1099 © IWA Publishing 2013 Water Science & Technology: Water Supply | 13.

4 | 2013

Removal of natural organic matter and trihalomethane

formation potential in a full-scale drinking water
treatment plant
Ekaterina Vasyukova, René Proft, Johanna Jousten, Irene Slavik
and Wolfgang Uhl

ABSTRACT
Ekaterina Vasyukova (corresponding author)
A multidisciplinary approach was applied in this work to characterise natural organic matter and evaluate
René Proft
the performance of a full-scale waterworks treating organic-rich surface water. It was shown that the Johanna Jousten
Irene Slavik
combination of the treatment processes considered efficiently removed the dissolved organic matter, Wolfgang Uhl
Technische Universität Dresden,
including its specific fractions. Most of the dissolved organic carbon and nitrogen (DOC and DON), Institute of Urban Water Management,
Chair of Water Supply Engineering
biodegradable DOC and DON, as well as assimilable organic carbon were removed by coagulation/
01062 Dresden,
sedimentation. However, the coagulation process was not likely to be optimised for the removal of all Germany
E-mail: [email protected]
molecular weight compounds. The breakdown of high molecular weight compounds into others of low
molecular weight, as well as the production of biodegradable organic matter during ozonation, proved to
enhance their removal efficiency by subsequent biological activated carbon filtration. The specific
trihalomethane formation potential decreased during treatment, indicating a decrease in reactivity of
DOC with chlorine across the treatment train. Fractionation experiments demonstrated that high and
medium molecular weight organics were likely to be the main precursors for the formation of
trihalomethanes. However, other disinfection by-products (such as haloacetic acids) should also be
controlled, as the chlorine demand pattern did not necessarily follow that of trihalomethane formation.
Key words | biodegradable organic matter, dissolved organic carbon, dissolved organic nitrogen,
fractionation, liquid chromatography with online organic carbon and organic nitrogen
detection, trihalomethane

INTRODUCTION AND OBJECTIVES

The removal of natural organic matter (NOM) from water is a Frimmel et al. ; Steinberg ; Huber et al. ). How-
significant issue for the drinking water industry as it causes ever, still little information is known about the fate of
adverse effects within treatment processes and distribution sys- NOM fractions and their removal across full-scale drinking
tems. NOM is considered a precursor for the formation of water treatment plants (Lee et al. ; Baghoth et al. ;
potentially carcinogenic disinfection by-products during Liu et al. ). The objectives of this study were: (i) to find
chlorination, and can decrease the effectiveness of disinfec- out how the composition and behaviour of NOM changes
tants (Singer ). Additionally it may affect the within a real waterworks, (ii) to evaluate the performance
performance of membrane systems by contributing to mem- of the treatment processes in terms of NOM removal, and
brane fouling (Zularisam et al. ), as well as favour (iii) to determine the impact of treatment on the potential
microbial growth and regrowth in distribution systems. for trihalomethane (THM) formation. The main originality
Much research has already been conducted on the of this work was to combine, for the first time on the same
characterisation of NOM in natural waters (Thurman ; object, different techniques to better understand the character

doi: 10.2166/ws.2013.095

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1100 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

and removability of NOM during full-scale water treatment. A of Ca(OH)2 and CO2 into the sedimentation basin’s clear
waterworks treating organic-rich, low-turbid surface water water run-off. A gravity granular bed filtration step, consist-
was chosen for these investigations. Liquid chromatography ing of a dual-media arrangement with 0.6 m anthracite
with online organic carbon and organic nitrogen detection (grain size: 1.4–2.5 mm) and 1.2 m fine quartz sand (grain
(LC–OCD–OND) was used for the characterisation and size: 0.71–1.25 mm), removes the remaining flocs and par-
quantification of dissolved organic carbon (DOC) and dis- ticles. Further treatment consists of an oxidation step with
solved organic nitrogen (DON) in water. Complementary ozone (0.65 ± 0.05 mgO3/L in the period investigated), fol-
experiments were conducted to assess the change in lowed by pressure BAC filtration over 2 m (empty bed
bioavailability of dissolved organic compounds, as well as contact time ∼15 min). Treated water is disinfected by a
the reactivity of NOM with chlorine to form trihalomethanes. combination of free chlorine and chlorine dioxide.

