbon addition on the removal of natural organic matter by coagulation. Desalination and Water Treatment, 56, 912-920. BOND, T., HUANG, J., GRAHAM, N. J. & TEMPLETON, M. R. 2014. Examining the interrelationship between DOC, bromide and chlorine dose on DBP formation in drinking water--a case study. Sci Total Environ, 470-471, 469-79. BROWN, I. J., TZOULAKI, I., CANDEIAS, V. & ELLIOTT, P. 2009. Salt intakes around the world: implications for public health. Int J Epidemiol, 38, 791-813. CHIANG, P.-C., CHANG, E. E., CHANG, P.-C. & HUANG, C.-P. 2009. Effects of pre-ozonation on the removal of THM precursors by coagulation. Science of The Total Environment, 407, 57355742. CRITTENDEN, J. C., TRUSSELL, R. R., HAND, D. W., HOWE, K. J. & TCHOBANOGLOUS, G. 2012. MWH's water treatment: principles and design, John Wiley & Sons. DE LA RUBIA, A., RODRIGUEZ, M., LEON, V. M. & PRATS, D. 2008. Removal of natural organic matter and THM formation potential by ultra- and nanofiltration of surface water. Water Res, 42, 714-22. DUAN, J. & GREGORY, J. 2003. Coagulation by hydrolysing metal salts. Advances in Colloid and Interface Science, 100-102, 475-502. EATON, A., CLESCERI, L., RICE, E. & GREENBERG, A. 2008. Standard methods for the examination of water and wastewater. EVANS, C. D., MONTEITH, D. T. & COOPER, D. M. 2005. Long-term increases in surface water dissolved organic carbon: Observations, possible causes and environmental impacts. Environmental Pollution, 137, 55-71. FABRIS, R., CHOW, C. W., DRIKAS, M. & EIKEBROKK, B. 2008. Comparison of NOM character in selected Australian and Norwegian drinking waters. Water Res, 42, 4188-96.
FISHER, I., KASTL, G. & SATHASIVAN, A. 2011a. Evaluation of suitable chlorine bulk-decay models for water distribution systems. Water research, 45, 4896-4908. FISHER, I., KASTL, G. & SATHASIVAN, A. 2012. A suitable model of combined effects of temperature and initial condition on chlorine bulk decay in water distribution systems. Water research, 46, 3293-3303. FISHER, I., KASTL, G. & SATHASIVAN, A. 2016. A comprehensive bulk chlorine decay model for simulating residuals in water distribution systems. Urban Water Journal, 14, 361-368. FISHER, I., KASTL, G., SATHASIVAN, A. & JEGATHEESAN, V. 2011b. Suitability of chlorine bulk decay models for planning and management of water distribution systems. Critical reviews in environmental science and technology, 41, 1843-1882. GHADIMKHANI, A., TORABIAN, A. & MEHRABADI, A. 2006. Preozonation and prechlorination effect on TOC removal in surface water treatment. Pakistan Journal of Biological Science, 9, 708712. HARFOUCHI, H., HANK, D. & HELLAL, A. 2016. Response surface methodology for the elimination of humic substances from water by coagulation using powdered Saddled sea bream scale as coagulant-aid. Process Safety and Environmental Protection, 99, 216-226. HSU, S. & SINGER, P. C. 2010. Removal of bromide and natural organic matter by anion exchange. Water Research, 44, 2133-2140. JABARI KOHPAEI, A. & SATHASIVAN, A. 2011. Chlorine decay prediction in bulk water using the parallel second order model: An analytical solution development. Chemical Engineering Journal, 171, 232-241. KASTL, G. & FISHER, I. 1997. Predicting and Maintaining Drinking Water Quality in Distribution Systems. WATER-MELBOURNE THEN ARTARMON-, 24, 35-38. KASTL, G., SATHASIVAN, A. & FISHER, I. 2016. A selection framework for NOM removal process for drinking water treatment. Desalination and Water Treatment, 57, 7679-7689. KASTL, G., SATHASIVAN, A., FISHER, I. & VAN LEEUWEN, J. 2004. Modeling DOC removal by enhanced coagulation. Journal-American Water Works Association, 96, 79-89. KRISTIANA, I., JOLL, C. & HEITZ, A. 2011. Powdered activated carbon coupled with enhanced coagulation for natural organic matter removal and disinfection by-product control: Application in a Western Australian water treatment plant. Chemosphere, 83, 661-667. MATILAINEN, A., GJESSING, E. T., LAHTINEN, T., HED, L., BHATNAGAR, A. & SILLANPAA, M. 2011. An overview of the methods used in the characterisation of natural organic matter (NOM) in relation to drinking water treatment. Chemosphere, 83, 1431-42. MATILAINEN, A., VEPSALAINEN, M. & SILLANPAA, M. 2010. Natural organic matter removal by coagulation during drinking water treatment: a review. Adv Colloid Interface Sci, 159, 18997. NHMRC, N. 2011. Australian drinking water guidelines paper 6 national water quality management strategy. National Health and Medical Research Council, National Resource Management Ministerial Council, Commonwealth of Australia, Canberra. NISSINEN, T. K., MIETTINEN, I. T., MARTIKAINEN, P. J. & VARTIAINEN, T. 2001. Molecular size distribution of natural organic matter in raw and drinking waters. Chemosphere, 45, 865873. PIVOKONSKA, L., PIVOKONSKY, M. & TOMASKOVA, H. 2008. Optimization of NOM Removal during Water Treatment. Separation Science and Technology, 43, 1687-1700. PIVOKONSKY, M., NACERADSKA, J., BRABENEC, T., NOVOTNA, K., BARESOVA, M. & JANDA, V. 2015. The impact of interactions between algal organic matter and humic substances on coagulation. Water Res, 84, 278-85. RAMSEIER, M. K., PETER, A., TRABER, J. & VON GUNTEN, U. 2011. Formation of assimilable organic carbon during oxidation of natural waters with ozone, chlorine dioxide, chlorine, permanganate, and ferrate. Water Res, 45, 2002-10.
REICHERT, P., VON SCHULTHESS, R. & WILD, D. 1995. The use of AQUASIM for estimating parameters of activated sludge models. Water Science and Technology, 31, 135-147. SHANG, F., UBER, J. G. & ROSSMAN, L. A. 2007. Modeling reaction and transport of multiple species in water distribution systems. Environmental science & technology, 42, 808-814. ŚWIETLIK, J., DĄBROWSKA, A., RACZYK-STANISŁAWIAK, U. & NAWROCKI, J. 2004. Reactivity of natural organic matter fractions with chlorine dioxide and ozone. Water research, 38, 547558. USEPA 2001. Controlling Disinfection By-Products and Microbial Contaminants in Drinking Water. VOLK, C., BELL, K., IBRAHIM, E., VERGES, D., AMY, G. & LECHEVALLIER, M. 2000. Impact of enhanced and optimized coagulation on removal of organic matter and its biodegradable fraction in drinking water. Water Research, 34, 3247-3257. WORRALL, F. & BURT, T. P. 2009. Changes in DOC treatability: Indications of compositional changes in DOC trends. Journal of Hydrology, 366, 1-8. XIE, P., CHEN, Y., MA, J., ZHANG, X., ZOU, J. & WANG, Z. 2016. A mini review of preoxidation to improve coagulation. Chemosphere, 155, 550-63. ZHAN, W., SATHASIVAN, A., JOLL, C., WAI, G., HEITZ, A. & KRISTIANA, I. 2012. Impact of NOM character on copper adsorption by trace ferric hydroxide from iron corrosion in water supply system. Chemical Engineering Journal, 200–202, 122-132. ZHANG, Y., ZHAO, X., ZHANG, X. & PENG, S. 2015. A review of different drinking water treatments for natural organic matter removal. Water Science & Technology: Water Supply, 15, 442.
