Académique Documents
Professionnel Documents
Culture Documents
Available at www.sciencedirect.com
ar t ic l e i n f o abs tra ct
Article history: Based on the concept of hydraulic flocculator, a baffled-channel powdered activated carbon
Received 4 August 2006 (PAC) contactor, placed before the rapid-mixing basin, was designed and evaluated for
Received in revised form removal of taste and odor (T&O) in drinking water. PAC adsorption kinetic tests for raw
6 February 2007 water samples were conducted for selection of design parameters related to contact time
Accepted 7 February 2007 and degree of mixing. Within the tested range of velocity gradient (G) from 18 to 83 s1,
Available online 2 April 2007 mixing had a relatively minor effect on the adsorption kinetics of the PAC. The
Keywords: hydrodynamic characteristics of the pilot-scale horizontally and vertically baffled-channel
Taste and odor (T&O) PAC contactor were investigated by tracer tests. It was found that the plug flow fractions of
Velocity gradient vertically baffled-channel PAC contactor (vBPC) were higher than those of the horizontally
Powdered activated carbon (PAC) baffled-channel PAC contactor (hBPC) for the same bend width or bend height. However,
Optimal design the hBPC seems to be more appropriate than the vBPC in terms of construction and
Baffled-channel PAC contactor maintenance. The geosmin and MIB removal rate increased with the number of baffles, PAC
dose and contact time increased regardless of bend width in the pilot-scale hBPC. The pair
of full-scale hBPCs at Pohang water treatment plant, having a design capacity of
6.5 104 m3/d with 20 min of hydraulic retention time with a safety factor of 2, was
designed based on lab- and pilot-scale experimental results. Under a velocity gradient of
20 s1, the number of baffles to be installed was calculated to be 20 with a space of about
2 m between each baffle, resulting in a hydraulic head loss through the contactor of about
0.056 m. The successful application of hBPC for T&O removal from drinking water supplies
should provide momentum for developing more effective treatment methods.
& 2007 Elsevier Ltd. All rights reserved.
1. Introduction (PAC) has been widely used for T&O removal despite its
relatively high cost, due mainly to the advantage that PAC can
Taste and odor (T&O) problems caused by eutrophication of be applied only as needed (Suffet et al., 1995). In conventional
drinking water supplies have become increasingly important water treatment plants (WTP), PAC is most commonly added
because such aesthetic qualities are the primary measures by either at the intake, rapid-mixing basin or filter inlet. Another
which consumers estimate the quality of their drinking water point of addition that should be considered, although it is not
(Mallevialle and Suffet, 1987). Powdered activated carbon commonly used, is immediately before the rapid-mixing
Corresponding author. Tel.: +82 42 280 2535; fax: +82 42 284 0109.
E-mail address: baebu@dju.ac.kr (B.-U. Bae).
0043-1354/$ - see front matter & 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.watres.2007.02.005
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 1 (200 7) 225 6 – 226 4 2257
basin. It is also true that PAC is often added to the rapid- jar was filled with 2 L of raw water from Daechung Reservoir
mixing basin simultaneously with coagulants, alkaline che- at room temperature. PAC was added from a 2000 mg/L
micals and chlorine, and the question has been raised slurry stock solution to produce concentrations of 5, 10 and
whether the adsorption efficiency of PAC may be adversely 15 mg/L. TON and UV254 were determined for samples
affected by those chemicals (Najm et al., 1991). collected at 2, 5, 10, 20 and 30 min. PAC was removed from
To reduce this uncertainty and to maximize adsorption the samples prior to analysis of T&O using GF/C filter paper
efficiency of PAC for T&O removal by minimization of and 0.45 mm membrane filters (Millipore HA type, MA, USA).
interference with other water treatment chemicals, a sepa- The rpm values of the 2 L square jars were converted to
rated baffled-channel contactor based on the concept of a velocity gradient following the procedure proposed by Hud-
hydraulic flocculator was proposed for inclusion in a newly son (1981).
