Content uploaded by Inchio Lou
Author content
All content in this area was uploaded by Inchio Lou on Feb 14, 2015
Content may be subject to copyright.
Desalination and Water Treatment
www.deswater.com
1944-3994/1944-3986 © 2012 Desalination Publications. All rights reserved
doi: 10/5004/dwt.2012.3262
*Corresponding author.
45 (2012) 215–221
July
Turbidity removal improvement for Yangtze River raw water
Chin Nang Lei
a
, In Chio Lou
a,b,
*, Heng Un Song
b
, Pei Sun
b
a
Department of Civil and Environmental Engineering, Faculty of Science and Technology, University of Macau,
Av. Padre Tomás Pereira, Taipa, Macau SAR
b
Sino French Water Development Co. Ltd., 718 Avenda do Conselheiro Borja, Macau SAR., P.R.China
Tel. +853 8397 8469; Fax: +853 2883 8314; email: iclou@umac.mo
Received 9 September 2011; Accepted 8 November 2011
ABSTRACT
Coagulation-fl occulation followed by sedimentation and fi ltration is the most commonly used
water treatment process, in which turbidity or particle removal is strongly dependent on proper
coagulant dosage, fl occulation mixing time, mixing intensity (Gt), and effective size (ES) of
fi lter media. Jar tests and fi ltration column tests were preformed in this study to evaluate the
turbidity removal of the Yangtze River raw water that has medium turbidity and low dissolved
organic matters. The new internal standard of 1 NTU for settled water and 0.2 NTU for outlet
water were targeted. Operational conditions of primary fl occulation (coagulant amount, mix-
ing time and Gt), secondary fl occulation, and fi lter media ES, were tested. Results showed that
under the same amount of coagulant, longer fl occulation time and higher Gt with tapered mix-
ing can enhance the turbidity removal. The optimal dosage and Gt were estimated as 12 mg l
−1
PACL and 29,000, respectively. Secondary fl occulation further reduced the turbidity of settled
water by 80%, suggesting that the smaller particles retained in the primary settled water was
focculable. Compared to using 0.95 mm ES, using 0.65 mm ES as fi lter media obtained higher
turbidity removal and can lower the residual turbidity to 0.15 NTU.
Keywords: Coagulation; Filtration; Optimization; Turbidity; Particle size; Yangtze River
1. Introduction
Today, increasing regulatory pressure, cost compe-
tition and occurrence of process upset require that the
water business units consider more thoroughly on the
treatment process, ensuring that each water parameter
can meet the regulation with the lowest amounts of
chemicals and power consumed. Turbidity is the princi-
ple parameter, which is caused by the suspended matters
or impurities, interfering with the clarity of the water.
Positive correlation between turbidity and pathogens
has been reported in previous studies, and high residual
turbidity in the treated water may promote the re-
growth of pathogens in the distribution system, leading
to waterborne disease outbreaks [1,2]. Thus the US regu-
latory limit for treated water turbidity has reduced from
1 NTU in 1989 to 0.3 NTU in 2002, and some water utili-
ties are even committed to a lower internal guideline of
less than 0.1 NTU to guard against pathogen contamina-
tion [3]. However, ineffi ciency of turbidity removal in
conventional water treatment process (coagulation-fl oc-
culation-fi ltration) is occasionally observed, particularly
in the developing countries.
Coagulant dosage and hydrodynamic environment
are the two important factors affecting the effi ciency
of coagulation-fl occulation [4]. The amount of coagu-
lant added is determined by the levels of pH, salts and
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
216
alkalinity in raw water, while the degree or extent of fl oc-
culation is controlled by the applied velocity gradients
(G) and the time of fl occulation (t) [3]. If the mixing is too
mild, it is diffi cult for the fl ocs to grow and requires a lon-
ger fl occulation time; if the mixing is too intensive, the
already formed fl ocs may re-disperse again [5–7]. In prac-
tice, jar test is usually carried out to determine the opti-
mal coagulant amount and mixing intensity (
Gt) required.