Dissolved organic matter analysis

MATERIALS AND METHODS
Total DOC was measured by the catalytic combustion
Study site and sampling method using a LiquiTOC II organic carbon analyser (Ele-
mentar Analysensysteme GmbH). Ultraviolet absorbance
A waterworks facility which treats organic-rich, moor- at a wavelength of 254 nm (UVA254) was measured in a
influenced water from the Galgenteich Reservoir (Saxony, 5 cm quartz cell using a Genesys 10 UV spectrophotometer
Germany) was investigated in the period from August to (Thermo Spectronic). Prior to these measurements,
October 2011. The source water is characterised by a high samples were pre-filtered through 0.45 μm polycarbonate
organic matter content (DOC ∼ 8 mg/L), low turbidity filters. The specific UV absorbance (SUVA) was calculated
(∼2 NTU), slightly acidic pH (∼6.0), and low conductivity by dividing the UVA254 by the corresponding DOC
(∼70 μS/cm). Samples were collected before and after each concentration.
treatment stage (n ¼ 3) in pre-cleaned DOC-free vials and The characterisation and quantification of NOM frac-
stored untreated at 4 C until analysis, which followed tions by size exclusion chromatography was performed
W

within a week. with an LC-OCD-OND system (Model 8; DOC Labor,

The treatment train of the studied waterworks com- Germany) based on a Gräntzel thin-film reactor (Huber
prises coagulation/flocculation followed by sedimentation, et al. ). The method is used to fractionate DOC and
rapid sand filtration, ozonation, biological activated DON into five fractions according to their apparent molecu-
carbon (BAC) filtration, and chlorine disinfection. Before lar weight: biopolymers, humic substances, building blocks,
the mixing tank, the pH required for coagulation is adjusted low molecular weight (LMW) acids and LMW neutrals. Due
(6.7 ± 0.1 in the period investigated) and the hardness to interferences with nitrate, only DON of the biopolymer
increased by use of calcium hydroxide Ca(OH)2 and and humic fractions could be quantified. The total DON
carbon dioxide CO2. At the inlet of the coagulation basin, concentration was thus calculated as a sum of the two
ferric chloride is added as a coagulant (at a dosage of latter fractions. Accompanying software (ChromCALC)
4 3þ
3.7 ± 0.13 × 10 mmolFe /L in the period investigated) was used for the interpretation of the results.
and dispersed by rapid mixing. Downstream of the rapid
mix, a copolymer of acrylamide and sodium acrylate is Biodegradable dissolved organic matter and assimilable
added as a flocculation aid (0.72 mg/L of polymer on aver- organic carbon assays
age). Flocs and particles are separated from the process
stream in a sedimentation tank equipped with incline- Biodegradable DOC and DON (BDOC and BDON) assays
plates. Subsequently, iron and manganese are oxidised by were performed for two sampling campaigns in September
the application of potassium permanganate just prior to fil- and November 2011, respectively (n ¼ 2). According to the
tration. For this, the pH is increased through the addition method of Joret et al. (), inoculation was performed

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1101 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

with the natural assemblage of sessile bacteria that pre- with pre-filtered raw water and placed in 10 L clean DOC-
viously grew in water of the same origin during three free glass bottles filled with MilliQ. The MilliQ water outside
weeks of incubation. One litre of sample was incubated in the membrane was exchanged every 24 h, in such a way that
a glass bottle containing 100 mL quartz sand during 14 the final volume ratio of MilliQ to the dialysis compartment
days in the dark at a temperature of 22 ± 1 C. After incu- was >100 (Vasyukova et al. ).
W

bation, samples were taken for LC–OCD–OND analysis. The determination of trihalomethane formation poten-
The BDOC and BDON were determined by subtracting tial (THMFP) was performed according to a German
the DOC and DON concentrations after the incubation standard operation procedure (DVGW W–295 ).
time from the initial corresponding concentrations of a Samples were spiked with 10 mgCl2/L in 250 mL chlorine
sample. demand-free brown glass bottles using a commercial grade
The assimilable organic carbon (AOC) determination sodium hypochlorite solution. After 48 ± 2 h incubation at
was based on a two-species bioassay using two bacteria room temperature (22 ± 1 C) the samples were transferred
W

