modeling the treatment of drinking water to maximize dissolved organic matter removal and minimize...

ELSEVIER Desalination 176 (2005) 81-89

DESALINATION

www.elsevier.corpYlocate/de sal

Modeling the treatment of drinking water to maximize dissolved organic matter removal and minimize disinfection by-product

formation

J. v a n L e e u w e n a,d*, R. Daly a,b, M. H o l m e s ",c

°Co-operative Research Centre for Water Quality and Treatment, Australia ~Australian Water Quality Centre, SA Water, PMB3, Salisbury, South Australia

~United Water International Pry Ltd., P.O. Box 690, Modbury, SA 5092, Australia aCS1RO Land and Water, PMB 2, Glen Osmond 5064, Australia

Received 19 October 2004; accepted 29 October 2004

Abstract

Surface waters used for drinking purposes can vary markedly in their organic and inorganic content. High levels of variation occur in a range of water quality parameters such as turbidity, alkalinity, colour, natural organic matter, algae and micro-organisms. The removal of organic matter using inorganic coagulants is impacted by the character and concentration of the organics and the turbidity and alkalinity of the raw water. Mathematical models that relate the character and concentration of dissolved organic matter in raw water to inorganic coagulant dosing that maximize removal of dissolved organic carbon (DOC) have been developed. These models were used to predict alum doses that were subsequently applied to treat waters from two Australian drinking water sources (Googong and Middle River reservoirs) under jar test conditions and in pilot plant trials. Percentage removals of DOC were - 5 0 ~ 0 % with application of model predicted alum doses for maximizing removal of DOC when coagulation was performed at pH 6. Much higher coagulant dosing at similar pH resulted in comparatively minor additional removal of DOC. Trihalomethane formation potential (THMFP) under standard laboratory conditions was found to be proportional to the residual DOC concentrations and appeared to be linearly related. Formation of individual THMs was consistent in each water source but differed between the two sources.

Keywords: Enhanced coagulation; Modeling; NOM; DBP; THMFP

*Corresponding author. Present address: Tel. +61 (8) 8303-8721; Fax +61 (8) 8303-8750; email: [email protected]

Presented at the Seminar in Environmental Science and Technology: Evaluation of Alternative Water Treatment Systems for Obtaining Safe Water. Organized by the University of Salerno with support of NATO Science Programme.

September 27, 2004, Fisciano (SA), Italy.

0011-9164/05/$- See front matter © 2005 Elsevier B.V. All rights reserved

doi: 10.1016@ desal.2004.10.024

82 J. van Leeuwen et al. /Desalination 176 (2005) 81-89

1. Introduction

Water that is acceptable to consumers for drinking purposes should be safe, meaning it is free from pathogens and does not contain chemicals at concentrations that can cause harm. It should also be aesthetically pleasing having low colour and turbidity with no unpleasant taste or odours. In order to attain high quality drinking water a range of treatment technologies can be used including membrane filtration, coagulation/ flocculation/sedimentation or dissolved air flota- tion/filtration, ozonation, activated carbon, ion- exchange resins and disinfection by chlorine or chloramine and UV irradiation.

The most common water treatment process involves the use of metal-based coagulants for removal of colour, turbidity and organic compounds. Dissolved organic matter (DOM) in raw water used for drinking purposes comprises a wide range of organic compounds, some of which are amenable to removal by coagulation using metal based coagulants while others are recalcitrant to removal. In addition to the type of DOM (measured as DOC) present in raw water, the amount of DOM removed by coagulation is determined by the dose and type of coagulant used and the pH at which coagulation occurs. For a given coagulant there is an optimum pH at which coagulation is most efficient for removal of organics. Natural DOC could be considered as comprising of two fractions in relation to a particular coagulant, that which can be removed (coagulable) and that which is recalcitrant to removal. Those that are able to be removed are predominantly high molecular weight hydrophobic compounds that tend to have higher UV (at 254 nm) absorbance and colour while those that are recalcitrant are smaller molecular weight hydrophilic compounds.

