- Email: [email protected]
Optimizing and real-time control of biofilm formation, growth and renewal in denitrifying biofilter
Accepted Manuscript Optimizing and Real-time Control of Biofilm Formation, Growth and Renewal in Denitrifying Biofilter Xiuhong Liu, Hongchen Wang, Feng Long, Lu Qi, Haitao Fan PII: DOI: Reference:
S0960-8524(16)30232-2 http://dx.doi.org/10.1016/j.biortech.2016.02.095 BITE 16145
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
20 November 2015 19 February 2016 22 February 2016
Please cite this article as: Liu, X., Wang, H., Long, F., Qi, L., Fan, H., Optimizing and Real-time Control of Biofilm Formation, Growth and Renewal in Denitrifying Biofilter, Bioresource Technology (2016), doi: http://dx.doi.org/ 10.1016/j.biortech.2016.02.095
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
2
Optimizing and Real-time Control of Biofilm
3
Formation, Growth and Renewal in Denitrifying
4
Biofilter
5 6
Xiuhong Liu, Hongchen Wang *, Feng Long, Lu Qi, Haitao Fan
7
8 9
School of Environment & Natural Resources, Renmin University of China, Beijing, China,100872;
10
11
Corresponding author: Hongchen Wang
12
Phone: + 86-10-62510853
13
E-mail: [email protected]
14 15
1
16
ABSTRACT
17
A pilot-scale denitrifying biofilter (DNBF) with a treatment capacity of 600 m3/d was used
18
to study real-time control of biofilm formation, removal and renewal. The results showed
19
biofilm formation, growth and removal can be well controlled using on-line monitored
20
turbidity. The status of filter layer condition can be well indicated by Turb break points on
21
turbidity profile. There was a very good linear relationship between biofilm growth degree
22
(Xbiof) and filter clogging degree (Cfilter) with R2 higher than 0.99. Filter layer clogging
23
coefficient (Yc) lower than 0.27 can be used to determine stable filter layer condition. Since
24
variations of turbidity during backwash well fitted normal distribution with R2 higher than
25
0.96, biofilm removal during backwash also can be well optimized by turbidity. Although
26
biofilm structure and nirK-coding denitrifying communities using different carbon sources
27
were much more different, DNBF was still successfully and stably optimized and real-time
28
controlled via on-line turbidity.
29
30 31
KEY WORDS
32
Biofilm; formation; turbidity; real-time control; denitrifying biofilter
33
2
34
1. Introduction
35
Wastewater reclamation is one of the most effective ways to solve both water shortage and
36
pollution problems. Many wastewater treatment plants in water deficient regions, especially
37
large and medium-sized cities, have been up-graded to produce reclaimed water, and
38
advanced nitrate removal is required. Because denitrifying biofilter (DNBF) has advantages
39
of high treatment efficiency, small footprint, and low shock loading impact, it is widely
40
used for advanced nitrate removal. However, because biofilm growth cannot be easily
41
monitored, many biofilters do not provide stable performance. Moreover, some biofilters
42
can not be normally operated due to frequent clogging (Leverenz et al., 2009; Snowball,
43
2006) (Figure S1), which limits the engineering application and development of biofilter.
44
In fixed-film wastewater treatment systems (Corona et al., 2013; Ji et al., 2014; Wang et
45
al., 2015), biofilm is used to remove pollutants in wastewater; whereas, in other fields
46
including membrane treatment processes (Miura et al., 2006; Pang et al., 2005), drinking
47
water or reclaimed water distribution systems (Mathieu et al., 2014; Yang et al., 2015), and
48
food processing industry (Simões et al., 2010), biofilm formation is harmful and should be
49
avoided. It was reported that biofilm formation comprised a series of complicated steps
50
(Giaouris et al., 2014; Simões et al., 2010) including cell deposition, cell adsorption and/or
51
desorption, cell–cell signaling and EPS production, replication and growth, secretion of
52
biofilm matrix, and biofilm detachment or sloughing. To date, there is no effective method
53
to accurately measure or determine biofilm growth condition in a biofilter, especially in a
54
DNBF. Generally, during the start-up stage of a DNBF, biofilm was naturally cultivated
55
with nitrate and organic carbon. However, since biofilm can not be practically monitored or
56
measured during treatment, it is hard to determine the biofilter run time between
3
57
backwashes (backwash cycle), the backwash strength and duration. This may result in over
58
backwash and energy cost, or biofilter clogging if backwash is not performed in time. In
59
the last decade, most studies on biofilter process were focused on optimizing backwash
60
procedures (Amburgey and Amirtharajah, 2005; Slavik et al., 2013; Snowball, 2006), while
61
limited studies have been conducted on optimizing the biofilter operation based on the
62
extent of biofilm growth and removal.
63
Many on-line biofilm monitoring methods, such as differential turbidity measurement
64
(Métadier and Bertrand-Krajewski, 2012), microwaves (Saber et al., 2013), multi-channel
65
impedimetric and amperometric sensor (Pires et al., 2013) and thermal sensors (Reyes-
66
Romero et al., 2014), have been used to monitor biofilm growth. Among these methods,
67
turbidity is suitable for industrial applications. It was used to calculate urban storm water
68
pollutant concentrations and loads (Métadier and Bertrand-Krajewski, 2012). Backwash is
69
the key operational step to control biofilm growth, removal and renewal in treatment plants
70
that employ biofilter/filter as major unit process. However, the effective method to monitor
71
biofilm growth on-line in biofilter has not been well developed, and very limited studies
72
have been conducted on real-time control and optimization of biofilter-based biofilm
73
formation and growth.
74
Therefore, this study aimed to: a) determine the relationship between biofilm formation
75
and turbidity variation during the biofilm cultivation stage of a DNBF; b) control and
76
optimize biofilm growth, removal and renewal during filtration and backwash processes in
77
the DNBF via on-line monitored turbidity; c) test the stability of real-time control of
78
biofilm growth, removal and renewal in the DNBF using different typical carbon sources.
4
79
2. Materials and methods
80
2.1 Pilot-scale and lab-scale DNBF
81
A pilot-scale DNBF located at Beijing Gaobeidian Municipal Wastewater Treatment Plant
82
(WWTP) with a maximum treatment capacity of 600 m3/d was used in this study. The
83
DNBF was packed with expanded clay particles (4-6 mm) at a bed depth of 2.5 m (Table
84
S1). The pilot-scale treatment system consisted of a DNBF reactor, a carbon dosage system,
85
and a backwash system (Figure S2).
86
The control consists of four parts: detectors, a computer, interface cards and control units.
87
Nitrate (NO3-), turbidity, pH, dissolved oxygen (DO) and pressure sensors were installed in
88
the top section and the bottom of DNBF reactor, respectively. The values of the NO3-,
89
turbidity, pH, and DO were recorded every 0.5–10 minute and then transferred to a
90
programmable logic controller (PLC) and process control system (PCS). PCS was
91
programmed according to the control logic.
92
A lab-scale DNBF with a total volume of 30 L and filter media height of 1.1m was used
93
to study biofilm formation further. Except for volume and media height, the lab-scale
94
DNBF was the same as the pilot-scale DNBF.
95
2.2 Secondary effluent composition
96
The secondary effluent of Gaobeidian municipal WWTP that employs an anoxic/oxic (A/O)
97
process was continuously fed to the DNBF. The concentrations of the soluble chemical
98
oxygen demand (SCOD), NH4+-N, NO3--N, turbidity, and SS in the secondary effluent were
99
in the range of 23.28-76.03 mg/L, 0.33-2.36 mg/L, 20.40-34.12 mg/L, 0.48-13.10 NTU,
100
and 1.33-16.15 mg/L, respectively.
