Content uploaded by Bjørnar Eikebrokk
Author content
All content in this area was uploaded by Bjørnar Eikebrokk on May 24, 2016
Content may be subject to copyright.
Removal of natural organic matter (NOM) using different
coagulants and lightweight expanded clay aggregate
filters
Bjørnar Eikebrokk1, 2 and Torgeir Saltnes2, 3
1SINTEF Civil and Environmental Engineering, Dept. Water and Wastewater, N-7465 Trondheim, Norway
2Department of Hydraulic and Environmental Engineering, Norwegian University of Science and
Technology, Trondheim, Norway
3Optiroc a.s, Oslo, Norway
Abstract Chitosan and lightweight expanded clay aggregates were investigated as alternatives to traditional
metal coagulants and anthracite in pilot scale coagulation-filtration experiments for the removal of natural
organic matter (NOM) from drinking water. The raw water tested covered a range of colour and organic
carbon of 15–50 mg Pt/L and 2–5 mg NPOC, respectively. Aluminium sulphate, poly aluminium chlorides
with different calcium content, iron chloride sulphate, and chitosan coagulants were tested. The dual media
filter bed was built with 0.6 m of 0.8–1.6 mm lightweight expanded clay aggregates (Filtralite) above 0.35 m
of 0.4–0.8 mm sand. A conventional anthracite-sand filter with the same layer depths and grain sizes was
used as a reference. In general, the maximum permissible concentration level of 0.1-mg Me/L controlled the
minimum coagulant dose requirements when metal-based coagulants were used (Norwegian water quality
standard). Typical NOM removal efficiencies obtained with metal coagulants were in the range of 75–90%
and 40–70% with respect to colour and organic carbon, respectively. Chitosan was able to remove colour
quite effectively, but this coagulant was less effective with respect to organic carbon removal. Lightweight
expanded clay aggregate was a good alternative to anthracite, with lower rates of head loss build-up, and
increased filter run length and filter storage capacity. Only small differences in effluent water quality were
detected with the two filters.
Keywords Coagulation; contact filtration; NOM removal; dual media filters; pilot experiments; chitosan;
metal coagulants; lightweight expanded clay aggregates; anthracite; process optimisation
Introduction
Coagulation and direct filtration processes have traditionally focused on the removal of tur-
bidity, not natural organic matter (NOM). In recent years, however, new regulations and
process optimisation with respect to NOM removal has gained a lot of attention (Norwegian
Ministry of Health and Social Affairs, 1995; USEPA, 1998). The main motivation for this is
primarily concerns regarding colour, taste and odour, known and unknown disinfection by-
products (DBPs), regrowth in water distribution systems, and availability of micropollutants
associated with NOM. The concept “enhanced coagulation” has evolved, involving stricter
pH-control and elevated coagulant doses in order to increase NOM removal efficiency, and
thereby control the formation of known and unknown DBPs. Elevated coagulant dose
requirements and increased sludge production rates, however, are negative implications of
the increased NOM removal obtained. This in turn may lead to relatively short filter runs due
to early breakthroughs and more rapid head loss development, increased backwash water
consumption, and the need for increased sludge processing and disposal capacity. The impor-
tance of these factors is even more pronounced in direct and contact filtration systems.
Objectives
The objective of this study is to investigate the potential of using alternative coagulants and
filter media in NOM coagulation and filtration processes. More specifically, the abilities of
Water Science and Technology: Water Supply Vol 1 No2 pp 131–140 © IWA Publishing 2001
131
Water 1.2 K16 corr 3/27/01 11:50 AM Page 131
chitosan as an alternative to traditional metal-based coagulants, and lightweight expanded
clay aggregates (Filtralite) as an alternative to anthracite in dual media filters are evaluated.
The basis for the parallel and comparative studies are the challenges normally encountered
in coagulation-filtration processes for the removal of NOM, i.e. high coagulant dose
requirements, high sludge production rates, short filter runs due to early breakthroughs or
rapid head loss development, elevated metal coagulant residuals in treated water and
sludge, etc. In this context it is important to evaluate alternative types of coagulants with the
potential of obtaining reduced coagulant dose requirements and reduced sludge production
rates, as well as a more easily biodegradable and disposable sludge. In addition, new types
of filter media need to be considered that may be more adapted to or tailor-made for coagu-
lation-direct filtration processes primarily designed for NOM removal.