In water disinfected with chlorine, residual DOM can react to form disinfection by-products such as trihalomethanes (THM). Specified removal of organics to minimize disinfection by-products is a regulatory requirement in the USA, and water authorities in the USA, Britain and Europe are

required to meet regulatory limits on disinfection by-products. In Australia, guidelines exist for levels of disinfection by-products in drinking water [1 ].

Modeling of water treatment processes have been described by Bazer-Bachi et ai. [2], Ellis et al. [3], Girou et al. [4] and Ratnaweera and Blom [5]. These models are mostly based on relation- ships between raw and treated water quality and treatment conditions required to achieve a target water quality. Models and predictive tools for determination of the removal of organics from drinking water have been proposed by Edwards [6], Urfer et al. [7], Baxter et al. [8], Stanley et al. [9], van Leeuwen et al. [ ! 0,11 ] and Kastl et al. [ 12].

In this paper, mathematical models previously described [10,11,13] were applied for prediction of alum doses required to maximize removal of dissolved organic matter from two raw water sources, Middle River (South Australia) and Googong (Australian Capital Territory). Predicted alum doses were tested under jar test and pilot plant conditions. The formation potential of THMs from residual DOC and relative abundances of individual THMs are described.

2. Materials and methods

2.1. Samples

Raw water samples were collected from Middle River Reservoir, South Australia and from Googong Reservoir (ACT/New South Wales). At the time of sample collection, water from Googong Reservoir had a DOC concentration of -6.5 mg/1, alkalinity (as CaCO3) of 45 mg/l and turbidity o f - 2 NTU. Middle River water had a DOC o f - 1 2 - 1 4 rag/l, alkalinity -15 mg/I and turbidity -5 NTU.

2.2. Determination o f water quality parameters

Colour: Colour (Col), in Hazen units (HU)was determined by measuring the absorbance at

J. van Leeuwen et al. / Desalination 176 (2005) 81-89 83

456 nm using UV/VIS spectrophotometer (Model 918, GBC, Australia).

Dissolved organic carbon (DOC) analysis: DOC concentrations of water samples (filtered though 0.45 I~m) were determined using a total carbon analyser (Model 820, Sievers Instruments Inc., USA) and indirectly by measuring the absorbance at 254 nm using a UV/VIS spectrophotometer (Model 918, GBC, Australia)with a 1 cm quartz cell.

Turbidity: Turbidity, in nephelometric turbidity units (NTU), was measured using a Hach ratio turbidimeter (Model 2100 AN, Co., USA).

pH: Orion (Model 420A, MA. USA) and WTW pH 340i meters were used.

Trihalomethane formation potential ~HMFP) was determined by measuring trihalomethanes formed after dosing samples with chlorine (-20 mg/I) and a reaction time of 4 h at 35°C. THMs were measured using a gas chromatograph (Varian 3400) with electron capture detection. The method is based on [14; Section 6232 B] except that compounds were measured by headspace analysis using a Perkin-Elmer HS40 headspace autosampler. The sample temperature was 45°C.

2.3. Chlorine demand

The 72 h chlorine demand was determined by dosing a number of sub-samples with increasing amounts of chlorine. After 72 h, the chlorine residual of each was determined and the sample with a residual close to 0.5 mg/I was used to calculate the approximate CI 2 demand.

Chlorine decay curves were determined as follows: 1 I of treated water (pH 7.2) was dosed with an amount of chlorine equal to the approximate 72 h demand + 0.5 mg/L and then stored in an amber bottle in the dark at 20°C. At pre- determined times 100 ml samples were withdrawn for chlorine analysis over a period of 72 h.

The chlorine residual was determined using the DPD ferrous titration method. N,N-diethyl-p- phenylenediamine (DPD) is used as an indicator

in the titration procedure with ferrous ammonium sulphate (FAS) [ 14].