To further remove residual turbidity before fi ltration,
a secondary fl occulation for the primary settled water
can be considered to produce an effl uent with lower sus-
pended solids concentration [8]. During secondary fl oc-
culation, the surface charge of the small particles can be
neutralized and larger particles may be formed. How-
ever, additional secondary fl occulation tank with pump-
ing and mixing accessories is necessary to be installed
in the plants.
Another factor controlling the turbidity removal is
ES of the fi lter media. The performance of the fi ltration
process depends on two distinct steps: (1) the transport
of the particles to the surface of the solid-liquid inter-
face of the media and (2) attachment of these particles
onto the media or other particles which have previ-
ously been deposited on the media [9,10]. The trans-
port mechanisms of the particles within the fi lter media
include interception, diffusion and sedimentation, and
the size of the particles to be removed is a dominant
parameter determining the transport mechanism of the
particles [11]. For this reason, changing media size and/
or particle size may enhance the transport mechanism
[12]. Filter aids are sometimes added to increase the size
of the particles by inter-bridging the particles [13].
Due to the new Chinese National Drinking Water
Quality Standard (GB 5749-2006), in which a more
stringent monitoring scheme for turbidity and micro-
biological risk parameters was established, the coagu-
lation-fl occulation-fi ltration process of Changshu WTP
was evaluated in this study in order to improve the tur-
bidity removal of Yangtze River raw water, for comply-
ing with the new internal guideline set for the Chinese
Subsidiaries of Suez Environment, 1 NTU for settled
water and 0.2 NTU for outlet water. The amount of
coagulant, fl occulation mixing time and intensity, sec-
ondary fl occulation, and the ES of fi lter media were
investigated.
2. Materials and methods
2.1. Changshu full-scale WTP
Changshu WTP, located in the downstream of Yang-
tze River, is a subsidiary of SUEZ Environment WTP and
also the largest WTP in Changshu city, with a maximum
daily production capacity of 400,000 m
3
d
−1
. The raw
water has medium turbidity and low dissolved organic
matter, and is treated with the conventional coagulation-
fl occulation-sedimentation-fi ltration process. Polyalu-
minum Chloride (PACl) is used as coagulant, and the
fl occulator is designed to be in four compartments to
provide different mixing strengths, with the approximate
Gt value of 14,500. Filter media with an 0.95 mm ES are
currently used.
2.2. Jar tests
Coagulation jar tests were conducted in 1 l plexiglass
beakers using a programmable jar testing apparatus,
Model ZR4-6 (Zhongrun, China), and operated under
several mixing scenarios that mimicked different slow
mixing scenarios. Mixing intensity was quantifi ed by
the Gt value. Liquid PACl with Al
2
O
3
content of 10–11%
and 70–75% in base saturation degree (Tianshu Purifi ca-
tion Material Co. Ltd, Tianjin), were used as coagulant
and a stock solution of 10,000 mg l
−1
PACl was prepared
before the test. Supernatant samples were withdrawn at
2 cm below the water surface. Coagulation dosage was
measured by a calibrated pipette.
2.2.1. Determination of optimal coagulant dosage
The mixing simulated the in-practice operation by
using a 3 min rapid mixing followed by 20 min slow
mixing, with the corresponding fi xed rotation speeds of
250 rpm (G = 102.5 s
−1
) and 40 rpm (G = 11.9 s
−1
), respec-
tively. The Gt value for the fl occulation (slow mixing)
was about 14,500. The samples were then allowed to
stand for 20 min after mixing and the supernatant was
taken for turbidity analysis for particle number and size
distribution. PACl dosages of 8, 12, 16, 20, 24, 28 mg l
−1
were used by diluting the stock solution.