Materials and methods
Three types of raw water considered typical for most Norwegian conditions were tested.
Colour levels in the 3 raw waters (RW15, RW30 and RW50) were about 15, 30 and 50 mg
Pt/L, with correspondent non-purgeable organic carbon levels of about 2, 3.5 and 5 mg
NPOC/L. RW15 was tap water from Lake Jonsvatnet, the drinking water source for the city
of Trondheim, treated by screening and low dose chlorination (0.5 mg/L) only. The raw
waters RW30 and RW50 were prepared from tap water by adding 0.03 and 0.06% by vol-
ume of a highly concentrated regenerant solution from a nearby ion exchange water treat-
ment plant for NOM removal. Typical levels of raw water turbidity were < 0.4 NTU,
alkalinity < 0.3 mmol/L, and calcium < 6 mg Ca/L. The water temperatures varied between
6–12°C. Five different coagulant alternatives were tested:
• Aluminium sulphate (ALG)
• Poly-aluminium chloride (PAX14)
• A tailor-made calcium containing poly-aluminium chloride produced to our recommen-
dations (Ca-PAX)
• Ferric chloride sulphate (JKL)
• Chitosan (Chi)
Chitosan is a natural cationic biopolymer obtained from full or partial deacetylation of
chitin, the structure polymer of the outer skeleton of insects and crustaceans (shrimp and
crab shells). The production process involves processing of the shells, deacetylation, dry-
ing and milling. The cationic charge stems from protonation of the amino groups. The coag-
ulant was dissolved in hydrochloric acid, with a concentration of the dosing solution of 10
g/L. The direct filtration pilot plant used for the experiments was designed with two parallel
filter columns (3.5 m high, 0.144 m internal diameter). One of the columns was used as a
reference dual media filter throughout the entire experimental program, with 0.6 m of
0.8–1.6 mm anthracite and 0.35 m of 0.4–0.8 mm sand. The second column contained light-
weight expanded clay aggregates (Filtralite) and sand with layer depths and grain sizes
equal to the reference filter. Filtralite is produced by burning clay at high temperatures
(1100ºC). This material can be produced with different porosities, i.e. dry densities. In the
experiments presented here, the Filtralite used was normal density crushed (F-NC), with a
wet density of approximately 1200 kg/m3. A relatively high water flow of 0.9 m3/hr was
coagulated to maintain high precision of the dosage systems. From this stream, a flow of
0.08–0.24 m3/hr was extracted and filtered. Filtration rates were in the range of 5–12 m/hr.
A saturated solution of Ca(OH)2, or CaCO3as a highly concentrated slurry of fine graded
marble particles (98% < 2 µm), CO2, HCl, and NaOH were used for pH and corrosion con-
trol. Total available head was 1.5 m H2O. Water samples were stored at 3–5°C and analysed
in accordance with the Norwegian standards for water and wastewater analyses. Pilot
plant configuration, raw water and coagulant characteristics are presented in more detail
B. Eikebrokk and T. Saltnes
132
Water 1.2 K16 corr 3/27/01 11:50 AM Page 132
elsewhere (Eikebrokk 1982, 1996, Eikebrokk and Fettig 1990, Østerhus and Eikebrokk
1994).
Results and discussion
Results on coagulation performance and coagulant dose requirements are based on on-line
water quality and analysis of water samples from filter runs that were terminated shortly
after stable effluent water quality (turbidity) was obtained, normally after 2–3 hours of fil-
tration. However, the overall process and filtration performance are evaluated on the basis
of complete filter runs, i.e. until the occurrence of breakthrough or head loss termination.
This includes filter run lengths normally in the range of 4–24 hours.
Conventional dual media anthracite/sand reference filter
Coagulation with metal coagulants. Figure 1 shows the removed fractions of colour and
organic carbon (NPOC) with ALG or JKL for coagulation of the 3 raw waters in different
molar doses. Both coagulants perform well, and the differences between the coagulants in
terms of dose requirements and colour removal efficiencies are rather small. JKL is more
effective than ALG, however, with respect to NPOC removal. The optimum pH levels for
ALG are normally within the range of 5.5–6.5, and 4.0–5.5 for JKL. The poly aluminium
chlorides (PAX14 and Ca-PAX) allow a somewhat higher and wider range of optimum pH
than ALG. The removal efficiencies presented in Figure 1 are quite high, and doses of less
than 0.05 mmol Me/L would be sufficient to comply with the colour and organic carbon
removal requirements in most cases. The results demonstrate however that residual metal,
not colour or NPOC, controls the minimum metal coagulant dose requirements. This is
illustrated in Figure 2 where RW50 is coagulated with ALG. Close to 90% and 60% colour
and NPOC removal is obtained with the minimum dose required complying with the resid-
ual Al standard of 0.1 mg/L.