strains, Pseudomonas fluorescens (P-17) and Spirillum into 100 mL brown glass bottles, and chlorine reaction
(NOX), according to the procedure described by van der was quenched by addition of sodium thiosulphate solution.
Kooij & Hijnen (). One AOC assay (n ¼ 1) was per- THM concentrations were analysed by gas chromatography
formed in October 2011 for raw water and each treatment according to German standard DIN EN ISO 15680 by an
step of the studied waterworks facility. accredited laboratory (Dresden, Germany). Aliquots of
chlorinated samples were used to determine the remaining
free chlorine concentration using the DPD (N,N-diethyl-
Fractionation experiments and trihalomethane
1,4-phenylenediamine) colorimetric method according to
formation potential
German standard DIN EN ISO 7393–2.

Fractionation experiments were performed on 0.45 μm pre-

filtered samples using ultrafiltration and dialysis
membranes. Frontal ultrafiltration was performed using RESULTS AND DISCUSSION
single-use SpecraPor® polyethersulphone filters with a mol-
ecular weight cut-off (MWCO) of 100 and 10 kDa. The Raw water composition
filters were flushed with 50 mL MilliQ and 50 mL sample
prior to ultrafiltration. Glycerine-cleaned SpectraPor® poly- The considered reservoir raw water exhibited relatively high
ethersulphone dialysis membranes with a MWCO of 20 kDa organic matter content with mean DOC and DON concen-
were thoroughly washed in 0.1 M HCl and MilliQ water, so trations of 7.9 mgC/L and 296 μgN/L, respectively, and a
that the DOC blank remained <0.2 mg/L. Dialysis was per- mean SUVA of 5.1 L/(mg m) and UVA254 of 40 1/m
formed over 168–192 h using 30–90 mL dialysis bags filled (Table 1). Such values are typical for waters draining the

Table 1 | Mean values and standard deviation (SD) of DOC, DON, SUVA and UVA254 (n ¼ 3), BDOC and BDON (n ¼ 2), and AOC (n ¼ 1) across the treatment train. BAC stands for biological
activated carbon, nm stands for not measured

DOC ± SD DON ± SD SUVA ± SD UVA254 ± SD BDOC ± SD BDON ± SD AOC

Water sample (mg/L) (μg/L) (L/(mg m)) (1/m) (mg/L) (μg/L) (μg Acet.–C/L)

Raw water 7.9 ± 0.2 296 ± 32 5.1 ± 0.64 40.0 ± 2.0 2.0 ± 0.3 60 ± 19 12.9
Clarified 2.7 ± 0.2 48 ± 8 2.1 ± 0.03 6.1 ± 0.4 0.4 ± 0.3 13 ± 21 9.6
Sand-filtrated 2.6 ± 0.2 35 ± 10 2.1 ± 0.17 5.5 ± 0.4 0.3 ± 0.3 11 ± 4 7.3
Ozonated 2.5 ± 0.2 35 ± 5 1.1 ± 0.17 2.8 ± 0.2 0.7 ± 0.4 19 ± 4 22.5
BAC-filtrated 1.6 ± 0.1 26 ± 5 1.3 ± 0.15 2.2 ± 0.3 0.2 ± 0.3 1±4 10.0
Finished 1.2 ± 0.1 22 ± 5 1.2 ± 0.02 1.4 ± 0.1 nm nm nm

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1102 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