2.4. Coagulant dose determination using jar tests and pilot plant studies

Jar tests were performed on the above waters at ambient temperature [15] and pilot plant studies were conducted using flat-bed up flow clarifiers ffBC).

Jar tests: A SEM FMS6V variable speed, six paddle gang stirrer with 7.6 cm diameter flat paddle impellers and 2 L Gator jars were used. Chemical volumes required for coagulation pH adjustment (0.2 M NaOH, and 0.1 M H2SO4, ana- lytical reagent grade) were predicted using a model previously described [10] and the alkalinity of the raw water. Alum (Ai2(SO03.18H20 ) was added to each jar while rapid stirring at 230 RPM (G = 480 s -I). After flash mixing for 1 min. the speed was reduced to 25 RPM (G = 18 s -~) for 14 min. During this time the initial pH &each sample was determined and minor adjustments made as required. The floes were then allowed to settle for 15 min and the settled water turbidity measured. Samples were filtered (Whatman No. 1 ) to measure the filtered turbidity and subsequently through 0.45 ~tm to measure colour, UV, DOC and for metal residual analysis.

Pilotplants: The raw water from Middle River Reservoir was dosed with alum and pH control reagent (NaOH, technical grade) before entering a conditioning tank with a residence time of about 4 min. Water was then dosed with a flocculant aid before being injected atthe base of a2 m × 3.35 m high flat bottomed clarifier (FBC). A sludge blanket formed at the height of a sludge cone suspended about i.5 m from the surface of the clarifier. Excess sludge was intermittently removed from the sludge cone via a time-controlled valve. The clarifier was operated at a surface loading of 1.5 m/h (maximum flow rate of -2 ,700 L/h). The FBC used to treat Googong Reservoir water had a maximum flow rate o f - 9 0 0 L/h. It was

84 J. van Leeuwen et al. / Desalination 176 (2005) 81-89

operated in a similar manner as the plant used for Middle River water, except that hydrochloric acid (technical grade) was used for pH control.

3. Results and discussion

The removal of DOC with alum treatment at a controlled coagulation pH from Middle River Reservoir water is shown in Fig. 1. A similar DOC removal trend was found for the water collected from Googong Reservoir.

A feature of the Fig. l curve is that coagulant dose efficiency for removal of DOC is initially high and then decreases at the higher doses. This occurs as the higher molecular weight hydrophobic compounds are removed leaving smaller molecular weight hydrophilic compounds. This curve is a general trend of natural waters treated with an inorganic coagulant such as alum or ferric chloride. Selection of a dose for treatment of raw water for drinking purposes can be on the basis for maximizing DOC removal to ensure that chlorine residuals are better maintained and disinfection by-product formation is minimal. However, practical coagulant dose selection at a

large-scale water treatment facility is balanced by the need to remove organics with minimal chemical use and cost. Therefore, coagulant dose rates may be selected which result in near maximum removal of DOC or some acceptable target removal.

Models have been previously proposed that described the trend of removal of DOC with coagulant dosing at controlled pH [ 10, i 3].

One example is as follows:

DOC (residual in treated water) = A + B exp(-C × alum) ( 1 )

where A, B and C are constants for any one water and alum is in mg/l.

For the purposes of developing models that relate coagulant dose required for near maximum removal of DOC with the raw water DOC concentration and character, the dose (from jar tests conducted at optimum pH) at a gradient of-0.015 (ADOC/AAlum) was selected. This gradient, representing a reduction of 0.15 mg/1 DOC with an increase in alum dose of 10 mg/L was nominal- ly selected as a point where higher coagulant dosing would provide only minor further DOC

14" "14 t3. -13

~. 10-11"12" . . . . . . . -10"11"12

9- -9

8- -8

E3 7- -7

6.

5.