B. Eikebrokk and T. Saltnes
133
Figure 1 Obtained colour (left) and NPOC removal efficiencies when using ALG or JKL for coagulation
Figure 2 Minimum dose requirements (mg Al/L as ALG) to comply with the residual Al standard of 0.1
mg/L (line) compared to the dose requirements to give colour (left bars) and NPOC removal (right) in the
range of 70–90% and 40–60%, respectively, (RW50)
Water 1.2 K16 corr 3/27/01 11:50 AM Page 133
A summary on the minimum coagulant requirements for the tested metal-based coagu-
lants and raw waters are presented in Figure 3. It can be concluded that the molar dose
requirements (mmol of Me/L) are quite similar for ALG and JKL. The minimum dose
requirements of poly aluminium coagulants are 10–30% less. More details on metal-based
coagulation performance are presented elsewhere (Eikebrokk 1996, 1999).
In Figure 4 the dose requirements are related to the specific UV-absorption (SUVA) of
the 3 raw waters tested. It can be concluded from Figure 4 that residual aluminium controls
the minimum dose requirement for raw waters with SUVAs of 4.3 or higher unless the
required colour or organic carbon removal exceeds about 90% or 60%, respectively. As raw
water SUVA decreases below 4.3, indicating a shift towards lower molecular weight NOM,
colour and especially organic carbon will gain increased importance relative to residual
aluminium with respect to coagulant dose requirements. Even for SUVAs of 3.8, however,
removal requirements of about 70% and 50% with respect to colour and NPOC are needed
before NOM removal (i.e. colour or NPOC) and not residual Al controls the minimum
coagulant dose requirement.
Coagulation with chitosan. Chitosan is a biodegradable natural cationic biopolymer pro-
duced from deacetylation of chitin from shrimp or crab shells. Chitosan can be used as an
alternative to metal coagulants and synthetic polymers for coagulation purposes. The main
benefits of using chitosan are the ease of sludge treatment and disposal due to the low metal
content and the high biodegradability. Furthermore, reduced amount of sludge produced
more rapid filter ripening and prolonged filter runs are other benefits associated with this
coagulant. The main drawback is the high cost (100–400 NOK/kg). The colour and organic
carbon removal obtained with chitosan doses in the range of 1–7.5 mg/L for the raw waters
B. Eikebrokk and T. Saltnes
134
Figure 3 Minimum dose requirements (mmol Me/L) for the different metal coagulants as a function of raw
water colour (left) and organic carbon (right)
Figure 4 Effect of raw water SUVA on the minimum specific ALG doses required to comply with the resid-
ual Al standard (<0.1 mg/L), and to obtain given colour and NPOC removals of 70–90% and 50–60%,
respectively
Water 1.2 K16 corr 3/27/01 11:50 AM Page 134
RW15 and RW50 is presented in Figure 5. The absolute doses given in Figure 5 correspond
to specific dose levels of 0.2–2.1 mg chitosan per mg of raw water NPOC. Although chi-
tosan is able to remove colour quite effectively, the results show that organic carbon is
removed to a lesser extent. This is true even at specific dose levels as high as
1.5–2 mg of chitosan per mg of NPOC. It should be noted though that chitosan contains
some 60% NPOC and the actual NPOC removal obtained by the filter unit is relatively
high. To what extent the increased level of organic carbon in the filter effluent stems from
the NPOC contribution from the coagulant itself is not yet documented. However, this may
lead to a reduced level of biostability of chitosan-treated water when compared to water
treated by metal coagulants.
Sludge production. A comparison regarding sludge production from metal coagulants and
chitosan is presented in Figure 6. The values are obtained with raw waters RW15 and RW50
at close to optimum pH conditions, i.e. ALG: pH 5.9 ( 0.3, JKL: pH 4.9 ×0.5, Chi: pH 5.9 ×
0.4. The difference between the two metal coagulants regarding suspended solids produc-
tion at equal doses can be explained to a great extent by the differences in NOM removal effi-
ciency and molecular weights of Al- and Fe-hydroxides. No metal hydroxide is formed with
chitosan. The NOM removal efficiency is also lower than the values obtained with metal
coagulants. This explains the reduced sludge production rates associated with chitosan.
