Trihalomethane formation in Irish water supplies
02 May 2013
Authors: Professor Tom Casey, Aquavarra Research Limited; Hugh Kerr, Donegal County Council; and Pat Kearney, UCD School of Civil, Structural and Environmental Engineering
Although chlorine has been in use for drinking-water disinfection for over 100 years (USEPA, 2000), its reaction with NOM, leading to the generation of undesirable haloform byproducts, was not established until 1974, when Rook published the findings of an investigation at the Rotterdam Water Utility. This showed that haloforms formed at significant levels immediately following chlorination.
Later that year, Bellar et al reported significant levels of organochlorides in some US drinking waters. These findings prompted the US Environmental Protection Agency (USEPA) to initiate a national survey to determine to what extent organchlorides were present in waters nationwide – and especially the effect of chlorination on the formation of chlorinated organics.
The results of the survey (Symon et al) indicated that four trihalomethanes (THMs) – chloroform, bromodichloromethane, dibromochloromethane and bromoform – were present in a large number of the 80 city water supplies surveyed.
Bromide ions, the precursors of brominated THMs, are to be found in natural waters at low concentrations. Bromide is a conservative constituent in natural water and takes part in very few chemical reactions (Hutton & Chung, 1994). In the process of chlorination, bromide ions are oxidised to hypobromous acid (HOBr), which reacts more readily with organic matter than chlorine to form brominated THMs. It is noteworthy that water sources in coastal areas that are influenced by seawater intrusion may have elevated bromide levels.
Because of the suspected causative link between these chlorination-disinfection byproducts and various cancer forms, mandatory limit values have been set for their concentration in drinking water. The current EU and Irish drinking water limit value for total THM (TTHM) is 100μg/l. The TTHM drinking-water limit concentration in the United States is 80μg/l.
The complexity of THM precursor species makes it impossible to measure their exact concentration in natural water. Not all NOM will react with free chlorine to form THM. The general term ‘DBP precursor’ is used to represent the culprit fraction. The most widely used surrogate parameter is total organic carbon (TOC).
TOC is a collective measure of organic matter in water. However, it provides no information on the composition and distribution of the wide array of organic constituents. Most groundwaters contain less than 2mg/l TOC. The TOC of raw water grab samples collected from 12 Irish WTWs, using surface water sources, was found (Chua, 1996) to vary in the range 3.5-14.0mg/l.
Another parameter frequently used is ultraviolet absorbance (UVA, 1/cm) at 254nm. Organic compounds that are aromatic or that have conjugated double bonds absorb light in the ultraviolet region. UVA is considered a good potential surrogate parameter because humic substances contain aromatic moieties. Specific ultraviolet absorbance i.e. UVA/TOC (SUVA, l/cm.mg) is also used in the characterisation of NOM in water with reference to its potential for THM production in reaction with chlorine.
THM FORMATION POTENTIAL
The THM formation potential (THMFP) of water is measured in a standard test procedure, which allows ample time and chlorine availability for the reaction between chlorine and NOM. In the results reported here (Chua, 1996), the THMFP test involved seven days’ incubation at 25 oC, pH of 7.0, an applied chlorine dose of circa three times the TOC concentration, resulting in an RFC at the end of the seven-day incubation period in the range 2-5mg/l.
The test environmental conditions are important, as THM formation is observed to increase with temperature, pH and available free chlorine. Fig 1 presents the results of THMFP measurements for a set of raw and treated water grab samples taken from Irish WTWs. The plotted data illustrate the extent of THMFP removal by alum coagulation/filtration without pH adjustment for optimal TOC removal.
Since the data plotted in Fig 1 relate to grab samples, a wide value range is to be expected, reflecting both the temporal variability of the individual water sources as well as variability between sources. The measured raw water THMFP varied within the range 234-1058 μg/l, while the measured treated water THMFP varied within the range 87-295μg/l. Only two of the 13 treated waters had a THMFP ≤ 100μg/l while four had a THMFP ≥200 μg/l. The extent of THMFP removal by chemical coagulation/filtration was found to vary within the range 17-80%.