101
5
102
2.3 Experimental design
103
Experiments were conducted for 210 days, which was divided into 3 phases. Experimental
104
procedures, parameters, aims, and backwash operation were shown in Table 1 and tableS2.
105
In the first phase, the natural biofilm cultivation method was used in the pilot-scale and lab-
106
scale DNBF. Since biofilm formation in the pilot-scale DNBF was only operated for one
107
time and the filter media is difficultly taken out to observation, the lab-scale DNBF was
108
operated with Filtration velocity (FV) of 2.38m/h only to further confirm the results
109
obtained in the pilot-scale DNBF. In the second phase, DNBF was firstly operated for 40
110
days to establish backwash strength. Thereafter, variations of the effluent turbidity during
111
long-term filtration and backwash operation were studied on the following 38 days and 33
112
days, respectively. In the third phase, the stability of long-term operation of DNBF with
113
real-time control using turbidity as parameter was tested using two typical carbon sources.
114
When carbon source changed from sodium acetate to methanol, nitrate removal efficiency
115
of DNBF was recovered for 30 days.
116 117
2.4 Analysis
118
COD, NH4+-N, NO3--N, NO2--N, TN, PO43--P, and total phosphorus (TP) were measured
119
according to the standard methods given in APHA (APHA, 1998). DO, turbidity, NO3--N
120
and pressure were measured on-line using oxygen, turbidity, NO3--N meters (LDO,
121
SOLITAX sc, NITRATAX plus sc, Hach Company, USA) and pressure sensor,
122
respectively.
123
Extracellular Polymeric Substances (EPS) was extracted using formaldehyde-NaOH
124
according to the method described by Liu and Fang (Liu and Fang, 2002). Pre-treatment of
6
125
filter media for SEM observation was carried out using the modified method described by
126
Wang et al. (2013) (Wang et al., 2013). Visualization of the samples was conducted using a
127
Scanning Electron Microscope (Hoskin Scientific, Tokyo, Japan).
128
2.5 DNA extraction, PCR, cloning, sequencing and phylogenetic analysis
129
Filter media samples were taken from DNBF on day 100 and 203. Biofilm was removed
130
from filter media by shaker. The collected samples were freeze-dried in Labconco Freezone
131
1 L (Labconco, USA) and stored at −20 °C.
132
0.05–0.10 g of dry sludge sample was taken to extract genomic DNA using a FastDNA
133
SPIN Kit for soil (Qiagen, CA, USA). 1ul of extracted DNA was taken to measure DNA
134
concentration by Nanodrop Spectrophotometer (ND-1000, Thermo Fisher Scientific, USA).
135
Fragments of nirK genes were amplified with the primer set FlaCu/R3Cu (473 bp) as
136
described by Hallin et al. (1999) (Hallin and Lindgren, 1999).
137
The PCR was performed in a C1000TM thermal cycler (BioRad, USA). Cycle conditions
138
for the amplification were as follows: 2 min at 94°C; 35 cycles with each cycle consisting
139
of 30 sec at 94°C, 1 min at 57°C, and 1 min at 72°C; followed by a final 10-min extension
140
at 72°C. PCR products were visualized on 1.5% (w/v) agarose gel electrophoresis to
141
confirm the product size, and then, purified by Wizard® SV Gel and PCR Clean-Up
142
System (Promega, Madison, USA). The detailed information on cloning and sequencing
143
were described by Gao et al. (2014) (Gao et al., 2014).
144
All the sequences and their reference sequences obtained from NCBI BLAST were
145
aligned using MEGA 6.0 software (Tamura et al., 2013). The sequences sharing 98%
146
similarity were grouped into the same operational taxonomic unit (OTU) using software
7
147
mother (Schloss et al., 2009). Phylogenetic trees were generated by neighbor-joining (NJ)
148
with the Jukes–Cantor correction in MEGA (Tamura et al., 2013).
149
2.6 Calculations
150
Biofilm growth degree (Xbiof), filter clogging degree (Cfilter) and filter layer clogging
151
coefficient (Yc), were calculated based on the biomass yield coefficient and total effluent
152
turbidity as described in equation(1) – (3) , respectively. NT
{
Y NO3 − N ∑ [NO3 − N ]in - [NO3 − N ]out 153
}
n ∆t
× ∆t × Q n∆t
n =1
X biofilm =
(1)
Vfilter
154
where, YNO3-N is sludge yield coefficient (0.74 gVSS/gNO3-N), [NO3-N]in and [NO3-N]out
155
are nitrate concentrations in the influent and effluent (g/m3), n is data points on the profiles
156
of on-line monitored nitrate, T is the time of on-line monitored nitrate, NT is the data points
157
at the time of T, ∆t is the on-line monitored interval (day), Q is the influent flow rate (m3/d),
158
Vfilter is the volume of filter media (m3). NT
∑ {[Turb] - [Turb ] } in
159
160
C filter =
out n ∆t
× ∆t × Q n∆t
(2)
n =1
Vfilter
where, [Turb]in and [Turb]out are turbidity concentrations in the influent and effluent(NTU). NT
161
Yc =
∑ {[Turb] - [Turb] } in
1 YNO3 − N
out n∆t
n =1 NT
∑ {[NO
3
− N ]in - [NO3 − N ]out
n =1
8
}
n∆t
(3)
162
where, Yc is the slope of Cfilter/Xbiof.
163 164
3. Results and discussion
165
3.1 Correlation between biofilm formation and effluent turbidity during start-up of
166
DNBF
167
Figure 1 showed variations of COD, NO3--N, turbidity and pressure during biofilm
168
cultivation in pilot-scale DNBF. Because almost no denitrifying bacteria was cultivated in
169
the first 2 days, very small amount of nitrate was removed and the added carbon source
170
(CODadd=CODcarbon-CODin) was almost not consumed. From the 3rd day, nitrate removal
171
efficiency increased to 55%, and CODadd was still not sufficiently utilized. Meanwhile,
172
gases emitted from the effluent, indicating that denitrification occurred. On the following 2
173
days, CODadd was sufficiently utilized with removal efficiency higher than 98%. Figure S3
174
showed variations of CODremoved, NO3--Nremoved, turbidity and pressure during biofilm
175
cultivation in the lab-scale DNBF. In the first two days, variations of the removed COD,
176
nitrate and turbidity in lab-scale DNBF were very similar to these in the pilot-scale DNBF.
177
From day 3 to 4, with COD and nitrate slightly removed, turbidity gradually increased.
178
SEM results also found cells absorbed and grew on the rough surface of filter media
179
(Figure S4). Some of cells aggregated together and formed micro colonies. These results
180
indicated that with the cells replication and growth, since some discohesive cells and
181
produced EPS were desorbed from filter media into water (Busscher et al., 2010) under the
182
shear forces of the moving water with the filter media surface, turbidity were gradually
183
increased. On the flowing four days, with the increase of COD and nitrate removal,
184
turbidity was still gradually increased. On day 8, SEM results also demonstrated the
9
185
relatively integrated and stable biofilm were formed. Meanwhile, because the loose
186
structure biofilm were detached from filter media due to the hydraulic shear force, small
187
amount of macroscopic biofilm fragments were found in the effluent. After that, since the
188
excessive growth of biofilm caused the filter layer blocked, with the matured and aged
189
biofilm slough off and new biofilm formation, the hydraulic sheer force fluctuated caused
190
turbidity fluctuation. Therefore, based on the above results in pilot-scale and lab-scale
191
DNBF, turbidity in the effluent can well indicated biofilm formation and growth. To date,
192
few studies used turbidity to monitor biofilm formation and growth in a wastewater
193
treatment system.