upland regions of the Ore Mountains that are characterised bacterial production was small compared with the response
by upland moors and soils composed of sandy clay, podzols of respiration. Low AOC concentrations in our case can be
and brown earth. The mean pH value of 5.8 was indicative explained by two factors: (1) although phosphorous was not
of the acidic character of water draining peat lands (Thur- a limiting factor (addition to samples prior to incubation),
man ). The DOC/DON ratio averaged 27 mgC/mgN other essential inorganic nutrients (such as N, K, S, Ca,
and was relatively high for natural waters, indicating on a and Mg) might have been exhausted, taking into account
contribution of plants and soils (allochthonous material) the low levels of inorganic salts in the source water (conduc-
to a greater amount of NOM (Westerhoff & Mash ; tivity of ∼70 μS/cm; Charnock & Kjønnø ); and (2) the
Lee et al. ). specific character of the source water, i.e. the two bacteria
Humic substances were the prevailing fraction in the strains chosen for the bioassays, did not reflect the ‘real’
source water, contributing on average 60% to the bulk potential of microbial proliferation in the waters studied.
DOC which is comparable with the composition of lake
and river waters dominated by terrestrial run-off (Thurman Dissolved organic matter removal during treatment
). The biopolymer fraction represented 11% of the
bulk DOC and contained ca. 18% more DON than the Variations in the mean DOC, DON, SUVA, biodegradable
humics fraction. The DON/DOC ratio of the biopolymer DOC and DON as well as AOC values across the treatment
fraction averaged 0.22 mgN/mgC and was relatively high train of the waterworks considered are given in Table 1. In
for natural waters (Westerhoff & Mash ). The presence total, 82 and 93% of the initial DOC and DON were
of biopolymers either suggests a considerable contribution removed by the whole treatment, respectively, whereas
of autochthonous sources, such as bacteria or algae, or SUVA decreased by 76% and the UVA254 by 96%. BDOC
freshly introduced organic matter of terrestrial origin from and BDON were removed by 90 and 98%, respectively,
the catchment. Assuming a typical C:N ratio of 3:1 for pro- and AOC by 22%.
teins (Westerhoff & Mash ; Huber et al. ), the Dissolved organic matter removal mainly occurred
proportion of proteinic matter in the biopolymer fraction during coagulation/sedimentation (69 and 84% of the
averaged 63%, while 37% was comprised of polysacchar- initial DOC and DON removed, respectively). Clarifica-
ides. LMW acids were below the detection limit in all tion mainly removed the high and medium MW
samples. compounds in terms of both organic carbon and organic
The biodegradable dissolved organic matter in raw nitrogen (Figure 1): biopolymers were reduced by 90–
water averaged 2 mgC/L and 60 μgN/L, and appeared to 91% from the initial 0.81 mgC/L and 162 μgN/L, humics
be relatively high, representing 24 and 20% of the bulk by 76% from 4.35 mgC/L and 134 μgN/L, building
DOC and DON, respectively. DOC and DON removal by blocks by 47% from 1.05 mgC/L, and LMW neutrals by
bacteria after 14 days mostly occurred due to the biodegra- 44% from 0.73 mgC/L. The maximum decrease of SUVA
dation of biopolymer and humic substance fractions. and UVA254 across the treatment process also occurred
Similar values were reported for source waters of two by clarification (54 and 86%, respectively), reflecting the
other reservoir systems in the Eastern Ore Mountains by Fie- removal of refractory organic compounds (Figure 1(a)).
biger (). Although the incubation time lasted twice as Similar results for dissolved organic matter removal
long as the present study (28 days) the author found that during full-scale drinking water treatment were recently
the percentage contribution of BDOC28d to the bulk DOC reported by other authors (Baghoth et al. ; Liu et al.
averaged 21 and 25% for the reservoir waters of Carlsfeld ).
and Muldenberg, respectively. The AOC concentration of Biodegradable organic matter was almost completely
12.9 μg/L in this study was comparable with minimum removed by clarification (Figure 2): 80% of both BDOC
values of 7 μg/L reported by Fiebiger () for the two and BDON of the biopolymer fraction, as well as almost
above-mentioned reservoirs (median of 23–41 μg/L). The 100% of BDOC of the humics fraction were removed. At
AOC/BDOC ratio of 0.5% indicated that the response of the same time, BDON of the humics fraction did not show

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1103 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

Figure 1 | Mean DOC composition and SUVA (a) and mean DON composition and DON/DOC ratio of two fractions (b) of water during treatment. Error bars represent standard deviation
(n ¼ 3).

Figure 2 | Mean BDOC (a) and BDON (b) composition of water during treatment. Error bars represent standard deviation (n ¼ 2).