4. 3 i i , i ; 50 100 150

Alum dose (rag/L)

Fig. 1. Removal o f DOG from a sample o f M idd le River Reservoir water under jar test condi t ions at pH 5.5.


removal. The dose at this gradient was then plotted against the raw water UV absorbance with colour [UV × log (Col × 10)]. UV absorbance at 254 nm and colour can be used as indirect rapid measures of the concentration and character of dissolved organic matter (DOM) present in raw water. Compounds that are amenable to removal by coagulation, (high-molecular weight hydrophobic) typically have high UV absorbance and colour.

A relationship between coagulant dose at the ADOC/AAIum gradient o f - 0 . 0 1 5 with the character and concentration of DOC established using data from jar tests of waters ranging in DOC, colour, UV absorbance at 254 nm, turbidity and alkalinity has been previously described [13]. Coagulant dose requirements for near-maximum removal of DOC were also distinguished from that needed to remove turbidity, as fine clay.

One model that relates alum dose to achieve a ADOC/AAlum of-0.015 with the raw water UV absorbance at 254 nm and colour is shown in Fig. 2.

An example of a mathematical model of the above curve is as follows:

Alum = A + B [1 - exp (-C × UVCol)] (2)

where UVCol is UV × log (colour × 10), UV is absorbance at 254 nm cm -~, colour is in Hazen units and A, B and C are constants.

UV x lo~,Col x 10) )

Fig. 2. Relationship between alum dose (to achieve a ADOC/AAlum of-0.015) with the concentration and character of DOC (based on UV absorbance and colour).

Although alum doses predicted from the above model are unlikely to be exactly at the gradient of -O.015, these doses should be at gradients (ADOC/ AAlum) that reflect near maximum removal of DOC. Due to the low gradient used there is a potential for prediction of a coagulant dose in excess of that actually needed to give a high percentage removal of DOC and these should therefore be checked for suitability.

In aiming to maximize removal of organics from treated water, the coagulation pH needs to be considered. Maximum removal of organics occurs when coagulation is performed at the optimum pH for that coagulant, though practical constraints exist in relation to the use ofpH control chemicals, buffering capacity of the water and tolerances for treatment at low pH. For removal of dissolved organic matter from drinking water, the optimum pH range commonly reported for aluminium salts is between 5 and 6 [16]. However, lower coagulation pH can result in a higher concentration of residual aluminium (AP) being present in treated water. On that basis, selection of an alum dose and coagulation pH is influenced by the need to lower the DOC concentration and ensure that the residual aluminium concentration remains within guidelines or regulated levels.

Models that enable prediction of pH after treatment with alum for prediction of acid or alkali requirements to achieve a target coagulation pH have also been previously described [10]. These are based on knowledge of a coagulant's acidity and the raw water pH titration curve that is predicted from the raw water buffering capacity (alkalinity). The shift in pH with dosing by an acidic coagulant can be estimated as well as any additional requirements of acid or alkali to attain a target coagulation pH.

An example of the impact ofpH on the removal of DOC from Middle River Reservoir water by alum treatment in jar tests is shown in Fig. 3. The model predicted dose (MPD) was 105 mg/I alum and this is compared to removals with lower and higher doses. Lowest DOC residuals occurred

86 J. van Leeuwen et al. / Desalination 176 (2005) 81 89

6-

5

I I I .... I I I 75 75 105 105 105 160 160 160

Alum dose (mg/L)

Fig. 3. Residual DOC concentration in Middle River water after alum treatment at pH levels detailed in the columns. Results are from ajar test where the model predicted alum dose was 105 rag/1.

with highest alum dose at pH 5.5. However, this treatment condition would be considered imprac- tical on the basis of cost and acceptable coagulation pH levels at water treatment plants is mostly 6 or higher. The model predicted dose of 105 mg/i provided removal of about 60% which was similar to that obtained using the much higher dose. This MPD was also considered to be practically suit- able for application at the Middle River WTP. The Australian Drinking Water Guideline (ADWG) value for residual aluminium in treated water is 0.2 mg/I. Of the above jar tests, only the water sample with highest alum dose (160 mg/I) applied at pH 5.5 had a residual aluminium concentration that exceeded the guideline value. Water samples tested under the various jar test conditions detailed had filtered (Whatman No. 1 ) water turbidities of <0.1 NTU.