194
After the biofilm density gradually increased during the cultivation stage, backwash was
195
performed to scour the excessive biomass under higher hydrodynamic shear force, to renew
196
the biofilm with new denitrifying bacteria (Table S2). In Phase 2, on the first 40 days, if
197
backwash strength was controlled too low (Figure S5), during filtration, not only the
198
effluent nitrate was unstable even after adding sufficient carbon source, but also turbidity
199
was fluctuated significantly. Some large SS aggregates were also observed in the effluent
200
(Figure S6). Furthermore, because of more frequent filter clogging, backwash operation
201
became more difficult. However, high backwash intensity might also result in filter media
202
loss (Figure S7), and mix the supporting layer with filter media layer (Figure S8) in the
203
reclaimed WWTP. Therefore, appropriate backwash is the key to ensure long-term
204
operation of the biofilter.
205
Since biofilter is a closed system, the biofilm on the biofilter media can not be measured
206
directly. In addition, biofilter clogging has a time-lag property, which even increases the
10
207
difficulty to operate and maintain a biofilter. These undesirable backwash control methods
208
mainly resulted from not timely monitoring biofilm growth and filter layer clog condition.
209
In wastewater/water treatment plants, biofilters are normally operated based on operator's
210
experience, and these undesirable backwash control methods often occur, which slow the
211
application and development of biofilter. To date, there is no effective way to on-line
212
monitoring biofilm growth degree and determine filter media clog degree in fundamental
213
and applied research.
214
3.2 Correlation between effluent turbidity and biofilm renewal and removal during
215
normal operation
216
Based on the obtained results during biofilm formation, turbidity might also be used to
217
indicate biofilm growth and removal during filtration and backwash operation. To validate
218
this hypothesis, turbidity and nitrate concentrations in the effluent were monitored online
219
during filtration and backwash processes.
220
3.2.1 Variation in effluent turbidity indicating biofilm growth and filter layer condition
221
during filtration
222
Figure 2a showed typical variations of the effluent turbidity, CODadd, influent nitrate, and
223
effluent nitrate during long-term filtration. With the biodegradation of COD and the
224
reduction of nitrate, biofilm grew gradually and became thicker. Meanwhile, filter layer
225
was gradually blocked. At the beginning of filtration, the effluent nitrate varied with
226
CODadd and the effluent turbidity increase slightly. However, after continuous filtration for
227
about 49 h, since some microcolonies and/or biofilm fragments were sloughed off from the
11
228
surface of biofilm, both the effluent turbidity and nitrate increased significantly. Turb
229
break-1 and Nitrate break-1 points both appeared on the turbidity and ef-nitrate profiles,
230
respectively. Thereafter, the effluent nitrate still varied with CODadd, while the effluent
231
turbidity turned increase to decrease because of biosorption and low nitrate removal amount;
232
whereas, in the 63.25 h, because of higher hydraulic shear force caused by more serious
233
filter layer blockage, the effluent turbidity increased sharply again, and Turb break-2 point
234
appeared on the effluent turbidity. Subsequently, more and more filter layers were blocked
235
due to biofilm growth, and as a result, both the effluent turbidity and nitrate fluctuated
236
significantly. At the same time, both Turb shock and nitrate shock points appeared on the
237
turbidity and effluent nitrate profiles, respectively. Therefore, based on the above results,
238
Turb break points on turbidity profile were suitable for indicating the status of filter layer
239
condition, which can be divided into four phases, including stable, intergraded, clogged and
240
damaged phases.
241
Figure 2b showed relationships between Xbiof and Cfilter. It was found that Cfilter increased
242
with the increase of Xbiof. In each phase, there was a very good linear relationship between
243
Xbiof and Cfilter with R2 higher than 0.99. Yc kept constant in each phase, while, Yc
244
increased with the clogging status of biofilter transferred from one phase to the next.
245
Especially, Yc increased sharply from 0.54 in the clogging phase and to 2.46 in the
246
damaged phase. These results indicate that Yc can be used to determine the status of filter
247
layer clogging degree.
248
Therefore, during normal operation, if backwash cycle was controlled on-line between
249
Turb break-1 point and Turb break-2 point, not only DNBF kept stable operation and
250
avoided filter layer clogging, but also carbon dosage can be well controlled by the on-line
12
251
monitored effluent nitrate. Even when turbidity can not be monitored on-line for a long
252
time, filter layer clogging degree still can be determined by calculating Yc through
253
monitoring the effluent nitrate and turbidity for the final 2-3 h of filtration.
254
3.2.2 Effluent turbidity indicating the biofilm removal degree during backwash
255
Figure 3 showed the variations in the effluent turbidity and nitrate concentration during
256
filtration and backwash processes. During the filtration, when the Turb break-1 point lasted
257
for 30 min, the biofilter transformed from the stable phase to the intergraded phase. Then
258
the filtration was stopped and backwash was performed. During the air backwash, turbidity
259
increased significantly. Hereafter, during air + water backwash, most biofilm was detached
260
and carried out by backwash water. When the biofilm almost was completely detached,
261
turbidity reached a peak and then decreased; turbidity apex points appeared on the turbidity
262
profile. Thereafter, during water backwash, backwash water carried the residual detached
263
biofilm out of filter layer. Meanwhile, turbidity decreased sharply, followed by a gradual
264
decrease. After that, turbidity kept constant, which indicated that the detached biofilm was
265
completely removed by backwash water. Among three phases of backwash, air + water
266
backwash was the most intensive and decisive phase to keep the equilibrium between the
267
excessive biofilm growth and biofilm removal, in which, more than half of the detached
268
biofilm was removed. Furthermore, during air + water backwash, Turb apex indicated that
269
the loose biofilm was almost completely detached. If air + water backwash was
270
continuously performed, the tight formation of biofilm would be removed, which would
271
lead to the recovery of nitrate removal efficiency and biofilm renewal in the next filtration
272
operation. Therefore, Turb apex points and Turb platform states appeared on turbidity
13
273
profiles can be used to accurately determine air+water backwash time and water backwash
274
time, respectively.
275
Variations of turbidity during backwash well fitted normal distribution (R2=0.96313,
276
Figure 3). Backwash process was very similar to chromatographic separation, in which, air
277
and water, the mobile phase, carried the detached biofilm out of the filter media. Air mainly
278
enhanced filter media scour, while water flushed and carried the detached biofilm out of
279
biofilter. Hydrodynamic shear force was an effective tool to control of biofilm structures
280
(Morgenroth and Wilderer, 2000). However, in an engineering sense, as there is no
281
effective and simple method to measure the biofilm growth, the biomass removal, and the
282
strength of hydrodynamic shear force, some biofilters cannot be kept under stable condition
283
(Snowball, 2006). Furthermore, some filters had to remove filter media due to the extensive
284
clog (Figure S1). The results clearly showed that turb peak height, turbidity increase rate
285
and decrease rate indicated the strength of hydraulic shear force caused by air+water
286
velocity, air velocity and water velocity, respectively. Furthermore, turb peak area indicated
287
the removed biofilm amount. Therefore, biofilm removal during backwash process can be
288
well optimized using turbidity as control parameter.