Sample LT, which had the highest raw water THMFP at 1058μg/l, may be regarded as typical of a soft, highly coloured, peaty surface water (colour: 170 HU; TOC 13.9 mg/l). Sample LK, which had the second highest raw water THMFP at 641μg/l, may be regarded as typical of a lowland river reach with a hard alkaline water of moderate colour and high TOC (colour: 50 HU; TOC 8.9 mg/l).
Sample CI, which had the lowest raw water THMFP, may be regarded as typical of a good-quality upland impounded water of low colour (colour: 29 HU, TOC: 3.6mg/l).
As THM formation is a consequence of the reaction between chlorine and NOM in water, it is of interest to examine the correlation between THMFP and TOC and THMFP and UVA, the surrogate parameters commonly used as indices of the predisposition of waters to form THMs.
Fig 2 plots a correlation of THMFP and TOC for a set of grab samples taken from 14 Irish WTWs. Fig 3 presents a correlation of THMFP and UVA for the same sample set. While THMFP exhibits a positive correlation with TOC, it is clear that there is a significant variation in dependency, particularly in the commonly prevailing surface water TOC range of 3-6mg/l.
This is not surprising, as only a relatively small fraction of the TOC reacts with chlorine to form THM species, as demonstrated in Fig 4, which plots the fraction of TOC incorporated in THM species under THMFP testing conditions. The carbon incorporation in TTHM was found to vary from 0.3% to 0.6% of TOC for treated surface waters and from about 0.4% to 1% of TOC for raw surface waters. Hence, it is not surprising that TOC on its own is not a very precise predictive parameter for THMFP. The data plotted in Fig 3 indicate a reasonably linear correlation between UVA and THMFP for the raw waters examined but a more scattered relation for the treated waters.
Potential THM species comprise chloroform (CHCl3), bromodichloromethane (CHBrCl2), chlorodibromomethane (CHBr2Cl) and bromoform (CHBr3). The measured individual concentrations of these organchlorine species in a set of grab samples taken from Irish WTWs are plotted in Fig 5. Chloroform was found to be the dominant THM species formed in 11 of the 14 test samples, accounting for some 60% to 80% of the TTHM expressed on a μg/l concentration basis. Bromodichloromethane was found in all samples in concentrations ranging from 2-21μg/l. Chlorodibromomethane was detected in nine of the 14 samples with a significant level in samples CL and WD. Bromoform was not detected in any of the samples.
THM FORMATION KINETICS
The rate of THM formation following chlorination typically exhibits a decreasing growth rate profile similar to that of primary chlorine demand, albeit at a somewhat slower rate. A typical example is illustrated in Fig 6, which plots both TTHM formation and chlorine decay over time in a grab sample taken from WTW DA. The sample was found to have an RFC of 0.9 mg/l after a two-hour contact period; as shown in Fig 5, the RFC effectively decreased to zero after a contact period of circa nine hours. The TTHM concentration grew rapidly initially, reaching half its maximum value in about 2.5 hours while further TTHM development was found to negligible after 24 hours.
The typical TTHM decreasing growth rate pattern illustrated in Fig 6 can be modeled as a hyperbolic function having the form:
where TTHMt = TTHM concentration at time t, µg/l
TTHMmax = maximum TTHM concentration, µg/l
t = reaction time (h)
t50 = time taken to reach 50% of TTHMmax (h)
The hyperbolic model growth rate is defined by two parameters, namely TTHMmax and t50. In the example illustrated in Fig 6, the TTHMmax is 95.7 µg/l and the t50 value is 2.52h. The values of these parameters are not constants for a given treated water, increasing with increasing chlorine dose.
