Sei sulla pagina 1di 8

w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/watres

Comparison of NOM removal and microbial


properties in up-flow/down-flow BAC filter

Lineng Han a,*, Wenjun Liu a,*, Mo Chen a, Minglu Zhang b, Shuming Liu a,
Ruilin Sun c, Xiangqin Fei c
a
School of Environment, Tsinghua University, Beijing 100084, China
b
School of Food and Chemical Engineering, Beijing Technology and Business University, 100048, China
c
China Water Industry Investment Co., Ltd., Huai’an Water Supply Company, Huai’an 223002, China

article info abstract

Article history: The removal of natural organic matter (NOM) in term of CODMn by up-flow biologically
Received 10 December 2012 activated carbon filter (UBACF) and down-flow biologically activated carbon filter (DBACF)
Received in revised form was investigated in a pilot-scale test. The impacts of the molecular weight distribution
21 April 2013 of NOM on its degradation by the UBACF and DBACF were evaluated. The relationship
Accepted 15 May 2013 between biodegradation and the microbial properties in the UBACF and DBACF were
Available online 7 June 2013 approached as well. The feed water of the UBACF and DBACF were pumped from the
effluent of the rapid sand filtration (RSF) of Chengnan Drinking Water Treatment Plant
Keywords: (CDWTP), Huaian, Jiangsu Province, China. When the adsorption was the dominant
Up-flow biologically activated mechanism of NOM removal at the initial stage of operation, the CODMn removal efficiency
carbon by the UBACF was lower than the DBACF. However, with the microbes gradually accu-
Down-flow biologically activated mulated and biofilm formed, the removal of CODMn by the UBACF increased correspond-
carbon ingly to 25.3%, at the steady-state operation and was approximately 10% higher than that
NOM removal by the DBACF. Heterotrophy plate count (HPC) in the finished water of the UBACF was
Microbial structure observed 30% higher than that of the DBACF. The UBACF effluent had higher concentration
Pyrosequencing of detached bacteria whereas the DBACF harbored more attached biomass. The highest
attached biomass concentration of the UBACF was found in the middle of the GAC bed. On
the contrary, the highest attached biomass concentration of the DBACF was found on the
top of the GAC bed. Furthermore, a total of 9479 reads by pyrosequencing was obtained
from samples of the UBACF and DBACF effluents. The UBACF effluent had a more diverse
microbial community and more even distribution of species than the DBACF effluent did.
Alphaproteobacteria and Betaproteobacteria were the dominant groups in the finished water of
the UBACF and DBACF. The higher organic matter removal by the UBACF was attributed to
the presence of its higher biologically activity.
ª 2013 Elsevier Ltd. All rights reserved.

1. Introduction treatment process typically consists of flocculation, filtration


and disinfection, and NOM removal efficiency is only
Control of NOM has become one of the major issues in approximately 30% (Randtke, 1988). These treatment pro-
ensuring the safety of drinking water. The conventional cesses are ineffective in eliminating some specific NOM

* Corresponding authors. Tel.: þ86 15010351533; fax: þ86 (0)1062797643.


E-mail address: linenghan@gmail.com (L. Han).
0043-1354/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.watres.2013.05.022
4862 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8