The removal o f DOC from Middle River Reservoir water in pilot plant studies is shown in Fig. 4. The highest DOC removal (64%) occurred with highest alum dose at the lowest coagulation pH. However, the MPD at pH 6 removed a com- parable amount of DOC at 59%. Residual aluminium concentrations were less than the ADWG value

t13

E 0 o

70-

65-

60-

55

5O 5.5

45-

40-

35-

3O 0.67

5.~ 16.4 5.5 5.,' 5.~ 6. - ~ 5.7

I I 0.67 0.67 1 1 1 1.5 1.5

x Model predicted alum dose

Fig. 4. Percentage removal of DOC from Middle River water after alum treatment at pH levels detailed in the columns. Results are from pilot plant trials. MPDs were 95-105 mg/I alum.

when coagulation was performed at pH 6 or greater.

The coagulation pH generally impacted on DOC removal as evidenced by chlorine decay curves (Fig. 5). The curves of samples after coagulation using the MPD at pH 5.5 and 6 were similar which can be explained by the small difference in residual DOC found in this test (Table i ). There was a higher residual DOC in the sample coagulated at 6.4 resulting in higher chlorine decay and greater amounts of THMs being formed after 72 h (Table 1). For the chlorine decay curves (and the THMs detected after 72 h) water samples were adjusted to pH 7.2, so the results obtained are likely to be due to the differences in DOC concentrations and character. However, the basis for a lack of a trend in the formation ofbromoform in the three samples is unclear.

In pilot plant trials undertaken using Googong raw water, percentage removals of DOC at a range of alum doses (as indicated by multiples of MPD) and coagulation pH (values in columns) are shown in Fig. 6. Removal of DOC at the lowest dose

J. van Leeuwen et al. / Desalination 176 (2005) 81-89

Table 1 THMs (p.g/1) measured in the Middle River sample of the chlorine decay test, at 72 h

87

pH DOC (mg/1) TTHM Chloroform Brom~ Dibromo- Bromoform dichloromethane chloromethane

5.5 4.6 154 4 28 69 53 6 4.7 17t 21 54 72 24 6.4 5.2 272 23 71 125 53

O l t • A lumMPDpH5.5 ] I~ .... o-- Alum MPD pH 6.0

11 ~ ............ Alum MPD pH 6.4 4 ii'. 1 "'~' "'"

..~ 4 - " ........ "~-: -~..~---~.

tO . ' ' " "~ ' . - . . .... • ......... - - ~ : - ~ - ~ I

5 - " " • ....... ...... A ......................................

6 6 '1'o 2'o 3'o 4 ' o ' ~ ' o 6 ' o ' z ' o ~'o Time (h)

Fig. 5. Chlorine decay curves of Middle River water after alum treatment.

applied, equivalent to 0.4 oo MPD (20 rag/l) was comparatively high compared to the model predicted dose of 41-45 mg/1 alum, verifying the need to test lower doses from that predicted, for suitability. However, jar test data of water collected at the same time showed lower DOC removals at the dose o f - 2 0 mg/1 alum; i.e. ~26% removal compared with ~42% removal with the model predicted dose at pH 6.2. Residual aluminium concentrations in water samples tested under pilot plant conditions after filtration (Whatman No. 1) were all less than 0.1 mg/1. Filtered turbidities were all less than 0.2 NTU.