The rate of THM formation is influenced by both temperature and pH. As with most chemical reactions, THM formation is found to increase with increasing temperature. Chua (1996) found that THMFP formation increased approximately linearly with temperature, resulting in an increase in the range 50-60% per 10 oC rise in temperature. Engerholm and Amy (1983) reported the increase in THM formation per 10 oC rise in temperature to be within the range 25-50%. As the Irish surface water seasonal temperature variation is at least 10 oC, a significant seasonal variation in THM formation is to be expected with higher values in summer than in winter. Seasonal variation in NOM also contributes to the variation in THM formation.
THM formation has been found to increase with increase in pH (Kim et al, 2002; Oliver & Lawrence, 1979). Chowdhury & Champagne (2008) found that THM production increased by about 40% when the pH was changed from 6.5 to 8.0 for one source and increased by about 25% when the pH was changed from 6.2 to 7.4 for a second source.
INFLUENCE OF CHLORINE DEMAND
It has been found (Clarke, 1998; Bocelli et al, 2003) that TTHM production in a chlorinated drinking water at an RFC contact time t can be linearly related to the chlorine demand exerted at that time:
TTHMt = F.Dt + M
where Dt is the total chlorine demand at time t, F is a parameter relating THM formation to chlorine demand, and M is the TTHM present at t = 0 (typically M = 0 for primary chlorination, and M > 0 for booster chlorination).
Fig 7 plots TTHM as a function of chlorine dose for water samples taken from 10 Irish WTWs (Chua, 1996). The raw water grab samples (all of surface water origin) were treated in the laboratory by chemical coagulation/clarification/filtration for colour and turbidity removal and were chlorinated at the dose level required to generate an RFC of 0.5 mg/l at 25 oC. The TTHM concentration was measured after 48h contact. The RFC decay time varied from 4 to 10h; hence the chlorine dose may also be regarded as the chlorine demand in this case. The parameter F varied within the range 22-33 µg/mg for the 10 waters tested, with a mean value of 26 µg/mg (i.e. µg TTHM per mg chlorine demand). Bocelli et al (2003) reported F-values in the range 29-59 µg/mg for a similar set of treated water test samples. Thus, while TTHM may be linearly related to chlorine demand for a given water source, there is clearly a considerable variation in this correlation between waters derived from different catchments.
The practice of maintaining an RFC of at least 0.1mg/l throughout water distribution systems means that drinking water is in contact with free chlorine from the time of primary chlorination to the time of consumption. This means that chlorine demand, and hence also THM formation, increases with the water age, as measured from the time of primary chlorination. Thus, the risk of THM non-compliance is greatest for waters that have a high NOM/chlorine reactivity combined with a high age. The NOM/chlorine reactivity of treated surface waters can vary over a wide range as illustrated by the survey results presented in Table 2, which found that the maximum 2h chlorine demand was about three times the minimum 2h chlorine demand. While a high two-hour chlorine demand is a good indicator of longer-term chlorine demand, this is not always the case. A better indicator is the differential between the two-hour demand and the 30min. demand, which for the set of samples in Table 2 varied in the range 0.14 mg/l for sample DEE to a high of 0.51 for sample RD.
REDUCTION OF CHLORINE DEMAND
Surface waters are conventionally treated by a process combination of chemical coagulation/clarification/sand filtration to produce a treated water that meets drinking water standards as measured by control parameters, notably colour, turbidity and residual coagulant chemical. Where the raw water is of exceptional quality, such as impounded waters of low colour and turbidity, slow sand filtration (SSF) may provide an adequate treatment to deliver drinking water quality in respect of colour and turbidity levels.