fractions, such as disinfection by-product precursors and China. The pilot-scale process train (Fig. 1) consisted of two
synthetic organic chemicals (Nishijima and Speitel, 2004; parallel pilot systems of ozoneUBACF and ozoneDBACF.
Simpson, 2008). Because of the deterioration of source water The effluent of CDWTP rapid sand filtration (RSF) was pumped
quality and the public demand for safe drinking water in to the pilot test. The feed water qualities of pilot testing sys-
China, the combination of ozonation and biologically acti- tem are listed in Table 1.
vated carbon (BAC) (referred as O3BAC) technology has been The diameter of ozone contactor was 10 cm and the water
used as the main advanced drinking water treatment process depth was 3 m. Rapid sand filtration effluent was fed from the
to improve the water quality since 1985 (Chen et al., 2007). top and ozone gas was fed from the bottom of the ozone
At present, the down-flow BAC filter (DBACF) has been contactor. The hydraulic loading of ozone contactor was
widely applied because of its particle filtration capabilities and approximately 9 m/h, which provided a hydraulic retention
simplicity (Schideman et al., 2007). In China, the DBACF is time (HRT) of 20 min. The ozone gas was continuously bubbled
normally located following conventional treatment process for into water through a porous ceramic aerator. Ozone was
retrofitting the existing water treatment processes. But some produced from pure oxygen using CF-G-2-10G ozone gener-
disadvantages of the DBACF were found from the operational ator (Qingdao GuoLin Industrial Company Ltd.). Ozone dose
experience which hence limited its widespread acceptance. was approximately 2 mg/L, which was measured using an on-
The first major limitation of the DBACF is high head loss. line ozone gas analyzer (Zibo Ideal Measurement and Control
Typically, backwashing was required every 3e7 days to reduce Technology Co. Ltd, IDEAL-2000, USA).
the head loss. The second major limitation is that micro- The UBACF and DBACF used acrylic columns with diameter
organisms mostly grow and thrive in the upper layers of the and height of 8 cm and 5 m, respectively. The bottom of the
BAC filter and the organics are hence removed in the top UBACF and DBACF columns were lined with 20 cm sand fol-
portions of the bed (Velten et al., 2011a) while the lower layers lowed by 10 cm of gravel. The HPC testing samples of GAC
are not fully utilized. The DBACF is therefore not effective in particles were taken at the height of 10 cm, 160 cm and 190 cm
fully utilizing the entire activated carbon layer for NOM of the UBACF and 10 cm, 70 cm and 160 cm of the DBACF
removal. Thirdly, the leakage of bacteria-attached particles (Fig. 1). About 10 g GAC particles were collected from each
from the DBACF is a challenge to the downstream disinfection sampling point. The UBACF and DBACF had same packing
unit. To optimize the BAC based process, the authors proposed volume of the GAC. All GAC media were “virgin” at the start of
a new process which used the up-flow BAC filter (UBACF) to the experiments. All pipes for water flow were polyvinyl-
replace the DBACF and re-located O3UBACF in-between chloride (PVC).
sedimentation and rapid sand filtration. In this paper, the The UBACF and DBACF had a constant flow rate of 70 L/h.
UBACF maintained an expansion of less than 30%, which In the UBACF, influent was introduced to the bottom and
provided a stable expanded bed while maintaining a slight exited from the top of the column. While in the DBACF,
gross carbon mixing. The increased space among the GAC influent was fed from the top and exited from the bottom The
media allowed the introduction of particle containing influent HRT of the DBACF was 8.6 min whereas the HRT of the UBACF
without significantly clogging the bed (Lin et al., 2004), which was 10.7 min because of the 25% expansion of carbon layer.
slowed down the head loss increase over time. The following The activated carbon layer of the DBACF had no expansion.
rapid sand filtration (RSF) removed the particles and turbidity The detailed parameters of the UBACF and DBACF are listed in
from the UBACF thus maintaining finished water quality. The Table 2. Pervious literature concluded that the GAC particle
previous study suggested that backwash cycle of the UBACF size made inappreciable difference of NOM removal (Velten
ran every 4 weeks (Tan and Johnson, 2001). For the same et al., 2011b), and author’s preliminary test also found that
reason, the larger space improved the bacteria homogeneous the size of GAC did not impact NOM removal by the UBACF.
distribution in whole GAC bed. However, the current litera- Therefore, the authors chose smaller GAC to fill the UBACF to
tures lack comparison of NOM removal and biological activ- maintain the designed expansion ratio.
ities of UBACF and DBACF at same operational conditions. During the first period of operation (0e63 days, start-up
In view of the above, the objective of this paper was to phase), the UBACF and DBACF were directly supplied with
evaluate the NOM removal by the UBACF and DBACF. To un- feed water from the RSF without ozonation to culture bio-film
derstand the role of bacteria contributing to the removal of forming on the GAC. During the second period of operation
NOM by the UBACF and DBACF, 454 sequencing was used to (64e126 days, steady state period of biodegradation), ozona-
characterize the microbial properties. Tag pyrosequencing tion was applied. The CODMn and turbidity of each filter
technology as next generation DNA sequencing provided a effluent had been measured since the 6th day. The last week
new method for in depth analysis of microbial diversity. This of the steady state period, DOC, UV254, molecular weight dis-
method allowed the detection of rare microbial populations in tribution (MWD), heterotrophic plate count (HPC) and 454
water samples. pyrosequencing data collection were analyzed as well.
The influent and effluent CODMn concentration was re-
ported as a function of the number of bed volumes (BV), instead
2. Materials and method of the treated volume or operation time. Under the operational
conditions of the testing, 1000 BV were equivalent to 6 days of
2.1. Experimental set-up operation. Thus, about 10,500 BV water were treated by both
the UBACF and the DBACF during the first period of operation.
All pilot studies were conducted in Chengnan Drinking Water At the end of the study, the cumulative filtration BV was
Treatment Plant (CDWTP) located in Huaian, Jiangsu Province, approximately 21,000 for both the UBACF and the DBACF.
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8 4863