289
3.3 Real-time control of the biofilm growth and renewal in DNBF using the effluent
290
turbidity
291
Preliminary studies found that the effluent turbidity could effectively indicate the biofilm
292
growth, renewal and removal in the DBNF, which could be used to real-time control
293
filtration and backwash for DNBF. Therefore, in Phase 3, the DNBF was operated for a
294
long period of time using the effluent turbidity to control backwash. In this test, two
14
295
widely-used carbon sources, sodium acetate and methanol, were used in DNBF,
296
respectively.
297
3.3.1 Filter performance
298
Figure 4 and Figure 5 showed the variations in the effluent nitrate and turbidity during
299
filtrations and backwashes with real-time control using sodium acetate and methanol as
300
carbon sources, respectively. During the filtrations, turbidity gradually increased and nitrate
301
gradually decreased, which indicated that biofilm gradually grew and filter layer were in
302
the stable stage. After Turb break-1 point appeared on turbidity profiles for 1.5-2.5 h
303
(extend time), the filtration was stopped. As backwash cycles were well controlled between
304
Turb break-1 point and Turb break-2 point, filter layer were well controlled at the
305
intergraded stages with Yc lower than 0.27. Turbidity and Yc during filtration in DNBF
306
using sodium acetate as carbon source were little higher than that using methanol. During
307
backwash, air+water backwash time and water backwash time was controlled at Turb apex
308
points and Turb platform points or turbidity lower than 2.5 NTU points. Each backwash
309
did not affect nitrate removal efficiency in the following filtration, which indicated
310
backwashes were well controlled without excessive backwash.
311
3.3.2 Biofilm structure and diversity of denitrifying bacteria
312
Biofilm structure plays an important role in both overall performance of filter. Figure S9
313
showed SEM images of the biofilm structure when sodium acetate (SA-biofilm) and
314
methanol (M-biofilm) were used as carbon sources. SA-biofilm and M-biofilm were both
315
composed by organisms and the biofilm matrix. The compact SA-biofilm with massive
15
316
distinct porous and channels completely covered the surface of filter media, indicating that
317
SA-biofilm was not only continuous and uneven, but also complicate and thick. Some
318
microcolonies with loose structure attached on the biofilm matrix. However, M-biofilm was
319
relatively simple and has clear structure. The distribution of M-biofilm was not only
320
discontinuous and uneven, but also thinner than SA-biofilm. Clearly, the bottom layer of
321
M-biofilm was single and regular shape bacillus, which was coated and covered by EPS.
322
SA-biofilm contained a relatively higher EPS than M-biofilm, further indicating SA-
323
biofilm was thicker than M-biofilm. The cell cluster with loose structure was also attached
324
to the biofilm matrix. The phylogenetic analysis of nirK gene sequences showed that nirK-
325
encoding denitrifiers in M-biofilm and SA-biofilm had high biodiversity (Figure 6);
326
whereas, nirK-encoding denitrifiers in M-biofilm were much more different from that in
327
SA-biofilm. Mesorhizobium and Bradyrhizobium were dominant denitrifier species in SA-
328
biofilm, while, Hyphomicrobium and Paracoccus were key denitrifying populations in M-
329
biofilm.
330
The differences of biofilm structure and diversity of nirK-encoding denitrifiers between
331
M-biofilm and SA-biofilm directly resulted in different variations of turbidity profiles using
332
sodium acetate and methanol as carbon sources. Although Turb break-1 points were both
333
appeared on turbidity profiles whether using sodium acetate or methanol as carbon sources
334
during filtration, because thick SA-biofilm has relatively more and larger microcolonies
335
with loose structure than M-biofilm, at the final stage of filtration, turbidity in the effluent
336
of DNBF when using sodium acetate as carbon source was fluctuated, and a higher value
337
than that using methanol as carbon source was observed. During the following backwash
16
338
process, turb-apex points also appeared on turbidity profiles using both sodium acetate and
339
methanol as carbon sources. Furthermore, because difference biomass growth rate caused
340
by different denitrifying communities using sodium acetate and methanol as carbon sources
341
(Gómez et al., 2000), turbidity at turb-apex point was relatively higher when sodium acetate
342
was used. Thus, because real-time control strategy via turbidity variation well indicated
343
biofilm growth and removal degree, no matter what carbon source was used, characteristic
344
points were appeared on turbidity profiles during filtration and backwash, which can be
345
used to control the filter layer and prevent excessive backwash.
346
In biofilm technologies for treating wastewater, biofilm is too thick or matured, which
347
would cause bulk biofilm slough out of carrier in the moving bed process (Huang et al.,
348
2014) or biofilter clog in the fixed bed process. However, biofilm is too thin or immature,
349
which would cause the decrease of treatment efficiency. Thus, it is difficult to control of
350
biofilm growth in biofilm system. This study suggests that the proposed real-time control
351
method can not only indicate biofilm formation and growth degree, but also well control
352
and optimize biofilm removal and renewal. More importantly, this real-time control method
353
via on-line turbidity monitoring might also be used in other fields, such as, the moving-bed
354
process, rotating biological contactor, granular system and water pipe system, which should
355
be investigated further.
356 357 358
4. Conclusions
The main conclusions obtained in the pilot-scale DNBF (600m3/d):
17
359
Biofilm formation, growth and removal can be well detected and controlled using on-
360
line monitored turbidity during biofilm cultivation, filtration and backwash processes
361
in DNBF.
362
The status of filter layer condition can be divided into stable, intergraded, clogged and
363
damaged phases based on Turb break points on turbidity profile. Yc can be used to
364
determine filter layer condition in DNBF.
365
Although biofilm structure and nirK-coding denitrifying communities using different
366
carbon sources were much more different, DNBF was still successfully and stably
367
optimized and real-time controlled via on-line turbidity.
368
Acknowledgements
369
This research was supported by National Science and Technology Major Projects on
370
Water pollution Control and Treatment (2011ZX07316-001 and 2013ZX07314-001) and
371
National Natural Science Foundation of China (51508561). The authors are grateful to
372
Jianmin Wang and Guoqiang Liu from Department of Civil, Architectural and
373
Environmental Engineering, Missouri University of Science and Technology, USA, for
374
providing language help and giving helpful advices.
375 376
Supplementary materials
377
Filter media pictures of the serious clogged DNBF in wastewater reclaimed plant (figure
378
S1), parameters, materials and schematic diagram of the pilot-scale DNBF (Table S1 and
18
379
Figure S2), correlation between biofilm formation and effluent turbidity during start-up of
380
lab-scale DNBF(Figure S3-4), backwash operation during the entire experiments (Table
381
S2), effects of low and higher backwash strength on water quality and biofilter
382
operation(Figure S5-8), and biofilm structure in DNBF using sodium acetate and methanol
383
as carbon sources(Figure S9)
384
References:
385
1. Amburgey, J.E., Amirtharajah, A., 2005. Strategic filter backwashing techniques and
386
resulting particle passage. Journal of environmental engineering-asce, 131, 535-547.
387
2. APHA, 1998. Standard Methods for Examination of Water and Wastewater. American
388
Public Health Association, Washington, DC.
389
3. Busscher, H.J., Norde, W., Sharma, P.K., van der Mei, H.C., 2010. Interfacial re-
390
arrangement in initial microbial adhesion to surfaces. Current Opinion in Colloid &
391
Interface Science, 15, 510-517.