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1104 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

a clear trend (Figure 2(b)) and decreased by 66% after 14 of any NOM fraction (Figure 1). The amount of BDOC
days incubation in one assay, and increased by 20% in the and BDON almost did not change in sand-filtrated water
other, suggesting the possible production of soluble compared with water subjected to coagulation (Figure 2).
microbial products released by bacterial metabolism On the contrary, ozone treatment with a relatively low
during incubation (Liu et al. ). Clarification also signifi- average dosage of 0.65 mgO3/L (which corresponded to
cantly contributed to the reduction of AOC (26%). This is 0.25 mgO3/mgDOC related to the feed water of ozona-
in agreement with previously reported results, since the tion) resulted in a two-fold increase of both BDOC and
reduction in AOC by coagulation and clarification was indi- BDON as well as a three-fold increase of AOC (Figure 2).
cated in a range from 0 to 80% (Huck et al. ; Charnock In line with this fact, both SUVA and UVA254 were almost
& Kjønnø ). Charnock & Kjønnø () pointed out halved by ozonation, whereas the bulk DOC including its
that AOC molecules may be bound to humic material specific fractions remained fairly stable (Figure 1(a)). This
which could result in the easier removal of LMW non- suggests a decomposition of aromatic compounds by ozo-
humic compounds by coagulation. nation. Furthermore, the oxidation of humic substances
The coagulation process considered was likely to be into aliphatic, phenolic, and benzenecarboxylic acids in
controlled by the amount and character of NOM rather addition to normal alkanes and fatty acids makes this frac-
than by particles, as the SUVA was greater than 4 L/(mg m) tion more biologically available and might thus explain an
(Edzwald & Tobiason ) and turbidity of the source increase in BDOC, BDON and AOC in ozonated water
water was low (∼2 NTU). The pH of coagulation in this (Griffith & Schnitzer ). The BDOC increase was in
study averaged 6.7 ± 0.1, being somewhat beyond the pH agreement with Capellier et al. () who reported that
range of 5–6, which is generally considered to be optimal full-scale ozonation induced an increase of BDOC by
for coagulation with ferric chloride (Eikebrokk et al. 0.3–0.6 mg/L. A three-fold increase of AOC was consistent
). In the pH range proposed more positively charged with the data previously reported by Hammes et al. ().
coagulant species (Fe(OH)2þ and Fe3þ) are available, The production of biodegradable organic matter during
which increases the effectiveness of ferric chloride to ozonation has been proven to enhance its removal effective-
remove NOM. Considering the applied coagulant dosages ness by subsequent BAC filtration (Hambsch & Werner
4 4
of 3.7 × 10 ± 0.1 × 10 molFe3þ/L, amorphous ferric ). BDOC was reduced by 71% to 0.2 mgC/L, BDON
hydroxide was likely to be the dominant species in the con- by 95% to 1 μgN/L, and AOC by 56% to 10 μg/L (Figure 2).
sidered process. This can partially explain the fact that the A similar AOC removal efficiency was reported for a full-
coagulation procedure was not completely optimised for scale peroxidation/GAC filtration process by Volk ().
the removal of low molecular weight DOC fractions. At the same time, only 35% of the bulk DOC was removed
As mentioned above, DON was removed to a higher by BAC filtration which was mainly due to the removal of
percentage than DOC in the treatment train considered. LMW neutrals and, to a lesser extent, building blocks
This is in agreement with Lee et al. () who reported (50 and 25%, respectively; Figure 1(a)). DON decreased
that the addition of polymer as a flocculation aid led to by 26% due to the removal of biopolymers (by 63%;
the better removal of DON compared with DOC for 70% Figure 1(b)). These results indicate the preferential removal
of the 28 US waterworks facilities investigated. The N/C of biodegradable organic matter over refractory organic
ratio of the humic substances fraction of raw water was matter in this treatment step. The DOC of finished water is
0.03 on average, and remained constant during treatment somewhat lower than that of BAC filtrate. This can be
(Figure 1(b)). However, that of the biopolymer fraction explained by the fact, that only one of the two BAC filters
was initially 0.22 on average and increased gradually was sampled (no sampling point for the mixed filtrates
during treatment by 100% to 0.1, suggesting the better available) and the finished water is a mixture of two filtrates.
removal of proteins compared with polysaccharides. A decrease in SUVA and UVA245 in finished water
Within less than 0.1 mgC/L and 10 μgN/L, both sand compared with BAC filtrate can be explained by oxidation
filtration and ozonation had no impact on the removal of aromatic compounds due to chlorination.