constructed WTP in Korea. The hydraulic baffled-channel
contactor, which would use only hydraulic energy for mixing 2.2. Pilot-scale BPC
raw water and PAC, was to be located immediately before the
rapid-mixing basin. The advantages of such a baffled-channel Based on the concept of a baffled-channel flocculator using
contactor include good performance if the flowrate is reason- hydraulic energy for mixing (Schulz and Okun, 1984; Kawa-
ably constant, minimal maintenance due to lack of mechan- mura, 2000), a pilot-scale BPC with a working volume of 288 L
ical equipment and minimal short circuiting. Although some was manufactured, as depicted in Figs. 1 and 2. The plan area
information related to velocity gradient and retention time of the PAC contactor was 1600 mm 650 mm and the height
for flocculation in pilot- or full-scale baffled-channel floccu- was 400 mm, with a total water path of 3200 mm and effective
lators has been published (Bhargava and Ojha, 1993; Haarhoff, cross section of 300 mm 300 mm. To install various sized
1998; McConnachie et al., 1999; McConnachie and Liu, 2000), baffles, 25 grooves were provided on the inside walls of the
results from experimental research for T&O removal in contactor. The effluent was withdrawn over a weir located at
drinking water using hydraulic baffled-channel contactors the end of the contactor.
have not.
The goal of this research was to gather data to inform 2.3. Tracer tests of pilot-scale BPC
optimal design and operation of a full-scale baffled-channel
PAC contactor (BPC) specifically for T&O removal. The data The step input tracer tests were conducted for pilot-scale
were provided by experiments using two pilot-scale BPCs. In horizontally (horizontal-flow with around-the-end baffles)
one design, channel baffles were vertical, while in the other and vertically (vertical-flow with over-and-under baffles)
design, baffles were horizontal. In both designs, optimum BPC with a working volume of 288 L at 20 min of contact time
flow was the focus in experiments during which the number using 4 mg/L of phenol as a tracer. For each flow type, several
of baffles and the dimensions of the flow paths (bend height combinations of number of baffles, bend height (the distance
or width) were varied. Similarly, removal of given amounts of between the end of baffle and the bottom of contactor) and
introduced T&O compounds in a pilot-scale horizontally bend width (the distance between the end of baffle and the
baffled channel PAC contactor (hBPC) was the focus in side wall of contactor) were evaluated. Bend heights of 60 and
experiments during which PAC dose and contact time were 90 mm for the vertically baffled-channel PAC contactor (vBPC)
varied. and bend widths of 30, 60 and 90 mm for the horizontally
baffled-channel PAC contactor (hBPC) were each tested with
sets of 13, 20 and 25 baffles. Samples of effluent were taken
2. Materials and methods every 2 min and analyzed for the concentration of phenol.
Generally, step inputs were done to analyze the retention
2.1. Batch PAC adsorption kinetics time distribution (RTD) of effluent water in the PAC contactor.
Especially, the complete mixing and plug flow fractions were
Batch PAC adsorption kinetic tests were conducted to estimated using both index analysis and graphic analysis
obtain optimal mixing rate expressed by velocity gradient (Rebhun and Argaman, 1965).
(G, sec1) and contact time. Raw water samples were taken
from the Daechung Reservoir. The pH, turbidity, dissolved 2.4. Evaluation performance of pilot-scale hBPC for T&O
organic matter (DOC), UV254 and threshold odor number removal
(TON) of the raw water were 7.3, 3.1 NTU, 3.65 mg/L,
0.044 cm1 and 200, respectively. Because the analytical After tracer tests at the lab, the pilot-scale hBPC was moved to
instruments were not available during these tests, a sensory the Suji WTP in order to evaluate performance of pilot-scale
method was used. hBPC for T&O removal in drinking water supplies. At Suji
A domestic coconut-based PAC (Shinki Chemical Industrial WTP, taking raw water from the Paldang Reservoir, extensive
Corporation, Korea) was used in this study. For experiments studies on T&O removal were being conducted by a govern-
that required carbon slurries, slurry of 2000 mg/L was used. ment research institute. During this experiment, tempera-
These slurries were prepared by mixing 2 g of the carbon in 1 L ture, pH, turbidity, alkalinity, DOC and UV254 of raw water
of organic-free water. The device used for the batch kinetic were 15.672.6 1C, 7.770.4, 4.9771.02 NTU, 45.575.1 mg/L,
experiments was a jar test apparatus (PB900TM Program- 1.5870.19 mg/L and 0.02970.004 cm1, respectively. Raw
mable Jar Tester, 7790-912, Phipps & Bird, VA, USA) and water spiked with geosmin or MIB flowed continuously into
modified square jars (B-KER2, Phipps & Bird, VA, USA). Each the pilot-scale hBPC to an initial concentration range of
ARTICLE IN PRESS
2258 WAT E R R E S E A R C H 41 (2007) 2256– 2264
Fig. 1 – Schematic diagram of pilot-scale vertically baffled-channel PAC contactor: (a) cross sectional view and (b) plan view.