The relationship between the residual DOC in waters after alum treatment with the formation potential of THMs is shown in Fig. 7. A clear linear relationship is evident for Middle River Reservoir water and a similar general trend appears to be

6 0 -

o 50- 0 m

4 0 o

> 30 o E

2 0

0 .4

10

6 . 0 6 . 0

I 0.9 0 .8 0 .9 1 1.1 1.1 1.9

6 .1 6 . 3 6 . 0 6 . 0 6.4] 5 .~

x M o d e l p red i c ted a l u m d o s e

Fig. 6. DOC removals from Googong Reservoir water treated with alum in pilot plant trials. The coagulation pH levels are detailed in columns. Model predicted alum doses ranged from 41-45 mg/1.

present for Googong Reservoir water. However, Middle River Reservoir water has an overall greater potential for formation of THMs than Googoong Reservoir water. Middle River water had a yield (THMFP/DOC) of 47 ].tg/mg (S.D. 2.5 ~tg/mg) as compared to 16.8 ~tg/mg (S.D. 1.6 gg/mg). This may be explained by the different chemical characters of the remaining DOM after alum treatment. Regardless of the residual DOC concentration in each water sample, the proportions of individual THM compounds formed in the THMFP test were very similar for the same water type, but differed between the two waters studied (Table 2). The proportions of THMs formed in the THMFP test of Middle River water were similar to those formed at 72 h in the chlorine decay tests (Table 1).

88 I van Leeuwen et al. /Desalination 176 (2005) 81 89

7 0 -

65 ¸

~- '~60-

v -

~_ 5 5 -

-1- "

F.- 5 0 -

4 5 -

r= 0£8 /-

/ f J

~ f " I I

z

280 -

260 -

240 -

~ 2 2 0 -

v

2 0 0 -

~ 180-

160 -

140

r= 0.92

/ . /

/ 1 / •

/ / , /

• / /

, /

/

218 310 32 31, 316 318 410 315 ,i0 415 DOC (mg/L) DOC (mglL)

Fig. 7. Relationship between THMFP and residual DOC in alum treated waters from Googong (left) and Middle River (right) reservoirs.

Table 2 Mean (SD) percentages of THMs formed from residual DOC in THMFP analyses

Water source Chloroform Bromo- Dibromo- Bromoform dichloromethane chloromethane

Googong Mean 72.4 24.1 3.5 ND N = 6 S.D. 1.71 1.34 0.48 ND Middle River Mean 12.6 31.6 41.6 13.3 N = 13 S.D. 1.31 0.90 3.63 1.53

The predominant THM formed from Googong water was chloroform with minor formation of dibromochloromethane, and bromoform being undetected. In contrast, the predominant THM formed in the THMFP tests of Middle River Reservoir water was dibromochloromethane, indicating that this water had a higher bromide concentration than Googong water.

The data presented in Table ! and in Fig. 7 show that small differences in the residual concentration of DOC in treated water can have a marked impact on THMFP. Consequently, where the raw water DOC concentration is high, the application of enhanced coagulation, and models that predict conditions for this, can be justified.

The application of enhanced coagulation using alum to remove THM precursors was studied by

Vrijenhoek et al. [17]. They found significantly more THM precursors were removed when coagulation was performed at pH 5.5 than ambient pH (6.8 and 7.2), which is in agreement with the findings of this study. Waters studied by Vrijhoek et ai. [ 17] had comparatively much lower natural organic concentrations (Total Organic Carbon -3.8 mg/I and - 2.5 mg/I) than waters investigated in this study.

Although application of enhanced coagulation and model predicted coagulant doses, as previously described [ 10,12,13 ] can maximize removal of organics, there remains a proportion of organics that are highly recalcitrant to further removal. This fraction is a precursor to the formation of disinfection by-products and the extent of formation being dependent on the concentration of this fraction,


the chlorine dose applied, temperature and reaction time. The application of high coagulant doses for maximizing removal of organics lowers the subsequent formation of disinfection by-products when chlorine is used for disinfection. However, this may or may not be adequate to lower disinfection by-product formation to that which is required as per guidelines and regulations for drinking water supply. Where guideline or regulatory levels are exceeded, other forms of disinfection, e.g. chloramination, would be required. Alternatively, other treatment processes that remove a greater amount of organics would be needed.