Fig. 1 e Experimental set-up for treatment of O3LUp-flow BAC filter (referred as UBACF)/Down-flow BAC filter (referred as
DBACF), following conventional water treatment process.

2.2. Analytical methods weight cut offs of 1 kDa, 3 kDa, 5 kDa and 10 kDa, and each
fraction was collected for DOC determination.
2.2.1. Organic substance analysis
Chemical oxygen demand by manganese (CODMn) regulated 2.2.2. Heterotrophic plate count (HPC)
by China Drinking Water Standards was analyzed as Appropriate decimal dilutions of microbial suspensions were
described in the standard methods of Ministry of Health PRC obtained by sonication of GAC samples and water samples
(GB/T-5750.7-2006). with sterile homogenization solution (Camper et al., 1985).
Dissolved organic carbon (DOC) was measured with a TOC According to the method, homogenized samples were diluted
analyzer (SHIMADZU, TOC-V CPH, Japan). UV absorbance at further with sterile ultrapure water in triplicate and were
wavelength of 254 nm (UV254) was determined with a UV- incubated on R2A at 20  C for 7 days.
TU1901 spectrophotometer (Persee, TU-1901, China). SUVA254
was obtained by dividing UV254 by DOC. All water samples were 2.2.3. 454 Pyrosequencing and data analysis
filtered through 0.45 mm membrane (Millipore, USA) to remove 10e20 L effluents from the RSF, ozone contactor, DBACF and
water insoluble particles before DOC and UV254 test. UBACF were separately taken and filtered through Durapore
Molecular weight distribution was measured by a series 0.22 mm filter (Millipore, Ireland) using YY3009000 90 mm
of ultrafiltration membranes (Millipore, USA) with molecular

Table 2 e Operation information of UBACF and DBACF.


Table 1 e Feed water quality of ozone contactor.
Parameters Units UBACF DBACF
Parameter Unit Values
GAC depth (m) 2 2
Temperature ( C) 11.2e32.1
HRTs (min) 10.7 8.6
pH / 7.5e8.0
Filtering velocity (m/h) 14 14
CODMn (mg/L) 1.89e3.15
Expansion rate (%) 25 N/A
UV254 (cm1) 0.056e0.067
GAC effective size (d10) (mm) 0.4 0.7
Turbidity (NTU) 0.5e1.9
GAC uniformity factora / 1.9 1.6
Alkalinity (mg/L as CaCO3) 98e130
BET surface area (m2/g) 960 923
NH4 þ  N (mg/L) 0.00e0.34
Conductivity (ms/cm) 458e496 a GAC uniformity factor ¼ d60/d10.
4864 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8