Figure 1 schematics of experimental design coagulation with pre-oxidation
4.5
Rechlorination
DOC 7.80, FS 0
4.0 3.5
DOC 2.39, FS 35 DOC 0.95, FS 55 DOC 1.12, FS 80 Blank
Cl (mg/L)
3.0 2.5 2.0 1.5 1.0 0.5 0.0 0
50
100
150
200
250
300
350
400
Time (h) Figure 2 Chlorine decay in raw water (and after re-chlorination) and in water treated with various doses of ferric sulphate FS (35, 55 and 80 mg/L at pH 5.5). Lines connect consecutive experimental points.
FS 80 FS 80 Oz 2mg/L Oz 2 FS 45 AS 100 AS 100 PP 7.0 FS4 5 PP 7.0 FS4 5
4.5 4
Free Cl (mg/L)
3.5 3
2.5 2 1.5 1 0.5
0 0
100
200
Time (h)
300
400
Cl demand as determined from initial slow and fast reacting compounds (mg/L)
Figure 3 Chlorine decay experimental data (markers) and chlorine model (lines), FS ferric sulphate, AS aluminium sulphate, Oz ozone and PP potassium permanganate, number indicates the dose (details for each experimental condition are shown in Table 1)
25 Initial fast mg/L
20 Initial slow mg/L
15
10
5
0
FS 35 FS 55 FS 80 Oz 1 Oz 2 Oz 4 AS 70 AS 80 AS 100 PP 7.0 FS45 FS45 FS 45 FS 45
Figure 4 Initial concentrations of Fast and Slow chlorine decay components (FS – Ferric sulphate, O3 – ozone; each test followed by coagulant dose in mg/L)
Table 1 Residual DOC and metal concentrations after coagulation
Coagulant +conditions (experiment code)
Coagulant Coagulation Dose pH [1] (mg/L)
DOC (mg/L)
Ferric sulphate (FS 0) Ferric sulphate (FS 35) Ferric sulphate (FS 55) Ferric sulphate (FS 80) Ozone pre-dose 0 mg/L Ferric sulphate (Oz 0 FS S0)
0 35 55 80 0
6.75 5.80 5.60 5.70 6.75
7.65 2.39 0.95 1.12 7.85
Residual metal Fe or Al or Mn (mg/L) n/a 0.68 0.38 0.10 n/a
Ozone pre-dose 1 mg/L Ferric sulphate (Oz 1 FS 45)
45
5..44
0.64
0.09
Ozone pre-dose 2 mg/L Ferric sulphate (Oz 2 FS 45)
45
5.47
0.78
0.12
Ozone pre-dose 4 mg/L Ferric sulphate (Oz 4 FS 45)
45
5.15
0.92
0.40
Aluminium sulphate (AS 0) 0 7.90 n/a Aluminium sulphate (AS 70) 70 5.66 1.04 0.033 Aluminium sulphate (AS 80) 80 5.75 0.99 0.076 Aluminium sulphate (AS 100) 100 5.71 0.92 0.025 Potassium permanganate-dose 0 0 7.72 n/a mg/L Ferric sulphate (PP 0 FS 0) Potassium permanganate-dose 45 5.50 High colour n/a 3.5 mg/L Ferric sulphate not (PP3.5FS45) determined Potassium permanganate dose 7 45 5.50 1.27 0.50 mg/L Ferric sulphate (PP 7.0 FS4 5) All entries in Table 1 with zero doses of coagulant represent untreated water. These values are reasonably close in the range of 7.65-7.90 mg/L which indicates good reproducibility in the preparation of raw water and DOC analysis. The coagulant dose is expressed as Fe2(SO4)3 and Al2(SO4)3.
Table 2 Parameters of chlorine decay model (Fast and Slow initial concentration of chlorine reactive compounds) for waters coagulated under various conditions
Initial THM Coagulant Dose DOC % DOC Initial fast slow yield +conditions (mg/L) (mg/L) removal (mg/L) (mg/L) (mg/mg) FS 35 35 2.39 68.8 1.20 19.97 0.013 FS 55 55 0.95 87.6 0.47 9.01 0.023 FS 80 80 1.12 85.3 0.01 4.79 0.031 Oz 1 FS 45 45 0.64 91.8 0.52 9.60 0.012 Oz 2 FS 45 45 0.78 90.0 0.37 3.42 0.022 Oz 4 FS 45 45 0.92 88.3 0.27 11.85 0.017 AS 70 70 1.04 86.8 0.98 16.29 0.015 AS 80 80 0.99 87.4 1.24 9.88 0.022 AS 100 100 0.92 88.4 0.68 13.04 0.021 PP 7.0 FS 45 45 1.27 83.6 0.93 6.14 n/a Initial fast and slow chlorine reacting compounds in Table 2 are hypothetical concentrations of chlorine-consuming compounds present in variously treated waters. THM yield is the amount of THMs produced by the reaction of 1 mg of chlorine. Coagulant – FS is ferric sulphate, AS – aluminium sulphate, Oz – ozone and PP – potassium permanganate. The number after indicates the dose in mg/L.
Table 3 Comparison of concentration of Chlorine reactive compounds and DOC
Coagulant +conditions
Treated water DOC [mg/L]
O3 2 FS45
0.78
Treated water Chlorine reactive Fast + Slow compounds [mg/L] 3.79
FS80
1.12
4.80
KMnO4 FS45
1.27
7.07
FS55 O3 1 FS45
0.95 0.64
9.48 10.13
Alum80 O3 4 FS 45
0.99 0.92
11.12 12.11
Alum100 Alum70
0.92 1.04
13.72 17.27
FS 35
2.39
21.17
S0957-5820(17)30185-4 http://dx.doi.org/doi:10.1016/j.psep.2017.06.005 PSEP 1085
To appear in:
Process Safety and Environment Protection
Received date: Revised date: Accepted date:
30-12-2016 26-5-2017 7-6-2017
Please cite this article as: Kastl, G., Fisher, I., Sathasivan, A., Scan of water treatment processes to achieve desirable chlorine stability in water supply systems.Process Safety and Environment Protection http://dx.doi.org/10.1016/j.psep.2017.06.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Scan of water treatment processes to achieve desirable chlorine stability in water supply systems Kastl G.a, Fisher I.b and Sathasivan A.a a
School of Computing, Engineering and Mathematics, University of Western Sydney, Australia, (email: [email protected] phone +61415 819 514), bWatervale Systems, Sydney, Australia
Graphical abstract
Highlights
NOM removal with a pre-ozonation and coagulation with ferric salts
Chlorine decay test provides a better indication of water quality than DOC
Chlorine decay model parameters can be used for comparison of treatments
The chlorine decay model can predict chlorine and THMs (trihalomethane) profile
Abstract Removal of natural organic matter (NOM) is increasingly important in production and distribution of drinking water complying with health and aesthetic guidelines. The impact of treatment is often investigated using the resulting dissolved organic carbon (DOC) concentration and trihalomethane formation potential. Water treatment processes of ferric and alum coagulation with and without pre-oxidation by ozone and permanganate were scanned for their ability to treat synthetic raw water from peat extract. To compare the efficiency of the treatment methods the dissolved organic carbon (DOC) concentration and chlorine decay and trihalomethane (THM) formation tests – the important parameters needed in practice were used. A high level of DOC removal (> 80%) was achieved with all the treatments. A significant difference was observed in chlorine decay rates and THM production. Pre-oxidation with 2 mg/L of O3 and coagulation with 45 mg/L of ferric sulphate produced the most chlorine stable water followed by water treated with 80 mg/L of ferric sulphate. Chlorine decay test was found to be a more sensitive indicator of a successful treatment than DOC. The optimisation of the ozone dose is likely to identify the best water quality and chemical cost combination which is not likely achievable by the coagulation alone. .