Fig. 2 – Schematic diagram of pilot-scale horizontally baffled-channel PAC contactor: (a) cross-sectional view and (b) plan
view.
37.6–70.8 ng/L, respectively. PAC was added to the inlet of 2.5. Analysis
hBPC from a 10,000 mg/L slurry stock solution to produce
concentration of 5, 10, 15, 20 and 30 mg/L. Samples were The TON as sensory analysis was measured in accordance
collected after 10, 20 and 30 min of contact time, filtered and with Standard Methods (APHA, 1998). For measured TON, five
analyzed for geosmin or MIB before PAC was removed from different volumes of sample were added to 500 mL Erlen-
the sample using GF/C filter papers (Whatman, UK) and meyer flasks which contain sufficient odor-free water to make
0.45 mm membrane filters (Millipore HA type, MA, USA). final volumes of 200 mL. Several blanks containing 200 mL of
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 1 (200 7) 225 6 – 226 4 2259
odor-free water were also prepared. All the samples and procedure cited in Standard Methods (APHA, 1998). A TFE
blanks were heated to 60 1C. The samples were presented to a sleeve and a vial (50 mL) were used for extraction and repeated
panel of five or more persons, in random order. The samples extraction with decreasing amounts of CS2 (10, 5, 3 and 2 mL)
were shaken and the vapor sniffed by each member of the yielded a total of 20 mL. A 1 mL of extract was injected into a
panel and the odor or lack of odor recorded. The TON was GC/MS for quantification. Stock solutions of geosmin
calculated based on the dilution ratio until odor was no longer (077–01891, 1.0 mg/mL) and MIB (134–10581, 0.1 mg/mL) were
perceptible. purchased from WAKO Pure Chemical Industries (Japan).
Closed-loop stripping analysis (CLSA, Brechbühler AG, Internal standard solutions (1-chloroalkanes, such as Cl–C6,
Switzerland) combined with gas chromatography/mass spec- Cl–C8, Cl–C12 and Cl–C18) were provided by the Bureau of
trometry (Agilent 6890N GC/5973N MSD, USA) was used for Laboratory Service at the Philadelphia Water Department,
chemical analysis of geosmin and MIB. For the concentrated Philadelphia, USA.
geosmin and MIB in the sample using CLSA, each liter of UV254 was quantified using a UV/vis spectrophotometer
sample with 100 ng of internal standard was stripped for 2 h (UV-1601, Shimadzu, Japan), after filtering through 0.45 mm
at a water bath temperature of 45 1C and a trap temperature of filter paper (Millipore HA type, MA, USA). Phenol was
75 1C. After stripping, the carbon filter (1.5 mg) was removed measured in accordance with proposed Method 5530D in
from the holder and extracted with CS2, following the Standard Methods (APHA, 1998).
100
18 G_TON
80 36 G_TON
Percent remaining (%)
83 G_TON
18 G_UV254
60
36 G_UV254
83 G_UV254
40
20
PAC dose 5 mg/L
100
80
Percent remaining (%)
60
40
20
PAC dose 10 mg/L
100
80
Percent remaining (%)
60
40
20
PAC dose 15 mg/L
0
0 5 10 15 20 25 30
Contact time (min)
Fig. 3 – Percent remaining TON and UV254 as a function of contact time according to PAC dose and velocity gradient (G).