Filter Holder (Millipore, USA). DNA extraction from filters 3.2. Comparison of NOM removal in steady state period
and pellets was performed using the MoBio Ultra Clean
Soil DNA Isolation kit (MoBio, USA). Sequencing was per- The ozonation was applied once the biofilm inoculation phase
formed using the Roche 454 FLX pyrosequencing platform was completed. The RSF effluent’s CODMn concentration was
in USA. 2.65 mg/L on average. The ozonation brought about an
DNA sequenced data were processed using the Quantita- approximately 12% reduction of CODMn at the O3 dose of 2 mg/
tive Insights Into Microbial Ecology (QIIME) pipeline with L. As shown in Fig. 2, the concentration of CODMn of ozonated
default settings. Sequences were clustered into Operational water, the UBACF effluent and the DBACF effluent were
Taxonomic Units (OTUs) using a minimum identities of 97% approximately 2.34 mg/L, 1.66 mg/L and 1.93 mg/L, respec-
by UCLUST software. Taxonomy assignment, Equitability, tively. The UBACF had approximately 10% higher removal
Shannon and Simpson diversity indices were processed by efficiency of CODMn than the DBACF did.
QIIME as described in a previous literature (Telias et al., 2011). The data presented in Table 3 shows that ozonation led to
The authors deposited the 454 raw data in NCBI-SRA under the 25.8% decrease in term of SUVA. While in the following the BAC
accession number SRA 058732. Hierarchical clustering was treatment, the SUVA of the UBACF effluent (2.14 m1/(mg/L))
conducted by R (v 2.14.2, http://www.r-project.org/). Addi- and the DBACF effluent (2.07 m1/(mg/L)) were higher than that
tional details of the experimental procedure are described in a of the influents of the UBACF and DBACF (1.96 m1/(mg/L)).
previous literature (Zhang et al., 2012). According to the classification made by Edzwald and Tobiason
(1999), when the SUVA value was below 2, NOM was mainly
non-humic and non-hydrophobic composition. SUVA values
between 2 and 4 indicated that NOM consisted of a mixture of
3. Results and discussion aquatic humic and other intermediate molecular weight sub-
stances. In this case, the average SUVA value of the ozonated
3.1. Start-up phase: biofilm development and filter water was close to 2.Therefore, the aromaticity of NOM in water
transition samples was not so high and a higher biodegradability of ozo-
nated water can be anticipated. In this testing, the DOC removal
The experiments were initiated by applying unozonated efficiency of the UBACF and DBACF were 37.9% and 31.5%,
influent for biofilm formation. During that time, the virgin respectively, which were lower than the previous study
GAC was exposed to the microbial community present in the (Yapsakli and Cecen, 2010). The reason might be attributed to
water, which led to a rapid initial colonization. In this study, the high SUVA value of the samples (Hozalski et al., 1999).
CODMn, rather than DOC, was chosen to indicate the service Fig. 4 shows the variation of the molecular weight distri-
condition of GAC, because CODMn was the surrogate param- bution in the different treatment unit. The O3 process reduced
eter of organic matter in drinking water in the latest national high molecular weight (HMW) fraction (>w10 kDa) and con-
standard (GB5749-2006) of China. verted them into smaller parts, resulting in an increase of the
While the CODMn concentration of RSF effluent remained low molecular weight (LMW) fraction (<w1 kDa). As antici-
relatively stable over the investigation period, the ratio of pated, the reduction was significant for the <w10 kDa fraction
CODMn in the outlet and inlet increased continuously, that in the BAC, which decreased from 1.4 mg/L to 0.6 mg/L in the
owed to a gradually decreasing adsorption capacity of the UBACF and to 0.8 mg/L in the DBACF, respectively. The LMW
GAC (Fig. 2). The CODMn removal efficiency dropped from fraction after biodegradation represented metabolic products
initial value of approximately 80% to less than 15% in 8000 BV of microorganisms living in BAC (Matilainen et al., 2006), and
operation time. When the adsorption capacity of the acti- the UBACF effluent had more LMW fraction than the DBACF
vated carbon became saturated and GAC filter functioned effluent. Thus, the UBACF had better NOM removal that might
primarily as a BAC filter, the CODMn was biologically be contributed to better biodegradation effect than the DBACF.
removed. The authors concluded that adsorption could be Because the HMW fraction of the DOM was the major source
negligible after 10,500 BV (z63 days) and the combination of of UV254 (Lin et al., 2009), the BAC had very low removal effi-
bacterial respiration and biomass assimilation predomi- ciency of the HMW parts, and the non-UV254 DOM was rela-
nantly accounted for the CODMn removal from that time tively easily removed by the BAC, it is understandable that
because the removal efficiencies of CODMn were maintained SUVA values were raised after both the UBACF and the DBACF.
at a relatively stable level, referred to as the steady state of
biodegradation (Velten et al., 2011a). From Fig. 3, the DBACF 3.3. The impact of temperature on the removal of CODMn
was found to remove more turbidity by about 0.3 NTU than
the UBACF did in the start-up phase. Some parts of CODMn As discussed by the literature (Simpson, 2008), fluctuations in
were combined with particles and was removed with the temperature of the source water may impact the biological
turbidity together. Compared to the UBACF, the DBACF had a activity in the biofilm. In this study, the main purpose was to
little higher CODMn removal efficiency during the first compare the organics removal by the DBACF and the UBACF in
3500 BV operation, which might partially contribute to its same condition. The running testing of the DBACF and the
stronger interception of particles and turbidity. With the UBACF were performed with the same source water quality,
biofilm formed on the GAC media, the removal efficiency of and as such the temperature impact on the DBACF and the
CODMn by the UBACF became higher than that by the DBACF, UBACF was similar in some degree. During the start-up phase,
which was contributed to the more efficient biodegradation the removal of CODMn was mainly contributed to the
in the UBACF. adsorption (Fig. 2). The increase of the temperature of the
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8 4865