Chemical coagulation based on aluminium sulphate (alum), followed by clarification and rapid gravity filtration is by far the most widely used process combination for the production of drinking water from surface water sources in Ireland. Its effectiveness in reduction of THMFP (hence, also chlorine demand) has been illustrated in Fig 6, which presents the results of THMFP measurements for a set of raw water grab samples taken from 13 Irish WTWs. As is evident from examination of Fig 6, the removal percentage in THMFP across the sample set was found to be quite variable, falling within the range 40-80% for most samples. Perhaps of more significance is the residual THMFP, which was found to be highest for the hard alkaline raw waters, such as samples DA and LK. This reflects the fact that the effectiveness of alum coagulation in chlorine demand reduction is generally found to be pH-sensitive. Hence, process performance with low and medium alkalinity waters, where alum addition results in coagulation at an acid pH in the range 6-7, is more effective than with hard alkaline waters where the alkalinity is sufficient to maintain a coagulation pH in the range 7-8. The influence of coagulation pH on chlorine demand and THM production for raw water sample LK is illustrated in Fig 13. The upper curve relates to coagulation at an alum dose of 7 mg Al3+/l and a coagulation pH of 7.22. The lower curve relates to coagulation at an alum dose of 4mg Al3+/l and a pH of 5.51. The applied chlorine dose in both cases was that required to generate a residual of 0.5 mg/l at 30 minute contact time. Since the RFC reduced to zero in less than 6h the applied dose is also the chlorine demand in this case. The results show that the reduction in coagulation pH effected a ca. 20% reduction in chlorine demand and ca. 30% reduction in TTHM. It is also noteworthy that reduction in the coagulation pH resulted in a significant reduction in the required alum dose.
Thus, for hard alkaline surface waters, such as are found in the lower reaches of a number of Irish rivers, optimization of the coagulation pH to maximise TOC and chlorine demand removal is essential to avoid excessive THM production in water supplies derived from these sources. The THM reduction impact of improved chemical coagulation is confined to the CHCl3 species, as chemical coagulation does not effect removal of bromide ion.
SLOW SAND FILTRATION
Slow sand filtration (SSF) has traditionally been used to treat surface waters of a high quality that may be marginally outside drinking water standards in terms of colour and turbidity. Such sources are typically upland artificial impoundments or natural lakes derived from catchments that are free of peat cover.
The SSF process effects a marginal removal of colour, turbidity, TOC and chlorine demand. Chua (1996) reported a THMFP removal in the range 14-21% for an SSF pilot plant treating an impounded raw water that already was being successfully treated by a plant-scale SSF process over a long period. Collins & Eighmy (1988) reported a similar level of SSF performance with a THMFP removal in the range 9-27% and a TOC removal in the range 13-33%.
While it is probably the case that raw waters, that meet the quality requirements for treatment by SSF, are unlikely to have high TOC and chlorine demand levels, nevertheless the related issues of chlorine demand and THM formation should be included in the set of criteria used for assessment of SSF suitability in a particular application.
NOM removal by granulated activated carbon (GAC) generally requires a minimum empty bed contact time (EBCT=Vol/Q) of 10-20 minutes to achieve significant removal (Cummings & Summers, 1994). The effectiveness of a GAC contactor decreases with increasing operating duration. GAC adsorption capacity could be exhausted in as short as 41 days or as long as 182 days (Jacangelo et al, 1995). For example, a summary of the precursor removal performance of a GAC pilot plant, treating SSF filtrate (Chua, 1996) is presented in Table 3. The pilot plant was operated over a 260d period at an EBCT of 20 minutes.
Table 3: GAC contactor performance
While GAC adsorption is an excellent process for TOC/THMFP removal, the requirement for its regeneration/replacement is a major obstacle to its adoption as a viable process in water treatment practice.
The test results and analysis presented above point to the fact that chlorine demand is a key process variable in the production of drinking water from surface water sources, affecting both the disinfection process and the potential for undesirable THM formation. It is obviously an hygienic imperative that water leaving a WTW should be fully disinfected. In this regard, the primary chlorine demand of a treated water is a key design variable as it determines the minimum required chlorine dose that would provide the necessary CT for disinfection. For current discussion purposes, the primary chlorine demand is taken to mean the immediate demand as reflected in the free chlorine decay rate over a two-hour period following chlorination.