392
4.
Corona, F., Mulas, M., Haimi, H., Sundell, L., Heinonen, M., Vahala, R., 2013.
393
Monitoring nitrate concentrations in the denitrifying post-filtration unit of a municipal
394
wastewater treatment plant. Journal of Process Control, 23, 158-170.
395
5. Gao, J., Luo, X., Wu, G., Li, T., Peng, Y., 2014. Abundance and diversity based on
396
amoA genes of ammonia-oxidizing archaea and bacteria in ten wastewater treatment
397
systems. Applied Microbiology and Biotechnology, 98, 3339-3354.
398
6. Giaouris, E., Heir, E., Hébraud, M., Chorianopoulos, N., Langsrud, S., Møretrø, T.,
399
Habimana, O., Desvaux, M., Renier, S., Nychas, G., 2014. Attachment and biofilm
400
formation by foodborne bacteria in meat processing environments: Causes, implications,
19
401
role of bacterial interactions and control by alternative novel methods. Meat Science
402
Advancing Beef Safety through Research and Innovation: Prosafebeef, 97, 298-309.
403
7. Gómez, M.A., González-López, J., Hontoria-Garcı́ a, E., 2000. Influence of carbon
404
source on nitrate removal of contaminated groundwater in a denitrifying submerged
405
filter. Journal of Hazardous Materials, 80, 69-80.
406 407
8. Hallin, S., Lindgren, P.E., 1999. PCR detection of genes encoding nitrile reductase in denitrifying bacteria. Applied and Environmental Microbiology, 65, 1652-1657.
408
9. Huang, H., Ren, H., Ding, L., Geng, J., Xu, K., Zhang, Y., 2014. Aging biofilm from a
409
full-scale moving bed biofilm reactor:Characterization and enzymatic treatment study.
410
Bioresource Technology, 154, 122-130.
411 412 413 414 415 416
10. Ji, B., Wang, H., Yang, K., 2014. Nitrate and COD removal in an upflow biofilter under an aerobic atmosphere. Bioresource Technology, 158, 156-160. 12. Leverenz, H.L., Tchobanoglous, G., Darby, J.L., 2009. Clogging in intermittently dosed sand filters used for wastewater treatment. Water Research, 43, 695-705. 13. Liu, H., Fang, H.H.P., 2002. Extraction of extracellular polymeric substances (EPS) of sludges. Journal of Biotechnology, 95, 249-256.
417
14. Mathieu, L., Bertrand, I., Abe, Y., Angel, E., Block, J.C., Skali-Lami, S., Francius, G.,
418
2014. Drinking water biofilm cohesiveness changes under chlorination or
419
hydrodynamic stress. Water Research, 55, 175-184.
420
15. Métadier, M., Bertrand-Krajewski, J.L., 2012. The use of long-term on-line turbidity
421
measurements for the calculation of urban stormwater pollutant concentrations, loads,
422
pollutographs and intra-event fluxes. Water Research, 46, 6836-6856.
20
423
16. Miura, Y., Watanabe, Y., Okabe, S., 2006. Membrane Biofouling in Pilot-Scale
424
Membrane Bioreactors (MBRs) Treating Municipal Wastewater: Impact of Biofilm
425
Formation. Environmental Science & Technology, 41, 632-638.
426 427
17. Morgenroth, E., Wilderer, P.A., 2000. Influence of detachment mechanisms on competition in biofilms. Water Research, 34, 417-426.
428
18. Pang, C.M., Hong, P., Guo, H., Liu, W., 2005. Biofilm Formation Characteristics of
429
Bacterial Isolates Retrieved from a Reverse Osmosis Membrane. Environmental
430
Science & Technology, 39, 7541-7550.
431
19. Pires, L., Sachsenheimer, K., Kleintschek, T., Waldbaur, A., Schwartz, T., Rapp, B.E.,
432
2013. Online monitoring of biofilm growth and activity using a combined multi-
433
channel impedimetric and amperometric sensor. Biosensors & Bioelectronics, 47, 157-
434
163.
435 436
20. Reyes-Romero, D.F., Behrmann, O., Dame, G., Urban, G.A., 2014. Dynamic thermal sensor for biofilm monitoring. Sensors and Actuators A: Physical, 213, 43-51.
437
21. Saber, N., Ju, Y., Hsu, H., Lee, S., 2013. A feasibility study on the application of
438
microwaves for online biofilm monitoring in the pipelines. International Journal of
439
Pressure Vessels and Piping, 111, 99-105.
440
22. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B.,
441
Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B.,
442
Thallinger, G.G., Van Horn, D.J., Weber, C.F., 2009. Introducing mothur: Open-Source,
443
Platform-Independent, Community-Supported Software for Describing and Comparing
444
Microbial Communities. Applied and Environmental Microbiology, 75, 7537-7541.
21
445 446
23. Simões, M., Simões, L.C., Vieira, M.J., 2010. A review of current and emergent biofilm control strategies. LWT - Food Science and Technology, 43, 573-583.
447
24. Slavik, I., Jehmlich, A., Uhl, W., 2013. Impact of backwashing procedures on deep bed
448
filtration productivity in drinking water treatment. Water Research, 47, 6348-6357.
449
25. Snowball, M., 2006. Water treatment: Reducing backwash with air scouring. Filtration
450
& Separation, 43, 39-40.
451
26. Tamura, K., Stecher, G., Peterson, D., Filipski, A., Kumar, S., 2013. MEGA6:
452
Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and
453
Evolution, 30, 2725-2729.
454
27. Wang, H., Ji, G., Bai, X., 2015. Enhanced long-term ammonium removal and its ranked
455
contribution of microbial genes associated with nitrogen cycling in a lab-scale
456
multimedia biofilter. Bioresource Technology, 196, 57-64.
457
28. Wang, Y., Guo, G., Wang, H., Stephenson, T., Guo, J., Ye, L., 2013. Long-term impact
458
of anaerobic reaction time on the performance and granular characteristics of granular
459
denitrifying biological phosphorus removal systems. Water Research, 47, 5326-5337.
460
29. Yang, G., Feng, L., Wang, S., Zhou, J., Guo, C., Xia, T., Sun, W., Jiang, Y., Sun, X.,
461
Cao, L., Xu, X., Zhu, L., 2015. Potential risk and control strategy of biofilm
462
pretreatment process treating raw water. Bioresource Technology, 198, 456-463.