2.2.2. Flocculation under different fl occulation time
using 60 rpm
The PACL dosage was maintained as 12 mg l
−1
in this
test and the slow mixing speed used was 60 rpm (G =
20.5 s
−1
) with Gt value of 29,000, i.e., twice of that using
40 rpm. The fl occulation time examined varied from
5 min to 30 min. The samples were then allowed to settle
for 20 min before analysis.
2.2.3. Flocculation under different slow tapered mixing
Instead of using the fi xed velocity gradient, the
slow tapered mixing was introduced in this scenario
to simulate the WTP fl occulation operation, in which
four descending velocity G were used in the four cor-
responding compartments for the whole slow mixing
process. The operation of the slow tapered mixing was
shown in Table 1. The samples were then settled for
20 min before analysis.
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
217
2.2.4. Secondary fl occulation
The settled water from the WTP was collected before
the test and stirred at 60 rpm for 40 min, with the PACl
of 1, 2, 4, 6, 8 mg l
−1
added into the beaker, respectively.
Samples were then allowed to settle for 20 min before
analysis.
2.3. Filtration column test
The fi ltration column is made of PVC cylinder that
has 50 mm in diameter and 300 mm in height. Sand was
used as media, fi lled up to 90 mm high in the column.
The turbidity removal by 0.65 mm ES and 0.95 mm ES
media were examined in this test, with the addition of
2 mg l
−1
and 4 mg l
−1
of fi lter aid (PACl), mixed under
100 rpm (G = 40.7 s
−1
) for 3 min, before the test. The sand
media was backwashed with tap water before the test,
and fi ltration velocity was maintained as 7.5 m h
−1
. The
fi rst sample taken for turbidity analysis was collected
after 1000 ml of water was fi ltered to minimize the effect
of the backwash water. Sample was taken every 500 ml
thereafter.
2.4. Analytical methods
Turbidity and particle size measurement were mea-
sured using a HACH 2100AN turbid-meter (Hach Com-
pany, USA) and an IBR particle counter (IBR, USA),
respectively. Only the particles with sizes larger than
2 μm can be measured. Dissolved oxygen (DO), pH,
and conductivity were determined using HACH LDO
probe (Hach company, USA) meter, DKK-TOA HM-30R
pH meter and DKK-TOA CM-30R conductivity meter
(DKK-TOA corporation, Japan), respectively. Dissolved
organic carbon (DOC) was analyzed using UV-persul-
fate technique and the infrared carbon dioxide analyzer
(Phoenix 8000), and calibrated with potassium hydro-
gen phthalate as standard. UV-254 was measured by
following the organic constituents’ procedure using the
DR/2010 spectrometer (Hach Company, USA). Ammo-
nia and chemical oxygen demand (COD) measurement
followed standard procedures of the Chinese Environ-
mental Protection Bureau [14].
3. Results and discussion
3.1. Water characteristics of the WTP
Samples of the raw water and treated water were
obtained from May to August, 2009 and the character-
istics were measured and summarized in Table 2. The
results showed that the turbidity of the raw water in
Yangtze River was about 40 NTU, which was considered
as low to medium turbidity, and the value is a bit lower
than the previous results reported from other research
groups [15,16]. It is considered to be diffi cult to treat the
raw water with low turbidity using the traditional coag-
ulation-fl occulation process, as the concentration of par-
ticles in the water is too low to cause effective particle
collision and aggregation [17,18]. The DOC, UV254, and
specifi c ultraviolet absorbance (SUVA) values were 1.53
mg l
−1
, 4.06 m
−l
, and 2.65 l (mg m)
−1
, respectively, indicat-
ing that the water has a low potential to form the disin-
fection by products [19]. Thus this type of raw water may
not cause a disinfection problem when chlorine is used.