Keywords: NOM removal, DOC removal, coagulation, chlorine decay model, THMs formation, pre-ozonation
1 Introduction Natural Organic Matter (NOM) concentration in natural water resources is increasing globally (Worrall and Burt, 2009, Evans et al., 2005). The pressure on consistent disinfection and simultaneous control of disinfection by-products requires reduction of NOM by water treatment plants. Recently (Zhang et al., 2015) summarised a number of processes for NOM removal, including: 1. enhanced coagulation; 2. granulated Activated Carbon (GAC) and Powdered Activated Carbon (Kristiana et al., 2011); 3. ion exchange resins; 4. oxidation (Cl, O3, KMnO4, UV, H2O2 and ferrate) pre-reaction rather than removal; 5. separation of NOM by membrane; and 6. biological treatment. The most cost-effective method of NOM removal for a majority of cases is enhanced coagulation (USEPA, 2001). Enhanced coagulation refers to the use of a higher dose of coagulant than would be required for turbidity reduction alone. The advantage of enhanced coagulation is that it can be easily adopted in a classical water treatment plant using coagulation/flocculation, sedimentation and filtration. In most cases, it is more efficient for NOM reduction to use ferric rather than aluminium coagulants (Volk et al., 2000, Kastl et al., 2004) at low pH (5-6). After settling, the pH can be re-adjusted to higher values to oxidise and remove iron and manganese on media filters. However, the higher dose of coagulant needed for enhanced coagulation results in the production of more sludge, which has to be processed in an environmentally safe manner and contributes to increased salinity according to following equations. FeCl3 + 3H2O = Fe(OH)3 + 3HCl
Eq. 1
Subsequently, HCl has to be neutralised, usually by calcium hydroxide or sodium hydroxide or carbonate. As calcium is a cheaper source of alkali, it is usually used to increase pH. It is also considered healthier to supplement water with calcium rather than to increase its sodium load (Brown et al., 2009). Additionally, the increase in calcium carbonate saturation reduces metal and concrete corrosion. 2HCl + Ca(OH)2 = CaCl2 +2H2O
Eq. 2
HCl + Na2CO3 = NaCl+ NaHCO3
Eq. 3
Removal of NOM by ferric coagulants results from charge neutralisation and surface adsorption (Kastl et al., 2004, Duan and Gregory, 2003). Therefore, NOM is concentrated on ferric sludge in its original form and removed with the ferric sludge by sedimentation and filtration. Loading of ferric hydroxide is typically low; 20-40 mg/L of FeCl3 is expected to remove 2-4 mg/L of dissolved organic carbon (DOC), assuming that the raw water has DOC in the range 4-10 mg/L. Understandably, hydrophobic and larger molecules are easier to remove than small molecules (Matilainen et al., 2010). The purpose of NOM removal is to achieve better stability of chlorine and thus longer lasting disinfection and reduction of disinfection by-products(Matilainen et al., 2011, Bond et al., 2014). A conversion from a laboratory study into the application of a water supply system requires multidisciplinary input(Kastl et al., 2016). System/modelling engineers need to determine the level of NOM that will be acceptable in a given distribution system (in terms of chlorine stability and THM formation) and process engineers have to decide which process or a combination of processes can achieve the lowest/acceptable cost and operational robustness. The first step in conversion to practical application in a water supply system is to evaluate the chlorine reactivity and THM formation characteristics (Fisher et al., 2011b). While DOC is a good surrogate for chlorine stability in many cases, for more accurate determination of chlorine reactivity, a chlorine decay test is the ultimate measure. Simple variants of chlorine decay testing and THMs formation is a Distribution system simulation test (72h) and THMs formation potential test THMFP(de la Rubia et al., 2008). The Three Days Distribution System Simulation test aims at finding an initial chlorine dose which will result in 0.2-0.5 mg/L of free chlorine after 3 days. TMHFP measures the maximum amount of THMs produced after a large chlorine dose and sufficient reaction time at an elevated temperature. These tests provide useful indicators of chlorine stability and tendency to form THMs, but only a combined chlorine and THMs model (Kastl and Fisher, 1997) has the capacity to quantitatively predict chlorine and THMs profile in a distribution system under specific conditions. Kast et al. (2015) argued that requirements for the concentration of DOC in treated water or chlorine stability and THM formation characteristics are determined by a number of factors, with the most important being:
Regulations governing concentration of chlorine disinfectant in drinking water (max and min)
Water age in the distribution system Water temperature Maximum allowed concentration of disinfection by-products (DBPs) such as THMs and HAAs
In Australia, Australian drinking water guidelines (NHMRC, 2011) limit THMs to 0.25 mg/L. The DOC concentration in many raw waters is often high and it would be technologically challenging to reduce it to such a level that THMs can be consistently controlled near US EPA or EU guidelines of 0.08 or 0.1 mg/L respectively. The control of the formation of THMs in treated natural waters can be achieved in principle by either the removal of organic compounds that react with chlorine or by pre-reaction of the compounds with an oxidant that does not form THMs. These two strategies can be combined and, for example, compounds which cannot be easily removed by coagulation can be pre-oxidised. The limits of DOC reduction by enhanced coagulation have been described by a mathematical model (Kastl et al., 2004). According to the model, by increasing the coagulant dose at pH around 4, the remaining DOC in water is approaching the non-adsorbable DOC in the source water. Pre-oxidation before coagulation impacts the efficiency of DOC removal and consumes oxidant-reactive sites(Crittenden et al., 2012). In addition, both chlorine and ozone are effective in improving DOC removal by coagulation (Ghadimkhani et al., 2006, Xie et al., 2016). Chlorine has the disadvantage of producing DBPs (THMs and HAAs) and for that reason was not considered as a suitable pre-oxidant. Additionally, a pre-oxidant affects not only DOC concentration after coagulation but also chlorine stability of the remaining DOC (Agbaba et al., 2014, Crittenden et al., 2012). To scan potentially suitable processes compatible with a classical water treatment plant to produce treated water with sufficient chlorine stability and less THM, the following objectives were formulated: 1. Compare DOC removal with ferric sulphate, aluminium sulphate and investigate the effect of pre-oxidation (ozonation and potassium permanganate) 2. Measure chlorine decay and THM formation of the produced waters. 3. Compare methods of water quality characterization of the treated water (DOC with chlorine decay and THM formation)
2
Experimental Methods and Materials
2.1 Experimental design Experimental design to test DOC removal and chlorine stability after various coagulation treatments and coagulation with pre-oxidation is shown in Figure 1.
Figure 1 schematics of experimental design coagulation with pre-oxidation
The same synthetic water was used in all coagulation tests (Figure 1). Coagulation was performed with ferric sulphate and aluminium sulphate. The pH was adjusted with sodium
carbonate. The effect of pre-oxidation was tested on coagulation with ferric sulphate using ozone and potassium permanganate as pre-oxidants. The coagulant doses were selected on the basis of preliminary tests, to be sufficient to achieve formation of floc and high DOC removal.