463
22
Filtration
COD/(mg/L)
100
DNBFin
DNBFcarbon
DNBFout
(A)
75 50 25 0
0
1
2
3
4
5
6
(B)
20 10 3 0
(C)
Turbidity/(NTU)
Pressure
10 8
2
6
1
4 2
0 0
1
2
3
4
5
6
Pressure/(kpa)
NO3--N/(mg/L)
30
0
Time/(days)
Figure 1. Variation of COD, NO3--N, turbidity and pressure concentrations during the biofilm cultivation stage of DNBF
Figure 2. Typical variations of the effluent turbidity, CODadd, influent nitrate, and effluent nitrate during long-term filtration (a) and relationships between Xbiof and Cfilter (b) in DNBF
Backwash
Filtration Filtration Air Air+Water
Water
900 800 700
500
Y=12.58 + 897.5*exp(-2*((X-2047)/6.3)^2)
Turb Apex
600
9
500 400
6
300 200
400
3
2036 2038 2040 2042 2044 2046 2048 2050 2052 2054 2056 2058 2060 2062 2064
0
30 25 20
Turb Platform Nitrate Break
100
300 200
15
0
10
Time(min)
100
Turbidity Break
0 0
35
6
12
18 Time(h)
-
600
12
Turb Fit Curve
-
Turb(NTU)
700 Turb(NTU)
Adj.R-Square=0.96313
15
40
24
30
Figure 3. Typical variations of turbidity during filtration and backwash in DNBF
NO3 -N(mg/L)
800
-
NO3 -N
NO3 -N(mg/L)
900
5 0 36
NO3--NOUT
NO3--NIN
1000
Yc=0.26
Yc=0.18
Yc=0.25
Turbidity
35 Yc=0.21
Yc=0.17
Yc=0.20
30 800 25 20 20 15 10
Turb Break-1
Turb Break-1 Turb Break-1
Turb Break-1
Turb Break-1
Turb Break-1
NO3--N(mg/L)
Turbidity(NTU)
600
15 10
5
5
0
0 0
1
2
3
4 5 6 Time (Days)
7
8
9
Figure 4. Variations in the effluent nitrate and turbidity during filtrations and backwashes with real-time control using sodium acetate as carbon sources
-
Yc=0.21
Yc=0.13
35
Yc=0.11 Yc=0.12
Yc=0.11 Yc=0.13
800
30
600
25
400
20 15
15
-
Turbidity(NTU)
1000
Turbidity
NO3 -NOUT
NO3 -N(mg/L)
-
NO3 -NIN
Turb
Turb Break-1
10
Turb Break-1
Turb
Turb
Break-1
Break-1
Break-1
Turb Break-1
10
5
5
0
0 11
0
1
2
3
4
5
6
7
8
9
10
Time(days) Figure 5. Variations in the effluent nitrate and turbidity during filtrations and backwashes with real-time control using methanol as carbon source
OTU5 (9/27) OTU3 (1/29) OTU5 (9/27) Bradyrhizobium sp. D189(AB480442.1) Bradyrhizobium sp. D203a(AB480454.1) OTU3 (1/27) OTU16 (1/27) Bradyrhizobium japonicum USDA 110 DNA(BA000040.2) Bradyrhizobium sp. BTAi1(CP000494.1) OTU10 (1/27) Bosea sp. D257c(AB480483.1) Bosea sp.(HQ916684.1) 100 OTU11 (1/27) 100 Rhodanobacter sp. D206a(AB480456.1) Ochrobactrum sp. clone 1-40(GU136454.1) 100 Ochrobactrum sp. clone 2-80(GU136458.1) OTU15 (1/27) OTU12 (1/27) OTU1 (2/27) OTU9 (1/27) OTU6 (1/27) Mesorhizobium opportunistum WSM2075(CP002279.1) Mesorhizobium sp. GSM-373(FN600572.1) OTU9 (1/29) OTU2 (1/27) OTU14 (1/27) Chelativorans sp. BNC1(CP000390.1) OTU13 (1/27) 69 Paracoccus sp. R-26823(AM230847.1) 63 Paracoccus sp. R-28242(AM230886.1) 100 OTU5 (1/29) Paracoccus sp. R-26824(AM230857.1) OTU14 (1/29) OTU1 (1/29) OTU11 (1/29) OTU4 (10/29) 100 Rhodobacter sphaeroides forma sp.(AJ224908.1) Rhodobacter sphaeroides(U62291.1) OTU4 (3/27) Rhizobium gallicum(CP006880.1) OTU20 (1/27) Rhizobium sp. PY13(DQ096645.1) Rhizobium sp. R-24663(AM230832.1) Hyphomicrobium zavarzinii IFAM ZV-622(AJ224902.1) Hyphomicrobium nitrativorans NL23(CP006912.1) OTU14 (1/29) OTU2 (2/29) OTU7 (3/29) OTU13 (2/29) OTU8 (2/29) OTU6 (1/29) OTU10 (2/29)
100 47 46 49 67 68 49 27
90
42
13
81 13
74 25
4 15
100 46
19 29 23
24
75
100 39
92
16
100 52 60
7
100 69 67
100 62
75 67
96 55 47
Bradyrhizobium
Bosea Rhodanobacter Ochrobactrum
Mesorhizobium
Chelativorans
Paracoccus
Rhodobacter
Rhizobium
Hyphomicrobium
0.05
Figure. 6 NJ phylogenetic tree of the nirK -containing bacteria using methanol and sodium acetate as carbon sources. Sequences of nirK genes using methanol and sodium acetate as carbon sources are shown with "▲ OTU" and "♦OTU", respectively.
Table 1. Experimental procedure and aims Phases Duration Operation and main Parameters Aims (days) 1st 6/15* CS: sodium acetate Determine correlation between biofilm formation and turbidity during the FV:4.76m/h; FV:2.38m/h* start-up of the pilot-scale and lab-scale DNBF nd 2 9-120 CS: sodium acetate; FV: 4.76 m/h (40) Fixed time backwash cycle(48h) Effects of backwash on DNBF operation and establish backwash strength (38) Long-term filtration: Three times Determine the possibility of using turbidity as control parameter to indicate (Each time: filtration 5-6 days and biofilm growth and filter layer condition during long-term filtration. recovery 5days) Establish the methods of determining filter layer clogging degree. (33) Backwash cycle: Turb Break-1 point The correlation between the variations of the effluent turbidity and backwash (FV: 4.76-8.79m/h) rd 3 121-210 FV=8.79m/h Establish the real-time control approach of biofilm growth, removal and (30) CS: sodium acetate;; Real-time control renewal in DNBF via on-line monitoring turbidity. CS: methanol; Test the stability of the real-time control approach using two typical carbon (30) recovery stage sources. (30) Real-time control Analyze biofilm structure and diversity of denitrifying bacteria *in the lab-scale DNBF; CS—Carbon source; Real-time control—backwash was controlled via turbidity;
1
464 465
Optimizing and control of biofilm was studied in biofilter (600 m3/d).
466
Biofilm formation, growth and removal can be controlled using turbidity.
467
The filter layer status can be indicated by Turb break points on turbidity profile.
468
Filter layer clogging coefficient can be used to determine filter layer condition.
469 470
23
S0960-8524(16)30232-2 http://dx.doi.org/10.1016/j.biortech.2016.02.095 BITE 16145
To appear in:
Bioresource Technology
Received Date: Revised Date: Accepted Date:
20 November 2015 19 February 2016 22 February 2016
Please cite this article as: Liu, X., Wang, H., Long, F., Qi, L., Fan, H., Optimizing and Real-time Control of Biofilm Formation, Growth and Renewal in Denitrifying Biofilter, Bioresource Technology (2016), doi: http://dx.doi.org/ 10.1016/j.biortech.2016.02.095
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1
2
Optimizing and Real-time Control of Biofilm
3
Formation, Growth and Renewal in Denitrifying
4
Biofilter
5 6
Xiuhong Liu, Hongchen Wang *, Feng Long, Lu Qi, Haitao Fan
7
8 9
School of Environment & Natural Resources, Renmin University of China, Beijing, China,100872;
10
11
Corresponding author: Hongchen Wang
12
Phone: + 86-10-62510853
13
E-mail: [email protected]
14 15
1
16
ABSTRACT
17
A pilot-scale denitrifying biofilter (DNBF) with a treatment capacity of 600 m3/d was used
18
to study real-time control of biofilm formation, removal and renewal. The results showed
19
biofilm formation, growth and removal can be well controlled using on-line monitored
20
turbidity. The status of filter layer condition can be well indicated by Turb break points on
21
turbidity profile. There was a very good linear relationship between biofilm growth degree
22
(Xbiof) and filter clogging degree (Cfilter) with R2 higher than 0.99. Filter layer clogging
23
coefficient (Yc) lower than 0.27 can be used to determine stable filter layer condition. Since
24
variations of turbidity during backwash well fitted normal distribution with R2 higher than
25
0.96, biofilm removal during backwash also can be well optimized by turbidity. Although
26
biofilm structure and nirK-coding denitrifying communities using different carbon sources
27
were much more different, DNBF was still successfully and stably optimized and real-time
28
controlled via on-line turbidity.