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1105 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

Reactivity of NOM with chlorine and trihalomethane good agreement with Fabris et al. (), who found that
formation potential humic substances and building blocks correlated more
strongly with THMFP (r 2 ¼ 0.82 and 0.72, respectively)
The specific THMFP (calculated as the THM concentration than other molecular weight fractions in coagulated
divided by DOC concentration of the corresponding waters, emphasising the dominant role of humics.
sample) in the considered treatment train was mainly In order to understand the role of different organic com-
decreased by coagulation and sedimentation, where most pounds in THM formation, the fractionation of dissolved
of the DOC was removed, as well as by ozonation (Figure 3). organic matter in raw water was performed using dialysis
Compared with raw water, the specific THMFP in clarified and ultrafiltration membranes. The results of DOC fraction-
water decreased by 16%. Further ozonation was able to ation are shown in Figure 4(a). High and medium molecular
decrease the specific THMFP by 29% (compared with the weight compounds prevailed in the separated fractions of
previous treatment step). Within ±1% it was almost not 0.45 μm–20 kDa and 100–20 kDa, whereas medium and
affected by sand filtration or BAC filtration. LMW compounds prevailed in <100 kDa and <10 kDa
The specific THMFP of samples taken after different fractions. It was found out that the specific THMFP was
treatment steps positively correlated with the corresponding the highest for the high molecular weight fraction where
SUVA (r 2 ¼ 0.93; not shown) which is in agreement with both biopolymers and humic substances were present after
results of other authors who showed a positive relationship fractionation (0.45 μm–20 kDa). Moreover, the concen-
between these two parameters (Reckhow et al. ; Singer trations of biopolymers and humic substances exhibited
; Croué et al. ). The authors propose that the aro- the strongest correlations with the THMFP, with r 2 ¼ 0.96
matic compounds are the major precursors for the and 0.83, respectively (not shown). No correlation between
formation of THMs. Correlation between SUVA and specific the specific THMFP or SUVA was observed for building
THMFP is, thus, indicative of the influence of aromatic sub- blocks and LMW acids and neutrals. Overall, the specific
stances and compositional differences between NOM of THMFP decreased in the following sequence: [0.45 μm–
various samples on the formation of THMs during 20 kDa] > [100–20 kDa] > [<100 kDa] > [<10 kDa]. How-
chlorination. ever the chlorine demand did not follow this pattern,
Positive correlations were obtained between the total being the highest for fractions with the lowest specific
2
THMFP and concentrations of humic substances (r ¼ THM formation (Figure 4(b)). These results suggest that
2
0.73) and building blocks (r ¼ 0.77) in chlorinated waters high and medium MW compounds appear to be the major
of different treatment steps (not shown). However, biopoly- THM precursors in the surface water considered, whereas
mer and LMW neutral fractions revealed no correlation low MW compounds might be responsible for the formation
(r 2 < 0.4), suggesting that these fractions play a less impor- of low molecular weight (haloacetic) acids. This fact might
tant role in the formation of THMs. These results are in be worth investigating in further detail in the future.

Figure 3 | Mean specific THMFP and residual free chlorine across the treatment train after 48 ± 2 h incubation time. The error bars indicate standard deviation (n ¼ 3).

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1106 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

Figure 4 | DOC characterisation of different fractions separated using dialysis and ultrafiltration (a) and specific THMFP and specific residual free chlorine of these fractions (b). The error
bars indicate standard deviation (n ¼ 2).

SUMMARY AND CONCLUSIONS for the removal of low molecular weight compounds. The
removal of the biopolymer- and humic-bound organic nitro-
The entire treatment process of the waterworks facility gen in this treatment stage was closely associated with the
investigated achieved a relatively low level of remaining removal of the corresponding DOC fractions. Proteinic
NOM in treated water. The combination of factors including matter was preferentially removed compared with polysac-
the humic nature of NOM, coagulant dosage strategy and charides. Total and specific THMFP were at most reduced
the application of a polymer as a flocculation aid was by coagulation and ozonation. Ozone treatment alone did
likely to provide an efficient reduction in the bulk dissolved not significantly affect the bulk DOC or DON concen-
organic matter. However, whereas high and medium mol- trations, but considerably reduced the aromatic character
ecular weight fractions were to a greater extent removed, of NOM over the entire molecular weight range (which
the coagulation procedure was not likely to be optimised also reduced the specific THMFP). Furthermore, ozonation