The removal effi ciency of COD
Mn
(COD measurement
Table 1
Operation of the slow tapered mixing
Duration (s) Impeller speed of the jar tester (rpm)
Gt = 14,500 Gt = 21,750 Gt = 29,000
300 60 80 100
300 45 60 75
300 30 45 55
300 18 25 35
Fixed impeller speed (rpm) of equivalent Gt 40 – 60
Table 2
Raw water and treated water characteristics of WTP (average ± standard deviation)
Turbidit y
(NTU)
pH COD
Mn
(mg l
−1
)
Conductivity
(μs cm
−1
)
Total alkalinity
(mg l
−1
as CaCO
3
)
Raw water 40.3 ± 10 7.84 ± 0.13 2.65 ± 0.76 323 ± 50 91 ± 5
Treated water 0.3 ± 0.06 7.66 ± 0.10 0.96 ± 0.26 317 ± 25 86 ± 3
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
218
using Potassium Permanganate Method) in the WTP
was about 60%, which suggests that the organic matters
in raw water can be removed by this conventional treat-
ment process. The typical hourly PACl dosage and tur-
bidity of raw water, settled water and outlet water are
shown in Fig. 1. It has to be noted that the settled water
and outlet water residual turbidity was above 2 NTU,
and 0.15 NTU respectively, which can meet the current
Chinese regulation. However, the coagulant dosage was
as high as 40 mg l
−1
. To comply with the new internal
guidelines and the microbiological parameters closely
related to turbidity, as well as to reduce the chemical
costs, further improvement and optimization in turbid-
ity removal are necessary.
3.2. Jar tests to determine the optimal coagulant dosage
and slow mixing condition
3.2.1. Determination of optimal coagulant dosage
A preliminary jar test was conducted for the inlet
water to determine the optimal dosage of coagulant.
Water samples were taken from the inlet of the WTP and
the initial turbidity was measured as 52 NTU. Results
showed that the residual turbidity and particle number
decreased as the PACl dosage increased (Fig. 2). Under
PACl dosage of 8 mg l
−1
, colloid suspension was observed
in the supernatant after 20 min settling and the residual
turbidity was 5.3 NTU, which was beyond the current
internal standard of 3 NTU for the settled water. When
PACl dosage increased to 12 mg l
−1
or more, larger fl ocs
were formed after 5 min slow mixing and clearer super-
natant was observed. The turbidity and particle numbers
(> 2 μm) decreased to 2.9 NTU and 10,000 particle ml
−1
,
respectively at PACl dosage of 12 mg l
−1
. At PACl dosage
of 28 mg l
−1
, the residual turbidity could be lowered to
0.91 NTU, and no re-stabilization was observed within
the range of the PACl dosage change. Considering the
chemical costs as well as operational performance, PACL
dosage of 12 mg l
−1
was considered as the dosage for
optimization in the following further studies.
3.2.2. Flocculation under different fl occulation time
using 60 rpm
Effect of fl occulation time on the removal of turbid-
ity was explored. Using different fl occulation time, it
showed that the residual turbidity decreased linearly
as the fl occulation time increase from 5 min to 20 min
(Fig. 3), which suggests that a certain fl occulation time
was essential for maintaining satisfactory fl occulation.
When the fl occulation time is too short, there may not
be suffi cient collision between particles for fl ocs forma-
tion. When the time reached 20 min, the residual turbid-
ity was approximately 1 NTU, which can meet the new
internal standard of Suez Environment. Besides, as the
slowing mixing time increased to more than 20 min, the
turbidity did not decrease much. Even though the higher
velocity G would induce more turbulence in water, lead-
ing to more collisions among the particles within a given
time, there would be an ultimate fl oc size due to a con-
tinuous breakdown of the large fl ocs, and thus there will
be a limiting fl occulation time beyond which fl oc par-
ticles will not grow (Bratby 2006). Compared to the tests
using 40 rpm, under the same PACl dosage, the turbidity
0
5
10
15
20
25
30
35
40
45
50
0:00 3:00 21:0018:0015:0012:009:006:00
Time
PACL dosge (mg/L)
0
2
4
6
8
10
12
14
16
18
20
22
Turbidity (mg/L)
Raw Water PACL Dosage
Settled water Outlet water
Fig. 1. Hourly PACL dosage and turbidity of raw water,
settled water and outlet water.