2.1.1 Synthetic water model Synthetic model water was used to simulate consistent source water with moderately high DOC of approximately 7.5 mg/L. Water was prepared by extracting mountain fibre peat (Aqua exotic, Slovakia) in Milli-Q water. This extract contained humic substances (Hsu and Singer, 2010) commonly present in NOM, (Pivokonsky et al., 2015). The apparent molecular weight (MW) distribution of the humic substances, determined by high-performance size exclusion chromatography (HPSEC), shows that the majority of molecules fall in the range of 4.3-15.8 kDa with the highest UV signal for MW of 10.8 kDa. This MW is in a higher range than would be expected for natural waters (Zhan et al., 2012) and therefore higher removal of DOC by coagulation would be expected in such synthetic water. The DOC of the initial concentration of the HS extract was 18 mg/L. It was diluted with Milli-Q water to reach a DOC concentration of approximately 7.5 mg/L, which represents a DOC content common for surface waters (Fabris et al., 2008). Based on the distribution of MW it would be anticipated that most of the DOC would be removable by coagulation. 2.1.2 Chemicals used All chemicals used – ferric sulphate, aluminium sulphate, potassium permanganate and sodium carbonate – were of analytical grade and the sodium hypochlorite was of laboratory grade. 2.1.3 Pre-ozonation To test the effect of ozone pre-oxidation on DOC removal by coagulation, a 1L sample was dosed with a preselected dose of ozone. Pre-ozonation of synthetic raw water was carried out in a glass column of inner diameter 55 mm and height 800 mm. Ozone was generated from dry oxygen gas by LIFEPOOL 1.0/OXCW apparatus (manufacturer Lifetech, the Czech Republic). The generated ozone concentration was 1-10mg O3/L. The pre-ozonation was operated in a batch mode and the ozone dose was controlled by time of bubbling the ozone/oxygen mixture through the water column. Doses of ozone were set to 1, 2 and 4 mg/L.
2.1.4 Jar Test Coagulation was performed in a 2L round container with a variable speed mixer. The container was filled with 1L of synthetic water and dosed with the coagulant under rapid mixing conditions of 200 rpm for 60 seconds. Slow mixing was conducted at 20 rpm for 30 minutes. The pH was adjusted by dosing sodium carbonate solution in the early stages of slow mixing. Water was allowed to settle for 30 minutes. Settled water was clear (except in KMnO4 pre-oxidised samples) and contained minimal visible suspended particles. 2.1.5 Centrifugation Settled water pH was measured as pH determines DOC equilibrium on floc (Kastl et al., 2004), then the sample was centrifuged (at 3000rpm for 5 minutes) to simulate media
filtration. The main driver for centrifugation was to protect the DOC analyser from particulate matter. Additionally, it would be expected that the separation of particulate matter by centrifugation may (as would media filtration) marginally reduce the DOC concentration.
2.1.6 Analysis Samples of produced water were analysed for: 2.1.6.1 DOC The DOC was measured using a total organic carbon analyser (TOC-VCPH) (Shimadzu, Japan) with an estimated accuracy of 5%. The measurements were carried in triplicate and eventually repeated if the variation was too high. The NOM removal was determined as DOC before and after the treatment. The DOC was measured by converting all organic compounds in water to CO2 by combustion and by detecting the CO 2 produced. By subtracting inorganic carbon present in water in the form of carbonates from the total amount of carbon, the organic carbon was calculated. Water was centrifuged after coagulation (at 3000rpm for 5 minutes) to remove particles. Samples of centrifuged water were used for DOC determination and chlorine decay tests. 2.1.6.2 Chlorine decay and THM formation The treated water sample was dosed with nominal 3.5 mg/L of chlorine. A chlorine stock solution was prepared by dilution of nominal 15% NaClO (Sigma-Aldrich) in Milli-Q water. A sample of Milli-Q water was also dosed with an equal amount of chlorine and used as a blank for chlorine decay without DOC. The chlorine concentration in samples was determined using the DPD (N, N diethyl-paraphenylenediamine) colorimetric method (4500CI G. DPD Colorimetric Method (Eaton et al., 2008)). Chlorine concentration was monitored over time. At the end of chlorine decay tests, samples were analysed for THMs (using GCMS, headspace gas sample). 2.1.6.3 Derivation of chlorine decay and THM formation model parameters Reviews (Fisher et al., 2012, Fisher et al., 2011b, Fisher et al., 2011a) concluded that a tworeactant (2R) chlorine decay model is the simplest and sufficiently accurate description of chlorine reaction with organic compounds in natural waters. The formulation of chlorine decay and THM formation models is based on a description of the following reactions:
Cl + F = Cl- + THMs
Eq. 4
Cl + S = Cl- + THMs
Eq. 5
where: F is the fast reacting component S is the slow reacting component and THMs are trihalomethanes (CHCl3, CHBrCl2, CHBr2Cl and CHBr3).
The reactions described by Eq. 4 and Eq. 5 are of first order with respect to both chlorine and organic compounds so that the reaction rates can be described as follows:
dcCl k F cCl cF k S cCl cS k z dt
dcF k F cCl cF dt dcS kS cCl cS dt dcTHM dc yTHM (- Cl ) dt dt
Eq. 6
Eq. 7
Eq. 8
Eq. 9
where: kF and kS is the reaction rate constant for a given component [(mg/L) -1h-1] kz is the zero rate reaction rate constant for chlorine decomposition in pure water [mg/L/h] c is the concentration of a given component [mg/L[ yTHM is yield for THM [ mg THM/ mg Cl reacted] and t is time [h]. The concentrations of fast and slow reacting compounds can be interpreted as the maximum concentration of chlorine which can be consumed by organic compounds. Both compounds react with chlorine in parallel. Initially, the chlorine decay rate is dominated by the fast reacting compound. Once the fast reacting compound is consumed (usually within the first few hours) the chlorine decay rate is determined by reaction of the residual chlorine with the slow reacting compound (Jabari Kohpaei and Sathasivan, 2011). Water treatment can remove and, by pre-oxidation, modify various components of DOC. That explains the reduction of fast and slow compounds and also changes in selectivity of THMs (production of THMs per mg of reacted chlorine). As shown in Eq. 9, the formation of THMs is directly proportional to the amount of reacted chlorine. For this reason, it is not recommended to pre-chlorinate raw water. This practice is common, as pre-chlorination provides the benefit in the reduction of turbidity (Xie et al., 2016), but it invariably results in the production of THMs, which are often not removed by subsequent treatment processes and hence contribute to the final concentration of THMs in the distribution system. The modelling approach described above quantifies formation of THMs for any given dose of chlorine and position in the distribution system. Such an approach is superior to a THM
formation potential, which only determines the maximum concentration of THMs generated at an excessive chlorine dose and long contact time. It could also be noted that THM formation potential can be derived from the model (on the basis of the initial concentration of fast and slow reacting compounds and the THM yield). The described chlorine decay model was fitted to the derived data using AQUASIM software (Reichert et al., 1995, Fisher et al., 2011a). The software solves the differential equations Eq. 6 to Eq. 9 for an estimated value of the model parameters. It varies the estimated parameters until the best possible fit with the experimental data is achieved. Such a model can be used to simulate chlorine decay and THM formation tests for any initial chlorine dose and at any reaction time.