Fig. 2 e CODMn concentrations of influent (C0) and effluent (C ) of the UBACF/DBACF as a function of bed volumes (BV). The
influent of the UBACF/DBACF was unozonated water during start-up phase. And the influent of the UBACF/DBACF became
ozonated water during the steady state period.

influent from 11.2  C to 23.7  C did not increase the removal measured the biomass along the depth of GAC bed and the
efficiency of CODMn. On the contrary, the removal efficiency finished water of the UBACF and DBACF. The results of the
decreased because the GAC was gradually saturated with HPCs in the effluents of the UBACF and DBACF were plotted in
CODMn removed. During the steady state period, the temper- Fig. 5. The chemical destruction of the bacterial cells by ozone
ature of the influent was maintained to 22.5  C32.1  C, but caused the significant reduction of HPCs in ozone contactor
the removal efficiencies did not closely vary with the change effluent. HPCs in the UBACF effluent were 70e100 folds higher
of the temperature (Fig. 2). It was because the removal of than in the corresponding ozone effluent, while a 4565 folds
CODMn was mainly due to biodegradation by the biofilm on the increase was found in the DBACF effluent. It clearly shows
GAC media. The temperature at 22.5  C32.1  C was favorable that the detached bacteria number in the finished water of the
to the bioactivity and bacteria community structure at sta- UBACF was much higher than that of the DBACF.
tionary phase could maintain a stable removal of CODMn. Fig. 6 illustrates the profile of HPCs attached to BAC parti-
cles. From these observations, the conclusion can be drawn
that the UBACF and DBACF had different microbial distribu-
3.4. Biomass formed in the UBACF and DBACF
tion patterns, and the DBACF contained more attached bac-
teria than the UBACF. The HPCs in the DBACF decreased with
Because the biomass in the biologically activated carbon filter
the increase of the GAC filter depth because the bacteria
has a great impact on the removal of NOM, the authors had
growing in biofilms were influenced by nutrient limitation
(Servais et al., 1994). The highest biomass concentration of the
UBACF was found in the middle of the column while that of
the DBACF was found at the top layer. The highest biomass
concentration of the UBACF was lower than that of the DBACF,
which was attributed to the relatively higher friction induced
between particles in the UBACF. In addition, the lower
biomass concentration in the bottom of the UBACF was due to

Table 3 e Variation of water quality parameters in


treatment process.
RSF Ozone UBACF DBACF

SUVA 2.64  0.07 1.96  0.07 2.14  0.08 2.07  0.04


(m1(mg/L))
Turbidity (NTU) 1.0  0.4 1.4  0.7 0.6  0.2 0.3  0.1
Fig. 3 e The variation of turbidity in the UBACF/DBACF DOC (mg/L) 2.35  0.26 2.09  0.31 1.20  0.21 1.35  0.16
effluent during start-up phase.
4866 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8

Fig. 6 e Biofilm properties (defined as number of HPCs) of


BACF depending on bio-filter depth. (GAC samples were
drawn from h [ 10, 160 and 190 cm for UBACF and h [ 10,
Fig. 4 e DOC concentration and constituent distribution
70 and 160 cm for DBACF).
after each treatment unit.