It has been shown by an analysis of primary chlorine demand test results for a range of treated water grab samples taken from Irish WTWs that the primary demand could be reliably quantified by a simple power function of contact time. The tested sample set was representative of Irish surface waters in water treatment terms in that it included surface waters of varying colour, turbidity and alkalinity/hardness. All waters tested exhibited the same general primary chlorine demand/time profile, characterized by an initial rapid decrease in RFC over the initial few minutes of contact, followed by a decreasing rate that could be accurately represented as a simple power function of time. While all samples exhibited the same general chlorine demand/time profile, the two-hour chlorine demand varied in magnitude by a factor of about 3:1 between the treated water of lowest demand and the treated water of highest demand.
The primary chlorine demand of treated surface waters was found to be partially influenced by the prevailing free chlorine concentration but not to the extent inferred by a first-order reaction rate, as commonly assumed in chlorine decay rate modeling. However, it is important that this partial dependence on chlorine concentration is taken into account in measuring primary chlorine demand by selecting a chlorine dose ≥ the disinfection dose when experimentally determining the primary demand profile. Indeed, it would be useful to relate primary chlorine demand to a standard chlorine dose of 2mg/l (disinfection dose unlikely to exceed 2mg/l) and a standard temperature to provide an additional useful comparative index of treated water quality.
Water temperature is a significant variable in chlorine disinfection process design. Irish surface waters experience a seasonal temperature fluctuation of up to about 15 oC. The required disinfection CT-value increases as the water temperature decreases. In water treatment practice, the chlorination process is normally controlled by regulating the applied dose to achieve a set RFC in the outflow from the contact tank. However a fixed outflow RFC would deliver a reducing CT-value as the water temperature drops. Hence, it is important that the target outflow residual is modified as the water temperature changes.
The importance of having a properly designed contact tank at all WTWs cannot be overstated as it plays a key role in delivering safe drinking water. The key characteristics of the ideal contact tank are (a) fixed volume that provides a retention time RT= V/Q, and (b) near-to plug-flow hydraulic behavior, thus minimising short-circuiting. A fixed volume is easily provided by a fixed level overflow outlet weir or similar control to maintain a fixed water level. There is flexibility in selecting a design retention time, which would generally be in the range 0.5-two hours.
As outlined in the text, knowledge of the empirically determined chlorine demand coefficients kD and n enables calculation of the required chlorine dose and corresponding contact tank outflow RFC to achieve a selected design CT value. The chlorine dose required to achieve a target CT is inversely related to the retention time as is the outflow residual. Thus, a larger chlorine dose is required where the contact time is short. In general, this may not prove to be wasteful of chlorine since a residual has to be maintained in the distribution system.
The foregoing presentation has outlined the factors that influence THM formation in water supplies. THMFP data for a range of surface water types indicate that precursor organics are only partially removed by conventional water treatment process technology. The chlorine demand/time profile of treated waters provides a good index of the effectiveness of the applied treatment, particularly in relation to NOM removal. The requirement of maintaining an RFC of at least 0.1mg/l in water distribution systems means that the total chlorine dose must marginally exceed the demand of the water over its lifetime. Thus, the required total chlorine dose and THM generation both increase with water age.
In water supplies where chlorination is confined to a single upfront chlorine dose, the requirement to maintain an RFC of at least 0.1mg/l in the distribution system rather than disinfection considerations may be the determinant of the dose magnitude. Where the chlorine dose applied at the WTW exceeds that required for disinfection purposes, the provision of a purpose-designed contact tank may be deemed to be not necessary. In such circumstances, service reservoirs serve a dual purpose of contact tank and water demand buffer. In general, this is not best process design practice as service reservoirs tend to have a variable volume and uncertain residence time distribution, resulting in a low effective CT yield per unit of chlorine dose.