3 Results The results of DOC removal and the residual of metal coagulant after settling and centrifugation are shown in Table 1. The residual metal concentration suggests that most of the metal had been removed by settling and centrifugation. The implications of DOC results are discussed below. DOC measurements with a zero coagulant dose represent synthetic water without any treatment. The average DOC of all four prepared water is 7.78 mg/L and it is within ±2% of for all samples. This represents good reproducibility of preparation of water sample and DOC measurement. Table 1 Residual DOC and metal concentrations after coagulation
3.1 DOC results From Table 1 it is apparent that practically all coagulation conditions (except 35 mg/L dose of ferric sulphate and a sample with permanganate pre-oxidation) achieved exceptionally high DOC removal. DOC was approximately reduced from 7.8 mg/L to 1 mg/L which represents a removal of 87%. The high DOC removal is consistently reproducible which suggests that the used model water contained NOM fractions which were easier to remove than a typical natural source water where the highest reported removal is around 80%(Volk et al., 2000). High DOC removal in the tested water can be explained by the greater proportion of high molecular weight NOM (>4.3 kDa) fraction (Pivokonsky et al., 2015) than commonly encountered in surface waters (Zhan et al., 2012, Nissinen et al., 2001). DOC removal in all jar tests is generally high, and based on DOC results only, it is impossible to exactly rank the produced water quality as DOC is subjected to the experimental error and some remaining DOC may have been pre-reacted with an oxidant. A positive finding is that the chlorine decay results are sensitive enough to clearly differentiate the produced water quality. Pre-ozonated coagulation produced the lowest DOC concentration, using only 45 mg/L of ferric sulphate as a coagulant. The lowest DOC (0.64 mg/L) was also achieved with the lowest ozone dose (1 mg/L). It is likely that there is a local optimum in DOC removal for ozone pre-oxidation dose. This optimal dose can be between 0 and 2 mg/L of ozone.
Experiments with permanganate pre-oxidation did not fulfil expectations. Coagulation of permanganate treated water produced fine, poorly settling floc with unacceptably high concentrations of soluble manganese. Usually a low (<1 mg/L) dose of permanganate is applied in water treatment plants to oxidise manganese, but here the permanganate dose was much higher (> 3.0 mg/L), possibly the reason for the poor result.
3.2 Chlorine decay results The results of chlorine decay for samples coagulated with increasing doses of ferric sulphate are shown in Figure 2. It can be seen that coagulation with ferric sulphate dramatically reduces the chlorine decay with the increasing coagulant dose at pH around 5.5. Chlorination of raw water (DOC=7.8 mg/L) resulted in rapid chlorine decay, with the concentration dropping to near zero within 24h. The chlorine dose was repeated (2.7 mg/L), but even the second dose did not lead to a lasting chlorine concentration. Apparently, the chlorine demand of raw water was far too high for comparison with the same chlorine decay procedure used for coagulated samples. In other samples, a dramatic change in chlorine decay rates of raw and coagulated samples is consistent with high DOC removal efficiencies. The chlorine decay curves are more consistent with what could be expected from coagulation with increasing coagulant dose. While the residual DOC fluctuates with the increasing dose of coagulant (55 mg/L of ferric sulphate resulted in DOC of 0.95 mg/L and 80 mg/L dose of ferric sulphate produced DOC of 1.12 mg/L), the data for chlorine decay clearly indicate a consistent improvement in chlorine stability with the increasing coagulant dose. These results are against the traditional use of DOC as a surrogate for chlorine stability; i.e., the DOC values did not rank the water samples in the same way as the chlorine decay profiles (Table 3). Given the quick response of DOC measurements, DOC could be a good qualitative surrogate for chlorine decay if DOC is removed by physical means. If oxidants are used, they may change the nature of the remaining DOC and a direct comparison is no longer valid. DOC indicates the relative status in the treatment, but it is difficult to assess how such water would perform in a distribution system or, in other words, translate DOC to a chlorine decay curve.
Figure 2 Chlorine decay in raw water (and after re-chlorination) and in water treated with various doses of ferric sulphate FS (35, 55 and 80 mg/L at pH 5.5). Lines connect consecutive experimental points.
Chlorine decay curves from Figure 2 can be used to derive parameters of a chlorine decay model. The model parameters characterise a given sample of treated water and can be used to simulate chlorine decay; for example, chlorine concentration profiles in a distribution system can be predicted for various initial chlorine doses and consequently the adequacy of different water treatments(Fisher et al., 2016). The derivation of chlorine decay model parameters has the additional advantage of quantifying decay so that differently treated waters can be compared on a basis directly related to the initial “fast” and longer-term “slow” chlorine decay. As the rate of chlorine decay cannot be calculated based on water composition parameters such as DOC, this approach is the only feasible way to characterise chlorine decay in a specific water sample.
Figure 3 Chlorine decay experimental data (markers) and chlorine model (lines), FS ferric sulphate, AS aluminium sulphate, Oz ozone and PP potassium permanganate, number indicates the dose (details for each experimental condition are shown in Table 1)
It can be seen (Figure 3) that the chlorine decay model (Fisher et al., 2011a) describes chlorine profiles with an acceptable accuracy, from the initial fast decay in the first few hours to the slow decay over hundreds of hours. The most chlorine-stable water is produced with a pre-ozone dose of 2 mg/L and 45 mg/L ferric sulphate coagulation dose. This is understandable as this treatment option resulted in a lower DOC concentration than in other compared curves and ozone pre-oxidised chlorine-reactive sites on DOC molecules. The second most chlorine-stable water was produced by coagulation with 80 mg/L of ferric sulphate. Ferric salts at low pH (5-6) in higher doses are recognised as suitable for deep removal of DOC and excellent stabilisation of chlorine. The use of permanganate preoxidation followed by coagulation with 45 mg/L of ferric sulphate partially compensates for the reduction in ferric dose from 80 mg/L. Alum-treated water was the least stable even though DOC was not significantly higher. The derived chlorine decay parameters for water treated under different conditions are listed in Table 2.
Table 2 Parameters of chlorine decay model (Fast and Slow initial concentration of chlorine reactive compounds) for waters coagulated under various conditions
Concentrations of fast and slow reacting compounds for treated waters are shown in Figure 4. It can be seen that the variation in chlorine reactive compounds is much more discerning than values of DOC (Table 2 and Table 3). The THM yield is low on average. This fact may be attributed to the tested synthetic water which had a low bromide concentration (<0.020mg/L). Most variations in the THM yield can be explained by the changing DOC composition by coagulation. Increasing the dose of ferric sulphate increases THMs yield (FS 25, FS55 and FS80) consistently. This could be possible if organic compounds with predominantly chlorine reducing properties are removed preferentially to compounds which form chlorinated products. Amongst pre-ozonated samples, the highest yield was recorded in a sample with the highest chlorine stability. Interestingly, all THM yields of ozonated waters were low. A possibility may be that the maximum reduction of chlorine reactive compounds does not coincide with the proportional removal of THMs producing compounds. In principle, each treatment method can have different affinity for reduction of sites for oxidation and for halogenation.
Figure 4 Initial concentrations of Fast and Slow chlorine decay components (FS – Ferric sulphate, O3 – ozone; each test followed by coagulant dose in mg/L)
Figure 4 indicates that there is a significant difference in total amount of chlorine-reactive compounds in differently treated water samples. This comparison presents a straightforward and obvious indicator of chlorine stability. The best chlorine stability based on Figure 4 was for O3 2 FS45 mg/L, which is the water treated with 2 mg/L of pre-ozonation and coagulation with 45 mg/L of ferric sulphate. A comparison of DOC and chlorine-reactive compounds is shown in Table 3. Table 3 Comparison of concentration of Chlorine reactive compounds and DOC
Table 3 shows that DOC and chlorine-reactive compounds do not provide parallel conclusions about chlorine stability of the produced water. Excluding the lowest quality water (FS 35), DOC ranges from 0.64 to 1.27 mg/L – a ratio of approximately 2, while chlorinereactive compounds range from 3.79 to 17.27 mg/L – a ratio of 4.5. This suggests that for a similar relative error of the measurement, the conclusion from chlorine-reactive compounds is likely to be more reliable than DOC measurements.