ARTICLE IN PRESS
2260 WAT E R R E S E A R C H 41 (2007) 2256– 2264
1.0
3. Results and discussions
Figs. 4 and 5 show the results of tracer tests at the 60 and vBPC were higher than those for the 90 mm bend height (30%
90 mm of bend height or width, as well as 13 and 25 baffles in of total depth). The Morrill index (calculated using the index
the pilot-scale vBPC and hBPC. The indices, based on the analysis and expressed as t90/t10) showed a trend similar to
tracer tests, calculated using the index analysis and graphic the analysis results by Rebhun and Argaman (1965). The t10/T
analysis proposed by Rebhun and Argaman (1965) are values increased as the number of baffles increased and as
summarized in Table 1. As shown in the table, the plug flow the bend width decreased. In all cases examined, the percent
values for the 60 mm bend height (20% of total depth) in the of plug flow (p values) increased as the number of baffles
increased but as the bend dimension (bend height or width)
decreased, the percent of dead volume (m values) was nearly
1.0
zero. Rebhun and Argaman (1965) suggested in their study of
a sedimentation basin that the minus values of m had no
Relative concentration ratio(C/C0)
Table 1 – Index and graphic analysis of tracer tests for comparing vBPC and hBPC with varying numbers of baffles and
bend dimensions
Type Bend dimension—no. of baffles t90/t10 ta/T ti/T t10/T p (%) 1p (%) m (%)
Note: T ¼ theoretical retention time, t90 ¼ time when 90% of tracer is recovered, t10 ¼ time when 10% of tracer is recovered, ta ¼ average
retention time, ti ¼ time when tracer is initially detected, t90/t10 ¼ Morrill dispersion index, ta/T ¼ index of dead space, ti/T ¼ index of short
circuiting, t10/T ¼ t10 as percentage of T, p (%) ¼ fraction of active flow volume acting as plug flow, 1p (%) ¼ fraction of active flow volume
acting as mixed flow and m (%) ¼ fraction of total basin volume that is dead space.
20 mg/L of PAC was added to the pilot-scale hBPC at 10 min contactor was estimated to be about 0.056 m:
contact time, the geosmin removal rate reached 80%. If the
QrghL 1=2 rghL 1=2
desired number of baffles cannot be added, the PAC dose G¼ ¼ , (1)
mV mt
could be increased to achieve the same geosmin removal rate.
Fig. 7 shows the MIB removal rate with the same conditions as where G is velocity gradient (s1), Q is flow rate (m3/s), r is
performed in the above geosmin experiments. The MIB density of water (kg/m3), g is gravitational constant (9.81 m/s),
removal rate increased as the number of baffles and PAC hL is total head loss (m), m is dynamic viscosity (kg/m s), V is
dose increased, similar to the geosmin results. When the volume (m3) and t is hydraulic retention time (Q/V, s).
bend width was small, the MIB removal rate appeared to On the other hand, according to Schulz and Okun (1984), the
correspond with the number of baffles. Consequently, the relationship between the number of baffles and velocity
number of baffles was increased and the bend width gradient can be computed from Eq. (2). The number of baffles
decreased in order to maximize the design efficiency of the needed to achieve the desired velocity gradient (20 s1) is
hBPC. computed to be 49. However, the effect of bend width was not
considered
3.4. Design of full-scale hBPC for T&O removal in Pohang (" # )1=3
2mt ðHLGÞ 2
WTP n¼ , (2)
r 1:44 þ f Q
The pair of full-scale hBPCs at Pohang WTP having a design where n is number of baffles, H is depth of contactor (m), L is
capacity of 6.5 104 m3/d was designed based on the concept length (m) of contactor and f is friction coefficient of the
of a baffled-channel flocculator using hydraulic energy for baffles (0.02 for concrete). Thus, in this design, the number of
mixing (Schulz and Okun, 1984; Kawamura, 2000). Contact baffles was determined to be 20 and the distance between
time, velocity gradient, number of baffles, head loss and PAC each baffle to be 2 m. It was assumed that when head loss
dose were selected as design parameters for the hBPC. When accumulated through 20 bendings, reach a total head loss of
the design hydraulic retention time is 20 min, with a safety 0.056 m, the contactor would produce the desired velocity
factor of 2, the total volume of the contactors is calculated to gradient (20 s1).