Dual media expanded clay/sand filter
Lightweight expanded clay aggregates (Filtralite) and sand were investigated as filter
media by comparison to a conventional dual media anthracite-sand as a reference filter. The
composition of the tested filter beds is given below (Figure 7):
B. Eikebrokk and T. Saltnes
135
Figure 5 Effects of chitosan dose on colour and organic carbon removal. (RW15–50, pH 5.4–5.8).
(Regressions: Colour: y = 0.185 Ln (x)+0.44, R2: 0.75; NPOC: y = 0.145 Ln (x)+0.09, R2: 0.85)
Figure 6 Suspended solids production obtained with the different coagulants at optimum pH levels
Water 1.2 K16 corr 3/27/01 11:50 AM Page 135
• A conventional dual media anthracite-sand acting as a reference filter
• A dual media filter with lightweight expanded clay aggregate as an anthracite substitute.
Normal density, crushed Filtralite (F-NC) was applied.
A single medium lightweight expanded clay aggregate filter with two fractions of
Filtralite, i.e. normal density crushed (F-NC), and high density crushed (F-HC) was also
tested. The results from testing of this filter are presented elsewhere (Eikebrokk and Saltnes
2000).
Figure 8 (left graph) shows how the settling rates of grains of Filtralite NC increase due
to density increase as a result of water absorption during the first 40–50 days of storage. The
other types of grain (Filtralite HC, anthracite, and sand) are stable during storage in water.
The right graph of Figure 8 shows the grain size distribution of filter media samples taken
from the dual media Filtralite-sand and anthracite-sand filter beds. Three samples were
taken at different depths from the Filtralite layer, and one sample from the anthracite and
sand layers. The grain size distribution curves for the two filter beds are similar. The effec-
tive size for the anthracite and Filtralite grains are calculated to 0.82 and 0.81, and the uni-
formity coefficients to 1.32 and 1.31, respectively. Figure 9 shows a picture of Filtralite NC
and anthracite grains, illustrating differences in surface characteristics.
According to filtration theory, there is a close to linear relationship between head loss
development and time of filtration in cases where the filter is well adapted to the water to be
filtered (McEven 1998):
Ht=H0+ kC
0v t (1)
B. Eikebrokk and T. Saltnes
136
Figure 7 Schematic of the tested dual media filters, showing the anthracite-sand reference filter (left), and
the Filtralite-sand filter with expanded clay aggregates of normal density, crushed (F-NC) (right)
Figure 8 Measured average settling rates of grains (left) and size distribution of grain samples taken from
the layers of anthracite, sand and Filtralite NC
Water 1.2 K16 corr 3/27/01 11:50 AM Page 136
where Ht= head loss after time of filtration t (h)
H0= initial head loss at time 0 (“clean” bed)
k= constant
C0= concentration of particles in filter influent water
v= rate of filtration (m/h)
Thus, kC0vrepresents the slope of the line plotting Hversus t, and in a particular case of fil-
tration with constant conditions regarding inlet water quality, type and dose of coagulant,
rate of filtration, etc. the equation can be simplified:
Ht= H0+ K t (2)
where Kis a “constant” depending of the type of particles to be removed and properties of
the filter bed.