0
1
2
3
4
5
6
53025201510
PACl dosage (mg/L)
Residual Turbidity (NTU)
0
2000
4000
6000
8000
10000
12000
14000
16000
Residual Particle Number
(particle/mL)
Turbidity
Particle number
Fig. 2. Residual turbidity and particle at different PACl dosage.
0
0.5
1
1.5
2
2.5
3
05 353025201510
Flocculation Time (min)
Residual Turbidity (NTU)
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Residual Particle Number
(particle/mL)
Turbidity
Particle Number
Fig. 3. Residual turbidity under different fl occulation time.
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
219
removal using 60 rpm in this test increased from 2.9 NTU
to 1 NTU. Considering the power consumption, the opti-
mal fl occulation time was taken as 20 min.
3.2.3. Flocculation under different slow tapered mixing
Using different types of slow tapered mixing and
PACl dosage, it showed that the residual turbidity
decreased as Gt or PACl dosage increased (Fig. 4). Com-
pared to fi xed rate mixing that produce the settled water
of 2.9 NTU when Gt of 14,500 was applied (Fig. 1), the
tapered mixing can further reduce the residual turbid-
ity to 2.5 NTU. This result was consistent with previous
studies [20] that tapered mixing helps the fl occulation
process, as the bigger the particle aggregates form, the
more gentle agitation is required to avoid breaking up
the existing aggregates. While keeping the PACl dosage
of 12 mg l
−1
, increasing the mixing Gt value to 29,000 that
has the same value using fi xed 60 rpm can further lower
the turbidity to below 1 NTU.
3.2.4. Secondary fl occulation
A secondary fl occulation test was conducted to study
the fl occulability of the settled water after primary fl occu-
lation. The settled water with water turbidity of 2.4 NTU,
was used to investigate the fl occulation possibility, thus to
further increase the turbidity removal for the remaining
small particles, before fi ltration. It showed in Fig. 5 that the
residual turbidity decreased from 2.4 NTU to 1 NTU after
addition of second coagulant aid of 6 mg l
−1
PACl, suggest-
ing the secondary fl occulation can solve the high turbidity
problem in settled water to meet the new internal guide-
line of 1 NTU before fi ltration. It was also reported that
secondary fl occulation can improve the removal effi ciency
of algae and dissolved organic matters [8].
3.2.5. Correlation between particle numbers and turbidity
Large particles are more easily than small par-
ticles to form aggregate and would fi rst settle in the
sedimentation process, while the small particles form
colloids are still in the settled water. Thus turbidity is a
good indicator representing the particle number in the
water. Fig. 6 showed that there are strong correlation
between residual particle numbers and turbidity that
measured in the previous tests. Besides, under the same
particle number, the turbidity in the primary fl occula-
tion was higher than that in the secondary test, implying
the very small particles (<2 μm) that was not detected by
the particle counter and contributed to the turbidity, can
be removed during the secondary fl occulation, assum-
ing the light-scattering properties of small particle sus-
pension in both tests were the same.
3.3. Filtration column test
Settled water with the turbidity of 3.23 NTU and the
particle number of 8886 particles ml
−1
, was used as infl u-
ent for fi ltration. The results (Fig. 7) showed that more
than 60% of the initial turbidity can be removed for both
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
14500 2900021750
Flocculaiton Intensity (Gt)
Residual Turbidity (NTU)
PACl=6mg/L PACl=9mg/L
PACl=12mg/L PACl=15mg/L
PACl=18mg/L
PACl=21mg/L
Fig. 4. Residual turbidity under different Gt value and PACl
dosage.