be 900 m3. Considering a reasonable water depth of 3 m, the The head loss through each bend was calculated to be
length and width of each contactor were 25 and 6 m, 0.0028 m by dividing the total head loss (0.056 m) by the
respectively. number of baffles (20). The flow velocity through each bend
The velocity gradient of the full-scale hBPC was selected as can be computed
20 s1 for preventing the precipitation of PAC, because mixing 2
v
had a relatively minor effect on T&O compounds and organics h¼k , (3)
2g
removal by lab-scale PAC adsorption tests. According to Camp
and Stein (1943), the total head loss across the contactor is where h is head loss through each bend, v is flow velocity
determined by Eq. (1). The total head loss through the (m/s) through each bend and k is an empirical coefficient
ARTICLE IN PRESS
2262 WAT E R R E S E A R C H 41 (2007) 2256– 2264
100 100
Percent Geosmin remaining (%)
60 60
40 40
20
20
100
100
80
80
60
60
40
40
20
20
0
0 0 5 10 15 20 25 30
0 5 10 15 20 25 30 PAC dose (mg/L)
PAC dose (mg/L)
Bend width 30 mm - Contact time 10 min
Bend width 30 mm - Contact time 10 min Bend width 30 mm - Contact time 20 min
Bend width 30 mm - Contact time 30 min
Bend width 30 mm - Contact time 20 min
Bend width 90 mm - Contact time 10 min
Bend width 30 mm - Contact time 30 min
Bend width 90 mm - Contact time 20 min
Bend width 90 mm - Contact time 10 min
Bend width 90 mm - Contact time 30 min
Bend width 90 mm - Contact time 20 min
Bend width 90 mm - Contact time 30 min
Fig. 7 – Percent MIB remaining as a function of PAC dose in
the pilot-scale hBPC with various bend width and contact
Fig. 6 – Percent geosmin remaining as a function of PAC dose time: (a) 13 baffles and (b) 25 baffles.
in the pilot-scale hBPC with various bend width and contact
time: (a) 13 baffles and (b) 25 baffles.
Fig. 8 – Design view of full-scale horizontally baffled-channel PAC contactor at Pohang WTP.
Graham, M., Summers, R., Simpson, M., MacLeod, B., 2000. Najm, I.N., Snoeyink, V.L., Lykins Jr, B.W., Adams, J.Q., 1991. Using
Modeling equilibrium adsorption of 2-methylisoborneol and powdered activated carbon: a critical review. Am. Water Work
geosmin in natural water. Water Res. 34 (8), 2291–2300. Assoc. 83 (1), 65–76.
Haarhoff, J., 1998. Design of around-the-end hydraulic floccula- Rebhun, M., Argaman, Y., 1965. Evaluation of hydraulic
tors. J. Water SRT-Aqua. 47 (3), 142–152. efficiency of sedimentation basins. San. Eng. ASCE 91 (5),
Hudson, H.E., 1981. Water Clarification Process, Practical Design, 37–45.
and Evaluation. Van Nostrand Reinhold, New York. Schulz, C.R., Okun, D.A., 1984. Surface Water Treatment for
Kawamura, S., 2000. Integrated Design and Operation of Water Communities in Developing Countries. Wiley, New York.
Treatment Facilities, second ed. Wiley, New York. Suffet, I. H., Mallevialle, J., Kawczynski, E., 1995. Advances in
Mallevialle, J., Suffet, I. H., 1987. Identification and treatment of taste and odor treatment and Control. Report of American
tastes and odors in drinking water. Report of American Water Water Works Association Research Foundation. AWWA.
Works Association Research Foundation. AWWA. Denver, CO. Denver, CO.
McConnachie, G.L., Liu, J., 2000. Design of baffled hydraulic Teefy, S.M., Singer, P.C., 1990. Performance and analysis of tracer
channels for turbulence-induced flocculation. Water Res. 34 tests to determine compliance of a disinfection scheme with
(6), 1886–1896. the SWTR. Am. Water Work Assoc. 82 (12), 88–98.
McConnachie, G.L., Folkard, G.K., Mtawali, M.A., Sutherland, J.P., USEPA, 1989. Guidance Manual for Compliance with the Filtration
1999. Field trials of appropriate hydraulic flocculation pro- and Disinfection Requirements for Public Water Systems
cesses. Water Res. 33 (6), 1425–1434. Using Surface Water Sources. Washington, DC.