3.5. Bacterial community analysis


the presence of residual ozone in the influent (Urfer and Huck,
2001). The residual ozone decomposed rapidly prior to In this Study, bacterial communities of each unit’s effluent
entering the DBACF because the intake of the DBACF was open were elucidated by 454-pyrosequencing, which generated
to the air. However, the residual ozone decomposed relatively 18,545 DNA gene sequences from 4 samples and separated
slowly prior to entering the UBACF because the intake of the into 911 OTUs. The authors calculated the Good’s Coverage
UBACF was not open to the atmosphere. Estimator on the OTUs from each sample to assess the di-
In our study, the thickness of biofilm was approximately versity captured with the samples. Results indicated that this
105107 HPC/cm2. Pervious study (Gagnon and Huck, 2001) test captured 96%e98% of the species of all the samples. The
proposed the thickness of biofilm between 106e108 HPCs/cm2 detailed information is shown in Table 4.
did not limit rate of mass transfer either internally or exter- The Shannon diversity index, Simpson’s diversity index
nally in biofilm to utilize biodegradable organic matter (BOM). and equitability were calculated for each sample. When
And Monod half-velocity constant and Monod maximum treated by both the UBACF and the DBACF, the water exhibited
substrate utilization rate were the most significant parame- a higher diversity and equitability. Furthermore, the UBACF
ters affecting the BOM removal (Badriyha et al., 2003) which effluent had higher diversity and equitability than the DBACF
were related to biological activity. In the UBACF, the accu- effluent. The evenness in the structure of microbial diversity
mulated extracellular metabolites were easily washed out ensured that the community had more capacity to use its
allowing for enhanced biological activity, which led to the varied array of metabolic pathways (Wittebolle et al., 2009).
higher removal efficiency of NOM than in the DBACF. Thus, the higher equitability improved bacterial community’s
stability and increased NOM removal capacity (Boon et al.,
2011; Werner et al., 2011; Zhang et al., 2011). It was likely
that the UBACF had higher biodegradation ability of NOM
because the UBACF also had more even distribution of bac-
terial species than the DBACF.
Hierarchical cluster analysis of bacterial community is
shown in Fig. 7. The bacterial community relationship was
very close between the RSF effluent and the ozonated water,

Table 4 e The coverage, diversity and equitability indices


from pyrosequencing analysis.
Read OTUs Good’s Simpson Shannon Equitability
(%)

RSF 5210 304 96.9 0.66 3.40 0.41


Ozone 3856 307 96.0 0.73 4.02 0.49
UBACF 3855 303 97.0 0.96 6.08 0.74
DBACF 5624 409 97.2 0.84 5.09 0.59
Fig. 5 e HPC in the effluent of each treatment unit.
w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8 4867

in drinking water system. In this study, UBACF and DBACF


mainly decreased the relative abundance of Alphaproteobac-
teria but increased the relative abundance of Betaproteobacteria,
which was similar to the result in pervious literature (Pinto
et al., 2012). It suggested that Alphaproteobacteria was more
easily captured by BAC and Betaproteobacteria more easily
leaked from BACF with the effluent.
Evaluation at the genus level showed that members of the
genera Pedomicrobium was the predominant bacterial groups
of Alphaproteobacteria in the effluent of the RSF, ozonation and
the UBACF, and the relative abundance were 57.8%, 50.8% and
14.7%, respectively. In contrast, Bacteria of the genus Hydro-
genophaga constituted the main population of Betaproteobac-