29
30 31
KEY WORDS
32
Biofilm; formation; turbidity; real-time control; denitrifying biofilter
33
2
34
1. Introduction
35
Wastewater reclamation is one of the most effective ways to solve both water shortage and
36
pollution problems. Many wastewater treatment plants in water deficient regions, especially
37
large and medium-sized cities, have been up-graded to produce reclaimed water, and
38
advanced nitrate removal is required. Because denitrifying biofilter (DNBF) has advantages
39
of high treatment efficiency, small footprint, and low shock loading impact, it is widely
40
used for advanced nitrate removal. However, because biofilm growth cannot be easily
41
monitored, many biofilters do not provide stable performance. Moreover, some biofilters
42
can not be normally operated due to frequent clogging (Leverenz et al., 2009; Snowball,
43
2006) (Figure S1), which limits the engineering application and development of biofilter.
44
In fixed-film wastewater treatment systems (Corona et al., 2013; Ji et al., 2014; Wang et
45
al., 2015), biofilm is used to remove pollutants in wastewater; whereas, in other fields
46
including membrane treatment processes (Miura et al., 2006; Pang et al., 2005), drinking
47
water or reclaimed water distribution systems (Mathieu et al., 2014; Yang et al., 2015), and
48
food processing industry (Simões et al., 2010), biofilm formation is harmful and should be
49
avoided. It was reported that biofilm formation comprised a series of complicated steps
50
(Giaouris et al., 2014; Simões et al., 2010) including cell deposition, cell adsorption and/or
51
desorption, cell–cell signaling and EPS production, replication and growth, secretion of
52
biofilm matrix, and biofilm detachment or sloughing. To date, there is no effective method
53
to accurately measure or determine biofilm growth condition in a biofilter, especially in a
54
DNBF. Generally, during the start-up stage of a DNBF, biofilm was naturally cultivated
55
with nitrate and organic carbon. However, since biofilm can not be practically monitored or
56
measured during treatment, it is hard to determine the biofilter run time between
3
57
backwashes (backwash cycle), the backwash strength and duration. This may result in over
58
backwash and energy cost, or biofilter clogging if backwash is not performed in time. In
59
the last decade, most studies on biofilter process were focused on optimizing backwash
60
procedures (Amburgey and Amirtharajah, 2005; Slavik et al., 2013; Snowball, 2006), while
61
limited studies have been conducted on optimizing the biofilter operation based on the
62
extent of biofilm growth and removal.
63
Many on-line biofilm monitoring methods, such as differential turbidity measurement
64
(Métadier and Bertrand-Krajewski, 2012), microwaves (Saber et al., 2013), multi-channel
65
impedimetric and amperometric sensor (Pires et al., 2013) and thermal sensors (Reyes-
66
Romero et al., 2014), have been used to monitor biofilm growth. Among these methods,
67
turbidity is suitable for industrial applications. It was used to calculate urban storm water
68
pollutant concentrations and loads (Métadier and Bertrand-Krajewski, 2012). Backwash is
69
the key operational step to control biofilm growth, removal and renewal in treatment plants
70
that employ biofilter/filter as major unit process. However, the effective method to monitor
71
biofilm growth on-line in biofilter has not been well developed, and very limited studies
72
have been conducted on real-time control and optimization of biofilter-based biofilm
73
formation and growth.
74
Therefore, this study aimed to: a) determine the relationship between biofilm formation
75
and turbidity variation during the biofilm cultivation stage of a DNBF; b) control and
76
optimize biofilm growth, removal and renewal during filtration and backwash processes in
77
the DNBF via on-line monitored turbidity; c) test the stability of real-time control of
78
biofilm growth, removal and renewal in the DNBF using different typical carbon sources.
4
79
2. Materials and methods
80
2.1 Pilot-scale and lab-scale DNBF
81
A pilot-scale DNBF located at Beijing Gaobeidian Municipal Wastewater Treatment Plant
82
(WWTP) with a maximum treatment capacity of 600 m3/d was used in this study. The
83
DNBF was packed with expanded clay particles (4-6 mm) at a bed depth of 2.5 m (Table
84
S1). The pilot-scale treatment system consisted of a DNBF reactor, a carbon dosage system,
85
and a backwash system (Figure S2).
86
The control consists of four parts: detectors, a computer, interface cards and control units.
87
Nitrate (NO3-), turbidity, pH, dissolved oxygen (DO) and pressure sensors were installed in
88
the top section and the bottom of DNBF reactor, respectively. The values of the NO3-,
89
turbidity, pH, and DO were recorded every 0.5–10 minute and then transferred to a
90
programmable logic controller (PLC) and process control system (PCS). PCS was
91
programmed according to the control logic.
92
A lab-scale DNBF with a total volume of 30 L and filter media height of 1.1m was used
93
to study biofilm formation further. Except for volume and media height, the lab-scale
94
DNBF was the same as the pilot-scale DNBF.
95
2.2 Secondary effluent composition
96
The secondary effluent of Gaobeidian municipal WWTP that employs an anoxic/oxic (A/O)
97
process was continuously fed to the DNBF. The concentrations of the soluble chemical
98
oxygen demand (SCOD), NH4+-N, NO3--N, turbidity, and SS in the secondary effluent were
99
in the range of 23.28-76.03 mg/L, 0.33-2.36 mg/L, 20.40-34.12 mg/L, 0.48-13.10 NTU,
100
and 1.33-16.15 mg/L, respectively.
101
5
102
2.3 Experimental design
103
Experiments were conducted for 210 days, which was divided into 3 phases. Experimental
104
procedures, parameters, aims, and backwash operation were shown in Table 1 and tableS2.
105
In the first phase, the natural biofilm cultivation method was used in the pilot-scale and lab-
106
scale DNBF. Since biofilm formation in the pilot-scale DNBF was only operated for one
107
time and the filter media is difficultly taken out to observation, the lab-scale DNBF was
108
operated with Filtration velocity (FV) of 2.38m/h only to further confirm the results
109
obtained in the pilot-scale DNBF. In the second phase, DNBF was firstly operated for 40
110
days to establish backwash strength. Thereafter, variations of the effluent turbidity during
111
long-term filtration and backwash operation were studied on the following 38 days and 33
112
days, respectively. In the third phase, the stability of long-term operation of DNBF with
113
real-time control using turbidity as parameter was tested using two typical carbon sources.
114
When carbon source changed from sodium acetate to methanol, nitrate removal efficiency
115
of DNBF was recovered for 30 days.