When different filter beds are tested in parallel and receive the same water, different K-
values illustrate differences in filter bed properties, i.e. different distribution patterns of
deposited particles within the beds. The suspended solids storage capacity in different filter
beds can be calculated as C0vttot, where ttot is the total filter run length. Table 1 shows water
quality and filter head loss data (H0, k, K) for the Filtralite-sand (F2) and anthracite-sand
(F1) reference filters when treating the raw waters RW15 and RW50 with different coagu-
lants. Metal coagulants obtained removal efficiencies in the range of 75–96% and 53–83%
B. Eikebrokk and T. Saltnes
137
Figure 9 Grains of Filtralite NC (left) and anthracite (right)
Table 1 Filtered water quality and head loss data obtained with dual media filtration in anthracite-sand (F1)
and Filtralite-sand (F2). ALG, JKL or Chi was used for coagulation of the raw waters RW15 (upper) and
RW50 (lower)
Coag- Filtr. Sampl. pH Turb. Color Org. Removed Res-Al H
0
K k
filter rate (m/hr) (hr) (NTU) (mgPt/L) carbon fractions
(mgNPOC/L) Color Org. (mgAl/L) (mH
2
O) (mH
2
O/hr)(cmH
2
O
carbon /gSS/m
2
)
ALG-F1 10.5 8.4 6.6 0.08 3 1.02 0.77 0.53 0.06 0.50 0.049 0.12
ALG-F2 10.5 8.4 6.6 0.05 3 1.04 0.75 0.54 0.02 0.51 0.047 0.11
Chi-F1 10.0 8.8 6.4 0.08 6 1.79 0.57 0.24 – 0.48 0.070 0.29
Chi-F2 10.0 8.8 6.4 0.07 6 1.84 0.62 0.22 – 0.40 0.050 0.21
Coag- Filtr. Sampl. pH Turb. Color Org. Removed Res-Al H
0
K k
filter rate (m/hr) (hr) (NTU) (mgPt/L) carbon fractions
(mgNPOC/L) Color Org. (mgAl/L) (mH
2
O) (mH
2
O/hr)(cmH
2
O
carbon /gSS/m
2
)
ALG-F1 8.9 2.6 6.2 0.13 3 1.43 0.93 0.71 0.10 0.42 0.07 0.029
ALG-F2 8.9 2.6 6.0 0.14 3 1.35 0.96 0.72 0.12 0.38 0.06 0.024
JKL-F1 8.8 3.9 5.3 0.25 3 1.00 0.94 0.80 0.26 0.41 0.12 0.077
JKL-F2 8.8 3.9 5.2 0.21 3 0.82 0.95 0.83 0.27 0.41 0.09 0.061
Chi-F1 10.0 4.8 6.9 0.07 15 3.29 0.70 0.31 – 0.43 0.14 0.283
Chi-F2 10.0 4.8 6.7 0.07 17 3.27 0.66 0.32 – 0.35 0.10 0.205
Water 1.2 K16 corr 3/27/01 11:50 AM Page 137
with respect to colour and NPOC, respectively. Removal efficiencies for colour and NPOC
are lower for chitosan at the given doses, typically in the range of 57–70%, and 22–32%.
The differences between the filters are small with respect to effluent water quality as well as
initial head loss (H0). Filtralite, however, seems to have some benefits over anthracite in
terms of head loss development and head loss distribution.
The filter run lengths and solids storage capacities obtained at different filtration rates in
the two filters when treating RW15 and RW50 are presented in Figures 10 and 11, respec-
tively. Again, for both types of raw water and for all coagulants tested the dual media
Filtralite-sand filter (F2) performs better than the anthracite-sand reference filter (F1).
Filtralite in different size and density fractions can also be applied successfully as the only
filter media. A presentation of the results obtained with the combination of chitosan coagu-
lation and single medium Filtralite filters is given elsewhere (Eikebrokk and Saltnes 2000).
Conclusions
Typical colour and organic carbon removal efficiencies of 80–95% and 50–75% were
obtained at optimum conditions when Al- or Fe-based coagulants were used for the
removal of NOM from low turbidity raw waters with colour and organic carbon levels in
the range of 15–50 mgPt/L and 2–5 mg NPOC/L, respectively.
Maximum permissible level of coagulant residual (0.1 mg Me/L) determined the mini-
mum metal coagulant dose requirements. When using these minimum dose levels, the treat-
ed water quality fulfils the requirements regarding colour and organic carbon levels given
in the U.S. and in the Norwegian standards by a good margin. Typically, if a metal coagu-
lant dose level of 0.05 mmol of Me/L was sufficient to comply with the NOM removal
requirements, about 0.10 mmol/L was needed to comply with the residual metal
requirement.
Ferric chloride sulphate was effective over a broader range of pH when compared to alu-
minium sulphate. Although good results were obtained with both coagulants, iron was
more effective than aluminium in terms of colour and in particular organic carbon removal.
In terms of molar dose requirements as controlled by metal residuals however, no major
differences were detected between the two coagulants.
B. Eikebrokk and T. Saltnes
138
Figure 10 Filter run lengths and suspended solids storage capacities obtained in the dual media anthracite
(F1-left bars) and Filtralite (F2-right bars) filters when using ALG or Chi for coagulation (RW15, coagulant
doses of 1.0 mg Al/L for ALG, and 1.5 mg/L for Chi)
Water 1.2 K16 corr 3/27/01 11:50 AM Page 138
When using poly aluminium coagulants the dose requirements were reduced by 10–30%
in comparison to aluminium sulphate. Poly aluminium coagulants were effective over a
broader range of pH.