teria in the DBACF effluent and its relative abundance reached
38.8%.
Fig. 7 e Hierarchical cluster analysis of bacterial
According to Guidelines for Drinking-water Quality (fourth
community.
Edition, GDQ) from World Health Organization (WHO), Bur-
kholderia, Acinetobacter and Pseudomonas are considered to be
pathogens, which affect human health. In this study, they
which meant the ozone dosage of 2 mg/L did not change the were detected in finished water of the DBACF. Also, Bur-
bacterial community composition but changed relative kholderia was found in the UBACF effluent. The total relative
abundance. As compared to the DBACF, the microbial com- abundance of pathogens in the DBACF effluent and the UBACF
munity composition of the UBACF effluent was closer to the effluent were 2.8% and 0.6%, respectively. Hence, the risk of
RSF effluent, which was likely due to the more frequent pathogens in the BAC effluent should be closely monitored
backwash of the DBACF. and controlled. In addition, the ratio of Cyanobacteria in water
Fig. 8 shows the class composition of the four samples. The filtered by the DBACF and the UBACF were 10.5% and 15.9%,
four most abundant bacterial classes in the RSF effluent in this respectively. Cyanobacteria are concerned by WHO because
study were Alphaproteobacteria (62.8%), Actinobacteridae (9.9%), some species of Cyanobacteria produce cyanotoxins. However,
Holophagae (6.3%) and Betaproteobacteria (5.2%), which are also these three kinds of Cyanobacteria species recovered in this
major bacterial composition in ozonated water. But the study were not included in GDQ.
abundance varied to 55.5%, 9.6%, 6.4% and 6.5%, respectively.
The UBACF effluent was characterized by the domination of
Alphaproteobacteria (28.5%), Betaproteobacteria (17.4%) and Cya- 4. Conclusions
nobacteria (15.9%). Betaproteobacteria (46.8%), Alphaproteobac-
teria (19.5%) and Gammaproteobacteria (11.2%) were the The study compared the two different BAC filter types, UBACF
dominant groups in the DBACF effluent. According to the and DBACF, in NOM removal and microbial properties. The
literature (Hong et al., 2010; Revetta et al., 2010; Kwon et al., conclusions of this study include:
2011; Henne et al., 2012), Alphaproteobacteria, Betaproteobac-
teria and Gammaproteobacteria were found as dominant group 1. At the adsorption stage, compared to the UBACF, the DBACF
had a little higher removal of NOM in term of the CODMn.
Whereas at the steady state of biodegradation, the UBACF
averagely had 10% higher removal of NOM in CODMn and
6.4% higher in DOC, because the UBACF had stronger
biodegradation ability.
2. The highest biomass concentration in the UBACF was
found in the middle of the filter. While in the DBACF, the
highest biomass concentration was found on the upper part
of the filter. The DBACF had more attached biomass while
the UBACF effluent had more suspended bacteria. The
accumulated extracellular metabolites in the UBACF were
easily washed out allowing for enhanced biological activity,
which led to the UBACF having stronger biodegradation
ability than the DBACF.
3. The effluent of the UBACF and DBACF had different bacte-
rial community compositions. The UBACF effluent had
higher diversity and equitability than the DBACF effluent.
4. Alphaproteobacteria and Betaproteobacteria were dominant
groups in the BAC effluent. Alphaproteobacteria was more
Fig. 8 e Microbial population classification in the effluent of easily captured by the BAC and Betaproteobacteria would
each treatment (defined as class). more easily leak from the BAC with the water flow.
4868 w a t e r r e s e a r c h 4 7 ( 2 0 1 3 ) 4 8 6 1 e4 8 6 8