0
0.5
1
1.5
2
2.5
0108642
PACl dosage (mg/L)
Residual Turbidity (NTU)
0
2000
4000
6000
8000
10000
12000
14000
Residual Particle Number
(particle/mL)
Turbidity
Particle Number
Fig. 5. Residual turbidity and particle number under differ-
ent PACl dosage in the secondary fl occulation test.
y = 3278x - 523.69
R
2
= 0.9741
y = 5237x - 479.66
R
2
= 0.999
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0
Residual Turbidity (NTU)
Residual Particle Number
(particle/mL)
654321
original flocculation
secondary flocculation
Fig. 6. Correlation of particle number and residual turbidity
in the tests (Notes: in original fl occulation, water samples
were from the raw water while in secondary fl occulation,
the water samples were from settled water after the primary
fl occulation).
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
220
sizes of fi lter media. Without fi lter aid addition, the resid-
ual turbidity after 1000 ml water fi ltration was 1.1 NTU
for 0.95 mm ES and 0.48 NTU for 0.65 mm ES. The 0.95
mm ES fi lter has 20% less turbidity removal effi ciency
than 0.65 mm ES fi lter. However, the smaller the grain
size is, the slower the water moves through the media
and the smaller amount of water that can be fi ltered,
reducing the fi ltration fl ow rate, i.e., if equal amounts
of water is fi ltered, small ES media increases the head
loss and thus requires more frequent backwash. These
are disadvantage using smaller ES media, even though
it can increase the turbidity removal. In addition, as the
fi ltered volume increase the difference of removal effi -
ciencies became less and reached only 10% (0.58 NTU for
0.95 mm ES fi lter and 0.28 NTU for 0.65 mm ES fi lter),
after the 4000 ml fi ltrated volume. Furthermore, addition
of fi lter aid would greatly improve the turbidity removal,
e.g., using 2 mg l
−1
of PACl the residual turbidity after
the fi rst 1000 ml fi ltrated volume, the turbidity was 0.50
NTU for 0.95 mm ES fi lter and 0.29 NTU for 0.65 mm ES
fi lter, which was only about one half of that without PACl
addition. It was also noted that the 0.65 mm ES combined
with the PACl dosage of 4 mg l
−1
, the fi nal residual tur-
bidity can be reduced to 0.15 NTU, which meet the new
internal standard of 0.2 mg l
−1
for the outlet water.
To further understand the fi ltration mechanism of
both fi lters, water samples after 1500 ml, 3000 ml and
4000 ml fi ltrated volume, were collected to determine
the corresponding particle numbers and fl oc sizes
(Fig. 8), from which the particle and turbidity removal
effi ciencies can be calculated (Fig. 9). The tests were
performed without addition of coagulant aids. It was
observed that the residual turbidity and particle num-
bers decreased with increasing the fi ltrated volumes.
Large particles have high removal effi ciency than small
particles. 0.65 mm ES fi lter have higher particle and
turbidity removal than 0.95 mm ES fi lter that after 1500
ml of water sample was fi ltered, more than 90% of par-
ticles can be removed using the 0.65 mm ES. Using the
0.000
0.200
0.400
0.600
0.800
1.000
1.200
45003500 4000300025002000150010005000
Filtrated Volumn (mL)
Residual Turbidity (NTU)
d=0.65mm PACl=0mg/L
d=0.95mm PACl=0mg/L
d=0.65mm PACl=2mg/L
d=0.95mm PACl=3mg/L
d=0.65mm PACl=4mg/L
Fig. 7. Residual turbidity in the fi ltrate under different fi lter
media size and coagulant dosage.
Media size d=0.95mm
0
500
1000
1500
2000
2500
3000
3500
Source
water
1500 40003000
Filtrated volume (mL)
Source
water
1500 40003000
Filtrated volume (mL)
Particle Number
0
500
1000
1500
2000
2500
3000
3500
Particle Number
2-3um
3-5um
5-7um
>7um
2-3um
3-5um
5-7um
>7um
a.
Media size d=0.65mm
b.