4 Discussion DOC removals are very high considering that the commonly accepted range of DOC removal in natural water sources is up to 80%. High DOC removal may be due to the nature of DOC which may be easier to remove, as indicated by high molecular mass 4-15.8 kD of NOM derived from this source (Pivokonsky et al., 2015). In modelling DOC removal using a three-component model of nonpolar, polar and inert non-adsorbable compounds, Kastl et al. (2004) found that the inert non-adsorbable fraction in all tested waters was higher than 0.2, suggesting that the maximum DOC removal should approach 80%. In the case of humic substance, Humat Star80-Almedsa Agrochimic-Algeria both alum and ferric coagulant achieved practically complete removal of humic substances from water (Harfouchi et al., 2016). When it is not possible to directly predict an impact of DOC concentration on chlorine stability and disinfection by-products (DBP) formation, an empirical rule of the thumb is used to determine how far to increase the coagulant dose. For example, one such a rule requires that an increase of the coagulant dose by 10 mg/L results in the additional DOC removal by a minimum of 0.3 mg/L (Volk et al., 2000) to identify the point of diminishing return. US EPA uses prescriptive removal rates based on the raw water DOC concentration and alkalinity. The highest required DOC removal of 50% is specified for low alkalinity water with DOC >8 mg/L. A discussion of the application of these rules is presented by Kastl et al. (2016). In the absence of chlorine decay tests, the chemical oxygen demand of the treated water may be considered a better indicator of water quality than DOC. A comparison of DOC and chemical oxygen demand KMnO4 (CODKMnO4) removal with enhanced coagulation indicates that CODKMnO4 is removed more efficiently than DOC (Pivokonska et al., 2008). It was possible to remove 58% of DOC and 65% CODKMnO4 in one water and 41% DOC and 71% COD in another water sample. These results indicate that CODKMnO4, which is expected to be
closely related to the chlorine stability, is not removed at the same rate by coagulation as DOC is. Additionally, it may be argued that, while CODKMnO4 may be a better indicator of chlorine stability than DOC, it is still not a true representation of the chlorine stability which is the ultimate arbiter in the determination of process efficiency. CODKMnO4 has been widely used as a surrogate for chlorine stability before the introduction of automated DOC instrumentation. The reliability of DOC concentration as a chlorine decay predictor is seldom evaluated; therefore, this work also compares the use of DOC and chlorine decay and THM formation measurements. To improve the chlorine stability, pre-oxidation is an additional option to consume chlorinereactive molecules. It may be applied before, or as frequently is after flocculation/ sedimentation. It is often applied even after filtration when filtered water is, for example, ozonated and passed via biological activated carbon (BAC) to stabilise water to control biological regrowth in the distribution system. If pre-ozonation is followed by coagulation, it would be expected that pre-ozonation would oxidise functional groups amenable to chlorine oxidation. This action would also be expected to make organic molecules more hydrophilic and therefore less amenable to adsorption on ferric hydroxide or activated carbon surface. A typical application of ozonation after coagulation /sedimentation/ filtration would be consistent with this concept. The coagulation /sedimentation/ filtration would remove most of the DOC which can be removed by charge neutralisation and adsorption on metal hydroxide such as Fe(OH)3. The following ozonation could only react with those organics which are difficult to remove by coagulation and are susceptible to oxidation. This simple concept would suggest that post-ozonation would produce more chlorine-stable water using the same ferric and ozone dose. On the other hand, the experiments often show that pre-ozonation initially increases DOC removal in consequent coagulation and that higher pre-ozonation doses may lead to the reduction of DOC removal. Higher doses of ozone make DOC removal somewhat unpredictable (Agbaba et al., 2014). The explanation that ozone attacks double bonds and makes DOC molecules more hydrophilic sounds plausible, but the inference that this leads to better DOC removal by enhanced coagulation is not sufficiently supported, as more studies report better removal of hydrophobic DOC (Xie et al., 2016). The impact of preozonation on DOC removal usually shows an optimum at a relatively low ozone dose. An explanation of the improved DOC removal by enhanced coagulation in pre-ozonated water samples may be that forming additional organic acid groups on the surface of colloidal particles enhances the impact of charge neutralisation mechanisms which is more efficient in DOC removal than absorption on the ferric hydroxide surface (Xie et al., 2016). This hypothesis is consistent with the observation (Table 1) that, at higher pre-ozonation doses, DOC removal declines as larger DOC molecules are broken down to smaller and more hydrophilic molecules, which are less susceptible to both charge neutralisation by coagulants and adsorption on ferric hydroxide surface. Pre-ozonation before coagulation results in a higher reduction of THMs formation potential than coagulation alone (Chiang et al., 2009). This effect can be explained by increased DOC removal and pre-oxidation of chlorine-reactive compounds. An average THM yield (Table 2) is 0.020mg THMs/mg Cl with the minimum and maximum measured the yield of 0.012 and 0.031 mg/mg respectively. Under the chlorination test, all samples generated THMs below 0.06 mg/L, which is well below USEPA guideline limit of 0.08mg/L. Based on these results
the selection of a treatment process for this type of water would be driven by the required chlorine stability and operation cost.
4.1 Practical considerations Based on the conducted experiments and evaluations, pre-ozonation with 2 mg/L of ozone followed by coagulation with ferric sulphate (45mg/L) resulted in the most chlorine stable water. This treatment generated lower DOC and better chlorine stability than water treated with 80 mg/L of ferric sulphate. Assuming cost (US$ in 2016) of ozone at $10/kg and ferric sulphate cost at $0.8 /kg, then the chemical cost for the above two treatments would be: For 1 ML pre-ozonated water
45kg/ML*$0.8/kg + 2 kg/ML*$10/kg=36 + 20= $56/ML
For 1 ML 80mg/L ferric sulphate
80kg/ML*$0.8/kg = $64/ML
As shown by the chemical cost calculation, the use of ozone not only reduces the dose of the coagulant, and therefore the amount of sludge for disposal, but also results in lower cost of the chemicals and better chlorine stability in this case. Pre-ozonated water would benefit from a biologically active filter, either in the form of dual media or even more efficiently including GAC, as ozonation is known to generate assimilable organic carbon (AOC) (Świetlik et al., 2004, Ramseier et al., 2011). In this instance, the pre-ozonation dose is quite low (2 mg/L), but it is feasible that AOC formed by ozonation can support biofilm growth in the distribution system when chlorine disinfectant residual becomes low. A potential limitation of pre-ozonation may be bromate concentration generated by oxidation of bromide. Bromate has a limit of 0.01 mg/L WHO guideline or 0.02 mg/L (NHMRC, 2011). If not limited by bromate, the option with pre-ozonation appears attractive due to low coagulant dose, chemical cost, sludge production and production of the most chlorine-stable water. The implementation of the derived chlorine decay and THM formation models in a network model containing the EPANET MSX – Multi-Species eXtension – module (Shang et al., 2007) can predict chlorine and THM profiles in a specific distribution system for given hydraulic conditions. The ability to predict chlorine and THM profiles in a hypothetical distribution network for a treated water sample is a superior indicator of the treated water quality than any other, including DOC.