Nishijima, W., Speitel Jr., G.E., 2004. Fate of biodegradable


Acknowledgments dissolved organic carbon produced by ozonation on biological
activated carbon. Chemosphere 56, 113e119.
This work was supported by National Natural Science Foun- Pinto, A.J., Xi, C.W., Raskin, L., 2012. Bacterial community
dation of China (51178240) and the Grand Water Project of structure in the drinking water microbiome is governed by
filtration processes. Environmental Science and Technology
China (No. 2012ZX07404-002).
46, 8851e8859.
Randtke, S.J., 1988. Organic contaminant removal by coagulation
and related process combinations. Journal of the American
references Water Works Association 80, 40e56.
Revetta, R.P., Pemberton, A., Lamendella, R., Iker, B., Domingo, J.,
2010. Identification of bacterial populations in drinking water
Badriyha, B.N., Ravindran, V., Den, W., Pirbazari, M., 2003. using 16S rRNA-based sequence analyses. Water Research 44,
Bioadsorber efficiency, design, and performance forecasting 1353e1360.
for alachlor removal. Water Research 37, 4051e4072. Schideman, L.C., Snoeyink, V.L., Marinas, B.J., Ding, L.,
Boon, N., Pycke, B.F.G., Marzorati, M., Hammes, F., 2011. Nutrient Campos, C., 2007. Application of a three-component
gradients in a granular activated carbon biofilter drives competitive adsorption model to evaluate and optimize
bacterial community organization and dynamics. Water granular activated carbon systems. Water Research 41,
Research 45, 6355e6361. 3289e3298.
Camper, A.K., Lechevallier, M.W., Broadaway, S.C., McFeters, G.A., Servais, P., Billen, G., Bouillot, P., 1994. Biological colonization of
1985. Evaluation of procedures to desorb bacteria from granular activated carbon filters in drinking-water treatment.
granular activated carbon. Journal of Microbiological Methods Journal of Environmental Engineering 120, 888e899.
3, 187e198. Simpson, D.R., 2008. Biofilm processes in biologically active
Chen, C., Zhang, X., He, W., Lu, W., Han, H., 2007. Comparison of carbon water purification. Water Research 42, 2839e2848.
seven kinds of drinking water treatment processes to enhance Tan, L., Johnson, W., 2001. Removing organic color and by-
organic material removal: a pilot test. Science of the Total products from groundwater with ozone and pressurized
Environment 382, 93e102. biologically-active filtration. Ozone: Science & Engineering 23,
Edzwald, J.K., Tobiason, J.E., 1999. Enhanced coagulation: US 393e400.
requirements and a broader view. Water Science and Telias, A., White, J.R., Pahl, D.M., Ottesen, A.R., Walsh, C.S., 2011.
Technology 40, 63e70. Bacterial community diversity and variation in spray water
Gagnon, G.A., Huck, P.M., 2001. Removal of easily biodegradable sources and the tomato fruit surface. BMC Microbiology 11.
organic compounds by drinking water biofilms: analysis of Urfer, D., Huck, P.M., 2001. Measurement of biomass activity in
kinetics and mass transfer. Water Research 35, 2554e2564. drinking water biofilters using a respirometric method. Water
Henne, K., Kahlisch, L., Brettar, I., Hofle, M.G., 2012. Analysis of Research 35, 1469e1477.
structure and composition of bacterial core communities in Velten, S., Boller, M., Köster, O., Helbing, J., Weilenmann, H.,
mature drinking water biofilms and bulk water of a citywide Hammes, F., 2011a. Development of biomass in a drinking
network in Germany. Applied and Environmental water granular active carbon (GAC) filter. Water Research 45,
Microbiology 78, 3530e3538. 6347e6354.
Hong, P.Y., Hwang, C.C., Ling, F.Q., Andersen, G.L., Velten, S., Knappe, D.R.U., Traber, J., Kaiser, H., von Gunten, U.,
LeChevallier, M.W., Liu, W.T., 2010. Pyrosequencing analysis Boller, M., Meylan, S., 2011b. Characterization of natural
of bacterial biofilm communities in water meters of a drinking organic matter adsorption in granular activated carbon
water distribution system. Applied and Environmental adsorbers. Water Research 45, 3951e3959.
Microbiology 76, 5631e5635. Werner, J.J., Knights, D., Garcia, M.L., Scalfone, N.B., Smith, S.,
Hozalski, R.M., Bouwer, E.J., Goel, S., 1999. Removal of natural Yarasheski, K., Cummings, T.A., Beers, A.R., Knight, R.,
organic matter (NOM) from drinking water supplies by ozone- Angenent, L.T., 2011. Bacterial community structures are
biofiltration. Water Science and Technology 40, 157e163. unique and resilient in full-scale bioenergy systems.
Kwon, S., Moon, E., Kim, T.S., Hong, S., Park, H.D., 2011. Microbiology 108, 4158e4163.
Pyrosequencing demonstrated complex microbial communities Wittebolle, L., Marzorati, M., Clement, L., Balloi, A.,
in a membrane filtration system for a drinking water treatment Daffonchio, D., Heylen, K., De Vos, P., Verstraete, W., Boon, N.,
plant. Microbes and Environments 26, 149e155. 2009. Initial community evenness favours functionality under
Lin, D., Thömmes, J., Kula, M., Hubbuch, J.J., 2004. The influence of selective stress. Nature 458, 623e626.
biomass on the hydrodynamic behavior and stability of Yapsakli, K., Cecen, F., 2010. Effect of type of granular activated
expanded beds. Biotechnology and Bioengineering 87, carbon on DOC biodegradation in biological activated carbon
337e346. filters. Process Biochemistry 45, 355e362.
Lin, T., Chen, W., Wang, L., 2009. Removal characters of ozone- Zhang, D., Li, W., Zhang, S., Liu, M., Zhao, X., Zhang, X., 2011.
biological activated carbon process for typical pollutants in Bacterial community and function of biological activated
southern brooky regions of china. Huanjing Kexue 30, carbon filter in drinking water treatment. Biomedical and
1397e1401. Environmental Sciences 24, 122e131.
Matilainen, A., Vieno, N., Tuhkanen, T., 2006. Efficiency of the Zhang, M., Liu, W., Nie, X., Li, C., Gu, J., Zhang, C., 2012. Molecular
activated carbon filtration in the natural organic matter analysis of bacterial communities in biofilms of a drinking
removal. Environment International 32, 324e331. clearwell. Microbes and Environments 27, 443e448.

Potrebbero piacerti anche