THM formation is primarily minimized by the production of treated water having a low chlorine demand, following by a chlorination regime designed to minimise the total chlorine dose required for both disinfection and maintenance of a residual in the distribution system. For example, a chlorination regime relying on a single large dose inevitably results in a greater system chlorine demand than a regime where an initial disinfection dose is followed one or more booster doses. This arises from the fact that the chlorine decay rate is partly concentration-dependent. Maintaining unnecessarily high RFC levels in service reservoirs can add greatly to system chlorine demand and hence THM production.
- Chlorine demand has been shown to be a key parameter in disinfection process design and in the formation of THMs in water supply systems.
- Quantification of the initial short-term chlorine demand (contact time ≤2h) is essential for reliable design of primary disinfection processes. The initial 2h chlorine demand for a set of treated Irish surface water grab samples was found to vary in the range 0.6-1.9mg/l.
- It has been shown that the initial short-term chlorine demand of treated Irish surface waters could be accurately represented by a simple power function of contact time: Dt = kD.tn where Dt is the chlorine demand (mg/l) at time t (min), kD and n are empirical coefficients. The coefficient kD corresponds to the fitted model 1-minute chlorine demand value.
- The initial short-term chlorine demand was found to be partially chlorine concentration-dependent, but not to the extent that the decay rate could be represented as a first-order reaction with respect to RFC.
- As demonstrated in the text, the use of the derived model equation for short-term chlorine demand enables an accurate calculation of the CT-value generated in a contact tank, based on either the outflow RFC or the applied chlorine dose.
- It is highly desirable that all WTWs have a purpose-designed chlorine contact tank, having a fixed water volume, hydraulically configured to eliminate short-circuiting and sized to deliver the required CT to achieve primary disinfection.
- As with all chemical reactions, chlorine demand has been found to be influenced by water temperature as is the CT value required for primary disinfection. Since Irish surface waters may undergo a seasonal temperature fluctuation of up to 15oC, treated water temperature is an important primary chlorination process operational parameter, requiring a higher contact tank outflow RFC in winter than in summer.
- THMs are generated in all treated surface waters to which chlorine has been added, the resulting TTHM being directly related to the total chlorine added, which in turn is determined by the chlorine demand of the water.
- The total chlorine demand of treated surface water is related to its residual NOM content, which is conventionally measured as TOC. Only a very small fraction of TOC is incorporated into THMs. Hence, while THMFP is found to have a positive correlation with parameters such as TOC and UVA, the correlation is not sufficiently tight for general predictive use in the process design context.
- Conventional surface water treatment technology, incorporating chemical coagulation/clarification/filtration has a variable effectiveness in reducing THMFP, performing best with soft low-alkalinity waters and poorest with hard high-alkalinity waters. With the latter category of surface water, coagulation pH regulation by acid addition may be necessary to achieve the required THMFP and chlorine demand reductions necessary to ensure that the supplied water TTHM remains below the regulatory limit of 100µg/l.
- Slow sand filtration is found to effect a marginal reduction of chlorine demand and THMFP. Hence chlorine demand and THM formation should be included in the set of criteria used for evaluation of the suitability of SSF treatment.
- GAC adsorption has been found to be an excellent process for TOC/THMFP removal. However, the requirement for its regeneration/replacement is a major obstacle to its adoption as a viable process in water treatment practice.
- Chlorine demand has been shown to be a very useful and informative parameter in characterising treated surface water quality. The fact that it is easily measured using equipment available at all WTWs makes its more widespread use as a comparative treated water quality parameter readily feasible.
The authors wish to acknowledge the support of water analysis facilities provided by the UCD School of Civil, Structural and Environmental Engineering and Donegal County Council in carrying out the water analyses reported in the paper. The collaboration of the various Local Authorities that provided test water samples is also gratefully acknowledged.