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1107 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

resulted in an increase in BDOC, BDON and AOC by a and constructive review. We are also grateful to the staff
breakdown of refractory organic compounds into more of the waterworks involved for providing data and
easily biodegradable compounds, which could not be samples, and Jana Brückner for technical assistance.
detected by the LC-OCD-OND method alone. Subsequent
BAC filtration contributed to the biological stability of
water, reducing all three parameters as well as the REFERENCES
THMFP, but not the specific THMFP. Across the treatment
train, the SUVA pattern did not follow that of DOC removal, Baghoth, S., Sharma, S., Guitard, M., Heim, V., Croué, J.-P. &
Amy, G.  Removal of NOM-constituents as characterized
indicating the variable character of NOM.
by LC-OCD. Journal of Water Supply: Research and
One of the major concerns in finished water was the bio- Technology-Aqua 60 (7), 412–424.
degradable part of NOM. The treated water could meet the Capellier, M., Picoche, C. & Deguin, A.  Evolution of BDOC
biostability criteria with regard to the proposed limit of AOC in distribution networks. Two examples. Revue des Sciences
de l’Eau 5, 51–67.
<50–100 μg/L, supposedly due to the low initial AOC con-
Charnock, C. & Kjønnø, O.  Assimilable organic carbon and
centrations in the source water as only 22% of AOC was biodegradable dissolved organic carbon in Norwegian raw
removed by the entire treatment. BDOC was removed and drinking waters. Water Research 34 (10), 2629–2642.
Croué, J.-P., Korshin, G. & Benjamin, M.  Characterization of
more efficiently but not entirely. The characterisation of
Natural Organic Matter in Drinking Water. AWWA Research
the biodegradable organic matter showed that the pool of Foundation, USA.
indigenous bacteria utilised organic carbon from the entire DIN EN ISO 7393–2: Water quality – Determination of free
molecular weight spectrum. This implies that there is a suffi- chlorine and total chlorine – Part 2: Colorimetric method
using N,N-diethyl-1,4-phenylenediamine, for routine control
cient dynamic range of enzymes that enable the bacteria to
purposes (ISO 7393-2:1985); German version EN ISO 7393-
catabolise different sources of NOM. The need, therefore, 2:2000.
arises to control all dissolved organic matter fractions by DIN EN ISO 15680: Water quality – Gas-chromatographic
the treatment process in order to reduce the risk of determination of a number of monocyclic aromatic
hydrocarbons, naphthalene and several chlorinated
microbial proliferation in distribution systems.
compounds using purge-and-trap and thermal desorption
Overall, the results of NOM characterisation indicate (ISO 15680:2003); German version EN ISO 15680:2003.
the necessity to remove the organic matter of the entire mol- DVGW W–295  Determination of Potentials of THM-
ecular weight range, as it serves as: (1) a disinfection by- formation of Drinking Water, Water of Swimming Pools and
Baths (in German). DVGW Regelwerk, Bonn.
products precursor, which can increase the chlorine Edzwald, J. K. & Tobiason, J. E.  Chemical principle, source
demand and thus decrease the disinfection efficiency, (2) a water composition, and watershed protection. In: Water
DON source, which might lead to the production of Quality and Treatment (J. K. Edzwald, ed.). AWWA, Denver,
Colorado.
nitrogenous disinfection by-products, and (3) an energy
Eikebrokk, B., Juhna, T. & Østerhus, S. W.  Water Treatment
source for the (re)growth of microorganisms. With respect by Enhanced Coagulation – Operational Status and
to this, the importance of the combination of various Optimization Issues. TECHNEAU, The Netherlands.
methods for NOM characterisation is emphasised. Fabris, R., Chow, C., Drikas, M. & Eikebrokk, B.  Comparison
of NOM character in selected Australian and Norwegian
drinking waters. Water Research 42, 4188–4196.
Fiebiger, C.  Untersuchungen zu Eintrag, Verteilung und
ACKNOWLEDGEMENTS Verbleib natürlicher organischer Wasserinhaltstoffe aus
moorgeprägten Einzugsgebieten von Trinkwasserspeichern
(Studies on Import, Distribution and Fate of Aquatic Natural
This work was performed as part of projects REGKLAM Organic Matter from Bog Dominated Catchments of
and IWAS-Brazil funded by the German Federal Ministry Drinking Water Reservoirs). PhD Thesis, TU Dresden.
for Education and Research (BMBF) under grant numbers Frimmel, F., Abbt-Braun, G., Heumann, K., Hock, B., Lüdemann,
H.-D. & Spiteller, M.  Refractory Organic Substances in
FKZ 01LR0802B and 02WM1028/1166, respectively.
the Environment. Wiley-VHC Verlag, Weinheim.
The authors greatly appreciate this financial support. We Griffith, S. M. & Schnitzer, M.  Oxidative degradation of soil
are grateful to two anonymous reviewers for the insightful humic substances. In: Humic Substances II. In Search of