5 Conclusion This study searched for an efficient method of NOM removal by enhanced coagulation with ferric and aluminium salts, in combination with pre-oxidation (with ozone) or permanganate to achieve better chlorine stability and lower THM formation. The produced water quality was characterised by the measurements of DOC, chlorine decay and THM formation. The chlorine decay data were fitted to two-reactant (2R)(Fisher et al., 2012) model to estimate initial fast and slow reacting agents and their respective reaction rates. The treatment that included pre-ozonation with 2 mg O3/L and coagulation with 45 mg/L ferric sulphate resulted in the best chlorine stability and THMs below 0.06 mg/L. Coagulation with 80 mg/L ferric sulphate produced the second-best results. Aluminium sulphate treatment did not achieve chlorine stability of ferric sulphate treatment even at a higher dose (100 mg/L) and when the DOC concentrations were comparable.
DOC was confirmed to be a good qualitative indicator of the treatment process. However, the ranking of the treatment processes on the basis of DOC concentration does not identify the water with the highest chlorine stability. The discrepancy in DOC and chlorine decay results may be due to small differences in DOC relative to the measurement error and better differentiation of chlorine decay curves. The results obtained strongly suggest that in coagulation of medium-high DOC raw water (612 mg/L) with a ferric coagulant, better chlorine stability of the treated water can be achieved with a lower coagulant dose and an optimised pre-ozonation dose. The approach demonstrated in this paper is an efficient way to scan a broad range of potentially suitable coagulation-based water treatment processes for waters with a moderately high concentration of DOC.
6 Funding and potential conflict of interest This research was supported by The Australian Postgraduate Award scheme and did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors are not aware of any potential conflict of interest.
References AGBABA, J., TUBIĆ, A., DALMACIJA, B., WATSON, M., MOLNAR, J., RONČEVIĆ, S. & MALETIĆ, S. 2014. Investigation of the impact of ozone pretreatment and powdered activated carbon addition on the removal of natural organic matter by coagulation. Desalination and Water Treatment, 56, 912-920. BOND, T., HUANG, J., GRAHAM, N. J. & TEMPLETON, M. R. 2014. Examining the interrelationship between DOC, bromide and chlorine dose on DBP formation in drinking water--a case study. Sci Total Environ, 470-471, 469-79. BROWN, I. J., TZOULAKI, I., CANDEIAS, V. & ELLIOTT, P. 2009. Salt intakes around the world: implications for public health. Int J Epidemiol, 38, 791-813. CHIANG, P.-C., CHANG, E. E., CHANG, P.-C. & HUANG, C.-P. 2009. Effects of pre-ozonation on the removal of THM precursors by coagulation. Science of The Total Environment, 407, 57355742. CRITTENDEN, J. C., TRUSSELL, R. R., HAND, D. W., HOWE, K. J. & TCHOBANOGLOUS, G. 2012. MWH's water treatment: principles and design, John Wiley & Sons. DE LA RUBIA, A., RODRIGUEZ, M., LEON, V. M. & PRATS, D. 2008. Removal of natural organic matter and THM formation potential by ultra- and nanofiltration of surface water. Water Res, 42, 714-22. DUAN, J. & GREGORY, J. 2003. Coagulation by hydrolysing metal salts. Advances in Colloid and Interface Science, 100-102, 475-502. EATON, A., CLESCERI, L., RICE, E. & GREENBERG, A. 2008. Standard methods for the examination of water and wastewater. EVANS, C. D., MONTEITH, D. T. & COOPER, D. M. 2005. Long-term increases in surface water dissolved organic carbon: Observations, possible causes and environmental impacts. Environmental Pollution, 137, 55-71. FABRIS, R., CHOW, C. W., DRIKAS, M. & EIKEBROKK, B. 2008. Comparison of NOM character in selected Australian and Norwegian drinking waters. Water Res, 42, 4188-96.
FISHER, I., KASTL, G. & SATHASIVAN, A. 2011a. Evaluation of suitable chlorine bulk-decay models for water distribution systems. Water research, 45, 4896-4908. FISHER, I., KASTL, G. & SATHASIVAN, A. 2012. A suitable model of combined effects of temperature and initial condition on chlorine bulk decay in water distribution systems. Water research, 46, 3293-3303. FISHER, I., KASTL, G. & SATHASIVAN, A. 2016. A comprehensive bulk chlorine decay model for simulating residuals in water distribution systems. Urban Water Journal, 14, 361-368. FISHER, I., KASTL, G., SATHASIVAN, A. & JEGATHEESAN, V. 2011b. Suitability of chlorine bulk decay models for planning and management of water distribution systems. Critical reviews in environmental science and technology, 41, 1843-1882. GHADIMKHANI, A., TORABIAN, A. & MEHRABADI, A. 2006. Preozonation and prechlorination effect on TOC removal in surface water treatment. Pakistan Journal of Biological Science, 9, 708712. HARFOUCHI, H., HANK, D. & HELLAL, A. 2016. Response surface methodology for the elimination of humic substances from water by coagulation using powdered Saddled sea bream scale as coagulant-aid. Process Safety and Environmental Protection, 99, 216-226. HSU, S. & SINGER, P. C. 2010. Removal of bromide and natural organic matter by anion exchange. Water Research, 44, 2133-2140. JABARI KOHPAEI, A. & SATHASIVAN, A. 2011. Chlorine decay prediction in bulk water using the parallel second order model: An analytical solution development. Chemical Engineering Journal, 171, 232-241. KASTL, G. & FISHER, I. 1997. Predicting and Maintaining Drinking Water Quality in Distribution Systems. WATER-MELBOURNE THEN ARTARMON-, 24, 35-38. KASTL, G., SATHASIVAN, A. & FISHER, I. 2016. A selection framework for NOM removal process for drinking water treatment. Desalination and Water Treatment, 57, 7679-7689. KASTL, G., SATHASIVAN, A., FISHER, I. & VAN LEEUWEN, J. 2004. Modeling DOC removal by enhanced coagulation. Journal-American Water Works Association, 96, 79-89. KRISTIANA, I., JOLL, C. & HEITZ, A. 2011. Powdered activated carbon coupled with enhanced coagulation for natural organic matter removal and disinfection by-product control: Application in a Western Australian water treatment plant. Chemosphere, 83, 661-667. MATILAINEN, A., GJESSING, E. T., LAHTINEN, T., HED, L., BHATNAGAR, A. & SILLANPAA, M. 2011. An overview of the methods used in the characterisation of natural organic matter (NOM) in relation to drinking water treatment. Chemosphere, 83, 1431-42. MATILAINEN, A., VEPSALAINEN, M. & SILLANPAA, M. 2010. Natural organic matter removal by coagulation during drinking water treatment: a review. Adv Colloid Interface Sci, 159, 18997. NHMRC, N. 2011. Australian drinking water guidelines paper 6 national water quality management strategy. National Health and Medical Research Council, National Resource Management Ministerial Council, Commonwealth of Australia, Canberra. NISSINEN, T. K., MIETTINEN, I. T., MARTIKAINEN, P. J. & VARTIAINEN, T. 2001. Molecular size distribution of natural organic matter in raw and drinking waters. Chemosphere, 45, 865873. PIVOKONSKA, L., PIVOKONSKY, M. & TOMASKOVA, H. 2008. Optimization of NOM Removal during Water Treatment. Separation Science and Technology, 43, 1687-1700. PIVOKONSKY, M., NACERADSKA, J., BRABENEC, T., NOVOTNA, K., BARESOVA, M. & JANDA, V. 2015. The impact of interactions between algal organic matter and humic substances on coagulation. Water Res, 84, 278-85. RAMSEIER, M. K., PETER, A., TRABER, J. & VON GUNTEN, U. 2011. Formation of assimilable organic carbon during oxidation of natural waters with ozone, chlorine dioxide, chlorine, permanganate, and ferrate. Water Res, 45, 2002-10.