The natural cationic biopolymer chitosan is an alternative to traditional coagulants for
the removal of NOM from drinking water. With specific coagulant doses in the range of
1.5–2 mg of chitosan per mg of NPOC, typical colour and NPOC removal obtained is
50–70% and 30–40%, respectively.
Sludge production is reduced to about 50% with chitosan compared to metal coagulants.
Filter run lengths are also increased, and problems related to residual metal in finished
water are non-existent. Furthermore, sludge disposal is simplified due to increased
biodegradability and reduced metal content. However, the cost of chitosan is high.
Lightweight expanded clay aggregate (Filtralite) is a good alternative to anthracite in
dual media filters with sand as the second medium. With traditional dual media filter design
using 0.6 m of 0.8–1.6 mm Filtralite or anthracite (in the reference filter) above 0.35 m of
0.4–0.8 mm sand, the two tested filters performed equally well in terms of filter effluent
water quality.
In spite of the fact that only minor differences in filter grain size distribution were
detected from samples taken at different depths from the filter beds, the Filtralite-sand filter
rates of head loss development were 12–28% lower than the anthracite-sand reference
filter.
Filter run length and solids storage capacity is highly dependent on factors like raw
water quality, applied filtration rate, type of coagulant and coagulant dose applied. For the
range of experimental conditions investigated in this study, the average filter run length and
solids storage capacity values obtained with the Filtralite-sand filter were 33–36% higher
B. Eikebrokk and T. Saltnes
139
Figure 11 Filter run lengths and suspended solids storage capacities obtained with the dual media
anthracite (F1-left bars) and Filtralite (F2-right bars) filters when using ALG, JKL or Chi for coagulation
(RW50, coagulant doses of 3.1 mg Al/L as ALG, 6.1 mg Fe as JKL and 5.0 mg/L as Chi)
Water 1.2 K16 corr 3/27/01 11:51 AM Page 139
than the ones obtained in the parallel reference anthracite-sand filter operated under the
same conditions.
References
Eikebrokk, B. (1982). Coagulation/direct filtration for the removal of humic substances from drinking
water. Dr. Ing. thesis, Norwegian University of Science and Technology, Trondheim, Norway (In
Norwegian).
Eikebrokk, B. (1996). Removal of humic substances by coagulation. In: Chemical Water and Wastewater
Treatment V, Hahn, H.H., Hoffmann, E. and Ødegaard, H. (Eds.): Springer-Verlag, Berlin, Heidelberg,
pp. 173–187.
Eikebrokk, B. (1999). Coagulation-direct filtration of soft, low alkalinity humic waters. Water Science and
Technology, 40(9), pp. 55–62.
Eikebrokk, B. and Fettig, J. (1990). Treatment of coloured surface water by coagulation/direct filtration:
Effect of water quality, type of coagulant and filter aids. In: Chemical Water and Wastewater Treatment,
Hahn, H.H. and Klute, R (Eds.), Springer-Verlag, Berlin, Heidelberg, pp. 361–376.
Eikebrokk, B. and Saltnes, T. (2000). NOM removal from drinking water by chitosan coagulation and filtra-
tion through light-weight expanded clay aggregate filters”. Proceedings: Innovations in Conventional
and Advanced Water Treatment Processes, Amsterdam, 26–29 September.
McEwen, J.B. (Ed.) (1998). Treatment Process Selection for Particle Removal. AWWARF/IWSA-report,
Denver, Co. USA, ISBN 0-89867-887-0.
Norwegian Ministry of Health and Social Affairs (1995). Regulations on Water Supply and Drinking Water,
Oslo, Norway, February 1995.
USEPA (1998). Disinfectants and Disinfectants Byproducts. Final Rule. Fed. Reg., 63:241:69478, USA.
Østerhus, S.W. and Eikebrokk, B. (1994). Coagulation and Corrosion Control for Soft and Coloured
Drinking Water. In: Chemical Water and Wastewater Treatment III, Klute, R. and Hahn, H.H., (Eds.):
Springer-Verlag, Berlin, Heidelberg, pp. 137–153.
B. Eikebrokk and T. Saltnes
140
Water 1.2 K16 corr 3/27/01 11:51 AM Page 140