4. Conclusions

Models have been developed that relate coagulant dose to the concentration and character of organics present in natural raw waters. These models enable prediction of inorganic coagulant doses that maximize removal of organics at a particular coagulation pH. For two waters studied, the residual DOC concentration in alum treated water was directly related to the THMFP. Forma- tion of individual THMs in the THMFP test was found to be highly consistent for each water but varied markedly between the two waters.

The results of this study indicate that the application of enhanced coagulation to maximize removal of organics may not necessarily result in attainment of stringent levels of THMs in drinking water, where chlorine is used as disinfectant. This is due to the concentration of residual DOM that is recalcitrant to removal by coagulation.

Acknowledgements

The authors thank Prof. Don Bursiil (SA Water), Mr. Uwe Kaeding (UWI), Mr. Sid Fan'ell and Mr Eric Rajapakse (ACTEWAGL) for their support in this study. The contribution of the referees (Sureyya Meric and Natasa Nikolaou) is much appreciated.

References

[I] NHMRC, National Health & Medical Research Coun- cil and the Agricultural and Resource Management Council of Australia and New Zealand. Australian Drinking Water Guidelines, 1996.

[2] A. Bazer-Bachi, E. Puech-Coste, R. Ben Aim and J.L. Probst, Revue des Sciences de l'Eau, 3 (1990) 377- 397.

[3] G.W. Ellis, A.G. Collins, X. Ge and C.R. Ford, J. Environ. Engng., I 17(3) (1991) 308-319.

[4] A. Girou, M. Franceschi, E. Puech-Costes and L. Hum- bert, Reeents Prog. Genie Procedes, 6(20) (1992) 373- 385.

[5] H. Ratnaweera and H. Biota, Wat. Supp., 13(3/4) (1995) 285-289.

[6] M. Edwards, J. AWWA, 89(5) (1997) 78-89. [7] D. Urfer, RM, Huck, GA. Gagnon, D. Mutti and F.

Smith, J. AWWA, 91(3) (1999) 59-73. [8] C.W. Baxter, S.J. Stanely and Q. Zhang, J. Water SRT

-Aqua, 48(4) (1999) 129-136. [9] S. Stanley, C. Baxter, Q. Zhang and R. Shariff, Process

modelling and control of enhanced coagulation, AWWA Research Foundation and American Water Works Association, 2000.

[10] J.A. van Leeuwen, R. Fabris, L. Sledz and J.K. van Leeuwen, International Congress on Modelling and Simulation. MODSIM 2001, Canberra, Australia, pp. 1907-1912.

[It] J.A. van Leeuwen, C. Heidenreich, M. Holmes, G. Kastl, I. Fisher, Y. Nguyen, D. Oemcke, K. Craig and D. Bursill, Ozwater, Perth, Western Australia. 2003.

[ 12] Ca Kastl, A. Sathasivan, i. Fisher and J.A. van Leeuwen, J. AWWA, 98(2) (2004) 79-89.

[13] J.A. van Leeuwen, M. Holmes, C. Heidenreich, R. Daly, I. Fisher, G. Kastl, A. Sathasivan and D. Bursill. MODSIM Integrative Modelling of Biophysical, Social and Economic Systems for Resource Management Solutions, Townsville, 14-17 July 2003, pp. 1835- 1840.

[14] Standard Methods for The Examination of Water and Waste Water, A.E. Eaton, L.S. Clesceri and A.E. Greenberg (eds.), 20th ed., American Public Health Association, Washington, DC, 1998.

[15] J.A. van Leeuwen, C.W.K. Chow, R. Fabris, M. Drikas and K. Spark. AWWA 18th Federal Convention Proceedings, Adelaide, ! 1-14th April 1999.

[16] S.J. Randtke, J. AWWA, 80 (5) (1988) 40-56. [17] E.M. Vrijenhoek, A.E. Childress, M. Elimelech, T.S.

Tanaka and M.D. Beuhler, J. AWWA, 90(4) (1998) 139-150.

modeling the treatment of drinking water to maximize dissolved organic matter removal and minimize...

Documents