Fig. 8. Particle numbers and size distribution in the fi ltrate
for (a) media size d = 0.95 mm and (b) media size d = 0.65 mm.
Media size d=0.95mm
0
10
20
30
40
50
60
70
80
90
100
Source
water
1500 40003000
Filtrated volume (mL)
Source
water
1500 40003000
Filtrated volume (mL)
Particle/turbidity removal (%)
Particle/turbidity removal (%)
2-3um
3-5um
5-7um
>7um
Turbidity
2-3um
3-5um
5-7um
>7um
Turbidity
a)
Media size d=0.65mm
0
10
20
30
40
50
60
70
80
90
100
b)
Fig. 9. Particle and turbidity removal percentages in the fi ltrate
for (a) media size d = 0.95 mm and (b) media size d = 0.65 mm.
Downloaded by [University of Macau Library] at 01:04 16 May 2012
C.N. Lei et al. / Desalination and Water Treatment 45 (2012) 215–221
221
Acknowledgements
This project was supported by Sino French Water
Research Fund, Fundo para o Desenvolvimento das Cien-
cias e da Tecnologia (FDCT), under grant No. 016/2011/A
and Research Committee at University of Macau under
grant No. MRG002/LIC/2012. The authors are very
grateful to all staffs of Technical Department in Changshu
Sino-French Water Supply Co. Ltd., who provided full
supports to this research.
References
[1] M.W. LeChevallier and W.D. Norton, Treatments to Address
Source Water Concerns: Protozoa. Safety of Water Disinfec-
tion: Balancing Chemical and Microbial Risks, G.F. Craun
(ed.), ILSI Press, Washington, D.C., 1993.
[2] K.R. Fox, Turbidity as It Relates to Waterborne Disease Out-
breaks. Presentation at M/DBP Information Exchange, Cincin-
nati, Ohio, AWWA white paper, 1995.
[3] J. Bratby, Coagulation and Flocculation in Water and Wastewa-
ter Treatment, IWA publishing, London, UK, 2006.
[4] A. Amirtharagjah and C.R. O’Meli, Coagulation Processes:
Destabilzation, Mixing, and Flocculation. Water Quality
and Treatment: A Handbook of Community Water Supplies,
F.W.A.W.W.A. Pontius (ed.), McGraw-Hill, New York, 1990.
[5] K. Miyanami., K. Tojo and Y. Yokota, Effect of mixing on fl oc-
culation, Ind. Eng. Chem. Fundam., 21 (1982) 132–135.
[6] W.Y. Sheng, X.F. Peng and D.J. Lee, Coagulation of particles
through rapid mixing, Drying Technol., 24 (2006) 1271–1276.
[7] J. Churchill, M.W. Beutel and P.S. Burgoon, Evaluation of opti-
mal dose and mixing regime for alum treatment of Matthiesen
creek infl ow to Jameson Lake, Washington, Lake Reservoir
Manage., 25 (2009) 102–110.
[8] A.W. Timonthy and G.P. Nicholas, Optimizing fi lter perfor-
mance, J. New Engl. Water Works Assoc., 3 (1999) 6–21.
[9] K.J. Ives and J. Gregory, Basic concepts of fi ltration, Proc. Soc.
Water Treat. Exam., 16 (1967) 147–169.
[10] C.R. O’Melia and W. Stumm, Theory of water fi ltration,
J. AWWA, 59 (1967) 1393–1412.
[11] K.M. Yao, M.T. Habibian and C.R. O’Melia, Water and waste
water fi ltration: concepts and applications, Environ. Sci. Tech-
nol., 5 (1971) 1105–1112.
[12] J.S. Chang, S. Vigneswaran and J.K. Kandasamy, Effect of pore
size and particle size distribution on granular bed fi ltration
and microfi ltration, Sep. Sci. Technol., 43 (2008) 1771–1784.