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest
 1108 E. Vasyukova et al. | NOM removal and trihalomethane formation potential Water Science & Technology: Water Supply | 13.4 | 2013

Structure (M. H. B. Hayes, P. MacCarthy, R. L. Malcolm & R. Reckhow, D. A., Singer, P. C. & Malcom, R. L.  Chlorination
S. Swift, eds). John Wiley and Son, London, UK. of humic materials: by-product formation and chemical
Hambsch, B. & Werner, P.  The removal of regrowth interpretation. Environmental Science and Technologies 11,
enhancing organic matter by slow sand filtration. In: 1655–1664.
Advances in Slow Sand and Alternative Biological Filtration Singer, P. C.  Humic substances as precursors for potentially
(N. G. Collins, ed.). John Wiley and Sons, Chichester, UK. harmful disinfection by-products. Water Science and
Hammes, F., Salhia, E., Köster, O., Kaiser, H.-P., Egli, T. & Technology 40 (9), 25–30.
Gunten, U.  Mechanistic and kinetic evaluation of Steinberg, C. E.  Ecology of Humic Substances in Freshwaters.
organic disinfection by-product and assimilable organic Springer Verlag, Berlin-Heidelberg.
carbon (AOC) formation during ozonation of drinking water. Thurman, E. M.  Organic Geochemistry of Natural Waters.
Water Research 40, 2275–2286. M. Nijhoff and W. Junk Publishers, Dordrecht,
Huber, S. A., Balz, A., Abert, M. & Pronk, W.  Characterisation of The Netherlands.
aquatic humic and non-humic matter with size-exclusion van der Kooij, D. & Hijnen, W. A. M.  Regrowth of bacteria on
chromatography-organic carbon detection-organic nitrogen assimilable organic carbon in drinking water. Journal
detection (LC-OCD-OND). Water Research 45, 879–885. français d’hydrologie 16 (3), 201–218.
Huck, P. M., Fedorak, P. M. & Anderson, W. B.  Formation Vasyukova, E. V., Pokrovsky, O. S., Viers, J., Oliva, P., Dupré, B.,
and removal of assimilable organic carbon during water Martin, F. & Candaudaup, F.  Trace elements in organic-
treatment. Journal of the American Water Works Association and iron-rich surficial fluids of the Boreal zone: assessing
83 (12), 69–80. colloidal forms via dialysis and ultrafiltration. Geochimica et
Joret, J. C., Levi, Y., Dupin, T. & Gilbert, M.  Rapid method for Cosmochimica Acta 74, 449–468.
estimating bioeliminable organic carbon in drinking water. Volk, C. J.  Biodegradable organic matter measurement and
In: Proceedings of the AWWA Annual Conference, p. 1715, bacterial regrowth in potable water. Microbial Growth in
AWWA, Denver, CO. Biofilms – Part B: Special Environments and
Lee, W., Westerhoff, P. & Esparza-Soto, M.  Occurrence and Physicochemical Aspects 337, 144–170.
removal of dissolved organic nitrogen in US water treatment Westerhoff, P. & Mash, H.  Dissolved organic nitrogen in
plants. Journal of the American Water Works Association 98 drinking water supplies: a review. Journal of Water Supply:
(10), 102–110. Research and Technology-Aqua 51 (8), 415–448.
Liu, B., Gu, L., Yu, X. & Zhang, H.  Dissolved organic Zularisam, A. W., Ismail, A. F. & Salim, R.  Behaviours
nitrogen (DON) in a full-scale drinking water treatment of natural organic matter in membrane filtration for
plant. Journal of Water Supply: Research and Technology- surface water treatment – a review. Desalination 194,
Aqua 61 (1), 41–49. 211–231.

First received 23 October 2012; accepted in revised form 16 January 2013

Downloaded from https://iwaponline.com/ws/article-pdf/13/4/1099/414991/1099.pdf

by guest