REICHERT, P., VON SCHULTHESS, R. & WILD, D. 1995. The use of AQUASIM for estimating parameters of activated sludge models. Water Science and Technology, 31, 135-147. SHANG, F., UBER, J. G. & ROSSMAN, L. A. 2007. Modeling reaction and transport of multiple species in water distribution systems. Environmental science & technology, 42, 808-814. ŚWIETLIK, J., DĄBROWSKA, A., RACZYK-STANISŁAWIAK, U. & NAWROCKI, J. 2004. Reactivity of natural organic matter fractions with chlorine dioxide and ozone. Water research, 38, 547558. USEPA 2001. Controlling Disinfection By-Products and Microbial Contaminants in Drinking Water. VOLK, C., BELL, K., IBRAHIM, E., VERGES, D., AMY, G. & LECHEVALLIER, M. 2000. Impact of enhanced and optimized coagulation on removal of organic matter and its biodegradable fraction in drinking water. Water Research, 34, 3247-3257. WORRALL, F. & BURT, T. P. 2009. Changes in DOC treatability: Indications of compositional changes in DOC trends. Journal of Hydrology, 366, 1-8. XIE, P., CHEN, Y., MA, J., ZHANG, X., ZOU, J. & WANG, Z. 2016. A mini review of preoxidation to improve coagulation. Chemosphere, 155, 550-63. ZHAN, W., SATHASIVAN, A., JOLL, C., WAI, G., HEITZ, A. & KRISTIANA, I. 2012. Impact of NOM character on copper adsorption by trace ferric hydroxide from iron corrosion in water supply system. Chemical Engineering Journal, 200–202, 122-132. ZHANG, Y., ZHAO, X., ZHANG, X. & PENG, S. 2015. A review of different drinking water treatments for natural organic matter removal. Water Science & Technology: Water Supply, 15, 442.
Figure 1 schematics of experimental design coagulation with pre-oxidation
4.5
Rechlorination
DOC 7.80, FS 0
4.0 3.5
DOC 2.39, FS 35 DOC 0.95, FS 55 DOC 1.12, FS 80 Blank
Cl (mg/L)
3.0 2.5 2.0 1.5 1.0 0.5 0.0 0
50
100
150
200
250
300
350
400
Time (h) Figure 2 Chlorine decay in raw water (and after re-chlorination) and in water treated with various doses of ferric sulphate FS (35, 55 and 80 mg/L at pH 5.5). Lines connect consecutive experimental points.
FS 80 FS 80 Oz 2mg/L Oz 2 FS 45 AS 100 AS 100 PP 7.0 FS4 5 PP 7.0 FS4 5
4.5 4
Free Cl (mg/L)
3.5 3
2.5 2 1.5 1 0.5
0 0
100
200
Time (h)
300
400
Cl demand as determined from initial slow and fast reacting compounds (mg/L)
Figure 3 Chlorine decay experimental data (markers) and chlorine model (lines), FS ferric sulphate, AS aluminium sulphate, Oz ozone and PP potassium permanganate, number indicates the dose (details for each experimental condition are shown in Table 1)
25 Initial fast mg/L
20 Initial slow mg/L
15
10
5
0
FS 35 FS 55 FS 80 Oz 1 Oz 2 Oz 4 AS 70 AS 80 AS 100 PP 7.0 FS45 FS45 FS 45 FS 45
Figure 4 Initial concentrations of Fast and Slow chlorine decay components (FS – Ferric sulphate, O3 – ozone; each test followed by coagulant dose in mg/L)
Table 1 Residual DOC and metal concentrations after coagulation
Coagulant +conditions (experiment code)
Coagulant Coagulation Dose pH [1] (mg/L)
DOC (mg/L)
Ferric sulphate (FS 0) Ferric sulphate (FS 35) Ferric sulphate (FS 55) Ferric sulphate (FS 80) Ozone pre-dose 0 mg/L Ferric sulphate (Oz 0 FS S0)
0 35 55 80 0
6.75 5.80 5.60 5.70 6.75
7.65 2.39 0.95 1.12 7.85
Residual metal Fe or Al or Mn (mg/L) n/a 0.68 0.38 0.10 n/a
Ozone pre-dose 1 mg/L Ferric sulphate (Oz 1 FS 45)
45
5..44
0.64
0.09
Ozone pre-dose 2 mg/L Ferric sulphate (Oz 2 FS 45)
45
5.47
0.78
0.12
Ozone pre-dose 4 mg/L Ferric sulphate (Oz 4 FS 45)
45
5.15
0.92
0.40
Aluminium sulphate (AS 0) 0 7.90 n/a Aluminium sulphate (AS 70) 70 5.66 1.04 0.033 Aluminium sulphate (AS 80) 80 5.75 0.99 0.076 Aluminium sulphate (AS 100) 100 5.71 0.92 0.025 Potassium permanganate-dose 0 0 7.72 n/a mg/L Ferric sulphate (PP 0 FS 0) Potassium permanganate-dose 45 5.50 High colour n/a 3.5 mg/L Ferric sulphate not (PP3.5FS45) determined Potassium permanganate dose 7 45 5.50 1.27 0.50 mg/L Ferric sulphate (PP 7.0 FS4 5) All entries in Table 1 with zero doses of coagulant represent untreated water. These values are reasonably close in the range of 7.65-7.90 mg/L which indicates good reproducibility in the preparation of raw water and DOC analysis. The coagulant dose is expressed as Fe2(SO4)3 and Al2(SO4)3.
Table 2 Parameters of chlorine decay model (Fast and Slow initial concentration of chlorine reactive compounds) for waters coagulated under various conditions
Initial THM Coagulant Dose DOC % DOC Initial fast slow yield +conditions (mg/L) (mg/L) removal (mg/L) (mg/L) (mg/mg) FS 35 35 2.39 68.8 1.20 19.97 0.013 FS 55 55 0.95 87.6 0.47 9.01 0.023 FS 80 80 1.12 85.3 0.01 4.79 0.031 Oz 1 FS 45 45 0.64 91.8 0.52 9.60 0.012 Oz 2 FS 45 45 0.78 90.0 0.37 3.42 0.022 Oz 4 FS 45 45 0.92 88.3 0.27 11.85 0.017 AS 70 70 1.04 86.8 0.98 16.29 0.015 AS 80 80 0.99 87.4 1.24 9.88 0.022 AS 100 100 0.92 88.4 0.68 13.04 0.021 PP 7.0 FS 45 45 1.27 83.6 0.93 6.14 n/a Initial fast and slow chlorine reacting compounds in Table 2 are hypothetical concentrations of chlorine-consuming compounds present in variously treated waters. THM yield is the amount of THMs produced by the reaction of 1 mg of chlorine. Coagulant – FS is ferric sulphate, AS – aluminium sulphate, Oz – ozone and PP – potassium permanganate. The number after indicates the dose in mg/L.
Table 3 Comparison of concentration of Chlorine reactive compounds and DOC
Coagulant +conditions
Treated water DOC [mg/L]
O3 2 FS45
0.78
Treated water Chlorine reactive Fast + Slow compounds [mg/L] 3.79
FS80
1.12
4.80
KMnO4 FS45
1.27
7.07
FS55 O3 1 FS45
0.95 0.64
9.48 10.13
Alum80 O3 4 FS 45
0.99 0.92
11.12 12.11
Alum100 Alum70
0.92 1.04
13.72 17.27
FS 35
2.39
21.17