[13] H. Zhu, D.W. Smith and H. Zhou, Improving removal of tur-
bidity causing materials by using polymers as a fi lter aid,
Water Res., 30 (1996) 103–114.
[14] CEPB, Analysis Method for Monitoring Water and Waste.
Environmental Science Press, Beijing, China, 2002.
[15] A. Halawik, Effect of Aluminium and iron salts in coagula-
tion on turbidity removal of Yangtze River, J. Hehai Univ. (Nat.
Sci.), 29 (2001) 114–118.
[16] Y. Zhang, Y.X. Li and J. Jia, Studies on turbidity removal of tiny
polluted autumn Yangtze River raw water using composite
coagulants of polyaluminum chloride, Fine Chem., 26 (2009)
493–497.
[17] S.K. Dentel and J.M. Gossett, Mechanisms of coagulation with
aluminum salts. J. AWWA, 80 (1988) 187–198.
[18] W.P. Cheng, F.H. Chi and C.C. Li, A study on the removal of
organic substances from low-turbidity and low-alkalinity
water with metal-polysilicate coagulants, Colloids Surf., 312
(2008) 238–244.
[19] US EPA, Enhanced Coagulation and Enhanced Precipitative
Softening Guidance Manual, United States Environmental
Protection Agency, 1999.
[20] M. James, Water Treatment Principles and Design. Wiley-
Interscience, John Wiley & Sons, New York, 1985.
0.95 mm ES, the removal effi ciency of larger particles
(>7 μm) was much higher than that of smaller particles
(2–3 μm) at the beginning of the test, with the removal
effi ciencies of 85% and 53%, respectively. However,
when the fi ltration proceeded, the removal effi ciency
of smaller particles increased to 74% after 4000 ml
fi ltrated volume. These results supported the mecha-
nisms of straining, sedimentation and interception in
the fi ltration, as small ES media has small openings
that retain the particles.
Comparing the residual turbidity and the particle
removal effi ciencies (Fig. 9), it was found that using
both media the turbidity removal was lower than the
removal of particles with sizes greater than 3 μm, sug-
gesting that small particles contributed to turbidity
more than large particles. As the particle counter can
only determine the number of particles greater than
2 μm, the diffusion mechanism applied to small par-
ticles (typically <2 μm) cannot be verifi ed in the stud-
ies. However, using 0.95 ES media, the 2–3 μm particles
removal effi ciency is lower than the turbidity removal
effi ciency, which can be hypothesized that some portion
of the small particles less than 2 μm was removed by
other mechanism, probably by diffusion. The mecha-
nism of diffusion will be systematically investigated in
the future study. Thus, to increase the turbidity removal
in fi ltration, addition of coagulant acid to increase par-
ticle sizes, and using the smaller media fi lter will be the
effective approaches.
4. Conclusions
Jar tests and fi ltration column tests were performed
to improve the turbidity removal of the conventional
coagulation-fl occulation-fi ltration process in Changshu
WTP to meet the new internal standard for turbidity
with reducing chemical and power consumption. The
parameters investigated were coagulant dosage, fl oc-
culation mixing time, tapered mixing and fi lter media
size. The results showed that using PACl dosage of 12
mg l
−1
with the tapered mixing for 20 min (Gt = 29,000),
or implementing an additional secondary fl occulation
process with 6 mg l
−1
PACl, can reduce the settled water
turbidity to 1 NTU. Secondary fl occulation process can
further remove the smaller particles that retained in the
settled water. However it would need more space for
construction additional settling tank. It was found that
there were strong correlation between turbidity and par-
ticles. Besides, small particles contributed to turbidity
more than large particles in the fi ltration, and compared
to 0.95 mm ES, 0.65 mm ES with 4 mg l
−1
PACl fi lter aid
can reach the residual outlet water to below 0.15 NTU.
The approaches will be further studied in the full-scale
Changshu WTP.
Downloaded by [University of Macau Library] at 01:04 16 May 2012