Prof Tom Casey is the founder of Aquavarra Research Ltd; Hugh Kerr is water safety officer for Donegal County Council; and Pat Kearney is a staff member of UCD’s School of Civil, Structural and Environmental Engineering
Bellar et al (1974) The Occurrence of Organohalides In Finished Drinking Waters, J. AWWA, 66:12:703.
Boccelli, D., Tryby, M. E., Uber, J. G., and Summers, R. S. (2003): A reactive species model for chlorine decay and THM formation under rechlorination conditions, Water Research, 37, 2654-2666.
Brezonik, P. L. (1994) Chemical Kinetics and Process Dynamics in Aquatic Systems, CRC Press, Boca Raton.
Chowdhury, S. and Champagne, P. (2008): An investigation on parameters for modelling THM formation, Global NEST Jour., Vol. 10, No. 1, pp80-91.
Chua, K. H. (1996): THM Formation in Drinking Water, Ph.D Thesis, Department of Civil Engineering, UCD.
Clark, R. M. (1998): Chlorine demand and TTHM kinetics: a second order model. J. Environ Eng ASCE, 124(1), pp16-24.
Collins, M. R. and Eighmy, T.T. (1988): Modification to the slow sand filtration process for improved trihalomethane removal; Sect. 4.4, pp281-304 in Slow Sand Filtration: Recent developments in water treatment technology, Ed. N.J.D. Graham, Ellis Horwood.
Engerholm, B. A. and AMY G. L. (1983): A predictive model for chloroform formation from humic acid, JAWWA, 75(8), 418-423.
EPA(Irl) (2011) Water Treatment Manual: Disinfection, Environmental Protection Agency, Johnstown Castle, Co. Wexford.
EPA(Irl) (2011) The Provision and Quality of Drinking Water in Ireland, A Report for the Year 2010, Environmental Protection Agency, Johnstown Castle, Co. Wexford.
Fair, G. M., Geyer, J. C. and Okun, D. A. (1968) Water and Wastewater Engineering, Vol. 2, John Wiley & Sons, Inc.
Hrudey, S. E., Charrois, J. W. A. (2012): Disinfection By-products and Human Health, IWA Publishing, London.
Hutton, P. L. and Chung, F. I. (1994): Bromide distribution factors in THM formation. Jour. Water Resource, Planning and Management ASCE, Vol. 120, No. 1.
Kim, J., Chung, Y., Shin., D., Kim, M., Lee, Y., Lim, Y. and Lee, D. (2002): Chlorination by-product in surface water treatment process, Desalination, 151, 1-9.
Oliver, B. G. and Lawrence, S. (1979): Haloforms in drinking water: A study of precursors and precursor removal, JAWWA, 71(3),161-163.
Powell, J. C., Hallam, N. B., West, J. R., Forster, C. F. and Simms, J. (2000): Factors which control bulk chlorine decay rates, Wat. Res. Vol. 34, No. 1 pp. 117-126.
Rook, J. J. (1976) Haloform in Drinking Water, J. AWWA 68:3:168
Snoeyink, V. L. and Jenkins, D. (1980) Water Chemistry, John Wiley & Sons, Inc.
USEPA (1992) Water Treatment Simulation Program User’s Manual Version 1.21, EPA No. 811B92001.
USEPA (1999) Enhanced Coagulation and Enhanced Precipitative Softening Guidance Manual, EPA 815-R-99-012
USEPA (1999) EPA Guidance Manual, Alternative Disinfectants and Oxidants, EPA 815-R-99-0.14.
Warton, B., Heitz, A., Joll, C. and Kagi, R. (2006): A new method for calculation of the chlorine demand of natural and treated waters, Water Research, 40, 2877-2884.
WHO (1992) Guidelines for Drinking Water Quality: Vol. 2: Health Criteria and Other Supporting Information, Geneva.
WHO (2011) Guidelines for Drinking Water Quality, 4th. Edition, Annex 1: Water treatment and pathogen control: Process efficiency in achieving safe drinking water, Geneva.