116 117
2.4 Analysis
118
COD, NH4+-N, NO3--N, NO2--N, TN, PO43--P, and total phosphorus (TP) were measured
119
according to the standard methods given in APHA (APHA, 1998). DO, turbidity, NO3--N
120
and pressure were measured on-line using oxygen, turbidity, NO3--N meters (LDO,
121
SOLITAX sc, NITRATAX plus sc, Hach Company, USA) and pressure sensor,
122
respectively.
123
Extracellular Polymeric Substances (EPS) was extracted using formaldehyde-NaOH
124
according to the method described by Liu and Fang (Liu and Fang, 2002). Pre-treatment of
6
125
filter media for SEM observation was carried out using the modified method described by
126
Wang et al. (2013) (Wang et al., 2013). Visualization of the samples was conducted using a
127
Scanning Electron Microscope (Hoskin Scientific, Tokyo, Japan).
128
2.5 DNA extraction, PCR, cloning, sequencing and phylogenetic analysis
129
Filter media samples were taken from DNBF on day 100 and 203. Biofilm was removed
130
from filter media by shaker. The collected samples were freeze-dried in Labconco Freezone
131
1 L (Labconco, USA) and stored at −20 °C.
132
0.05–0.10 g of dry sludge sample was taken to extract genomic DNA using a FastDNA
133
SPIN Kit for soil (Qiagen, CA, USA). 1ul of extracted DNA was taken to measure DNA
134
concentration by Nanodrop Spectrophotometer (ND-1000, Thermo Fisher Scientific, USA).
135
Fragments of nirK genes were amplified with the primer set FlaCu/R3Cu (473 bp) as
136
described by Hallin et al. (1999) (Hallin and Lindgren, 1999).
137
The PCR was performed in a C1000TM thermal cycler (BioRad, USA). Cycle conditions
138
for the amplification were as follows: 2 min at 94°C; 35 cycles with each cycle consisting
139
of 30 sec at 94°C, 1 min at 57°C, and 1 min at 72°C; followed by a final 10-min extension
140
at 72°C. PCR products were visualized on 1.5% (w/v) agarose gel electrophoresis to
141
confirm the product size, and then, purified by Wizard® SV Gel and PCR Clean-Up
142
System (Promega, Madison, USA). The detailed information on cloning and sequencing
143
were described by Gao et al. (2014) (Gao et al., 2014).
144
All the sequences and their reference sequences obtained from NCBI BLAST were
145
aligned using MEGA 6.0 software (Tamura et al., 2013). The sequences sharing 98%
146
similarity were grouped into the same operational taxonomic unit (OTU) using software
7
147
mother (Schloss et al., 2009). Phylogenetic trees were generated by neighbor-joining (NJ)
148
with the Jukes–Cantor correction in MEGA (Tamura et al., 2013).
149
2.6 Calculations
150
Biofilm growth degree (Xbiof), filter clogging degree (Cfilter) and filter layer clogging
151
coefficient (Yc), were calculated based on the biomass yield coefficient and total effluent
152
turbidity as described in equation(1) – (3) , respectively. NT
{
Y NO3 − N ∑ [NO3 − N ]in - [NO3 − N ]out 153
}
n ∆t
× ∆t × Q n∆t
n =1
X biofilm =
(1)
Vfilter
154
where, YNO3-N is sludge yield coefficient (0.74 gVSS/gNO3-N), [NO3-N]in and [NO3-N]out
155
are nitrate concentrations in the influent and effluent (g/m3), n is data points on the profiles
156
of on-line monitored nitrate, T is the time of on-line monitored nitrate, NT is the data points
157
at the time of T, ∆t is the on-line monitored interval (day), Q is the influent flow rate (m3/d),
158
Vfilter is the volume of filter media (m3). NT
∑ {[Turb] - [Turb ] } in
159
160
C filter =
out n ∆t
× ∆t × Q n∆t
(2)
n =1
Vfilter
where, [Turb]in and [Turb]out are turbidity concentrations in the influent and effluent(NTU). NT
161
Yc =
∑ {[Turb] - [Turb] } in
1 YNO3 − N
out n∆t
n =1 NT
∑ {[NO
3
− N ]in - [NO3 − N ]out
n =1
8
}
n∆t
(3)
162
where, Yc is the slope of Cfilter/Xbiof.
163 164
3. Results and discussion
165
3.1 Correlation between biofilm formation and effluent turbidity during start-up of
166
DNBF
167
Figure 1 showed variations of COD, NO3--N, turbidity and pressure during biofilm
168
cultivation in pilot-scale DNBF. Because almost no denitrifying bacteria was cultivated in
169
the first 2 days, very small amount of nitrate was removed and the added carbon source
170
(CODadd=CODcarbon-CODin) was almost not consumed. From the 3rd day, nitrate removal
171
efficiency increased to 55%, and CODadd was still not sufficiently utilized. Meanwhile,
172
gases emitted from the effluent, indicating that denitrification occurred. On the following 2
173
days, CODadd was sufficiently utilized with removal efficiency higher than 98%. Figure S3
174
showed variations of CODremoved, NO3--Nremoved, turbidity and pressure during biofilm
175
cultivation in the lab-scale DNBF. In the first two days, variations of the removed COD,
176
nitrate and turbidity in lab-scale DNBF were very similar to these in the pilot-scale DNBF.
177
From day 3 to 4, with COD and nitrate slightly removed, turbidity gradually increased.
178
SEM results also found cells absorbed and grew on the rough surface of filter media
179
(Figure S4). Some of cells aggregated together and formed micro colonies. These results
180
indicated that with the cells replication and growth, since some discohesive cells and
181
produced EPS were desorbed from filter media into water (Busscher et al., 2010) under the
182
shear forces of the moving water with the filter media surface, turbidity were gradually
183
increased. On the flowing four days, with the increase of COD and nitrate removal,
184
turbidity was still gradually increased. On day 8, SEM results also demonstrated the
9
185
relatively integrated and stable biofilm were formed. Meanwhile, because the loose
186
structure biofilm were detached from filter media due to the hydraulic shear force, small
187
amount of macroscopic biofilm fragments were found in the effluent. After that, since the
188
excessive growth of biofilm caused the filter layer blocked, with the matured and aged
189
biofilm slough off and new biofilm formation, the hydraulic sheer force fluctuated caused
190
turbidity fluctuation. Therefore, based on the above results in pilot-scale and lab-scale
191
DNBF, turbidity in the effluent can well indicated biofilm formation and growth. To date,
192
few studies used turbidity to monitor biofilm formation and growth in a wastewater
193
treatment system.
194
After the biofilm density gradually increased during the cultivation stage, backwash was
195
performed to scour the excessive biomass under higher hydrodynamic shear force, to renew
196
the biofilm with new denitrifying bacteria (Table S2). In Phase 2, on the first 40 days, if
197
backwash strength was controlled too low (Figure S5), during filtration, not only the
198
effluent nitrate was unstable even after adding sufficient carbon source, but also turbidity
199
was fluctuated significantly. Some large SS aggregates were also observed in the effluent
200
(Figure S6). Furthermore, because of more frequent filter clogging, backwash operation
201
became more difficult. However, high backwash intensity might also result in filter media
202
loss (Figure S7), and mix the supporting layer with filter media layer (Figure S8) in the
203
reclaimed WWTP. Therefore, appropriate backwash is the key to ensure long-term
204
operation of the biofilter.
205
Since biofilter is a closed system, the biofilm on the biofilter media can not be measured
206
directly. In addition, biofilter clogging has a time-lag property, which even increases the
10
207
difficulty to operate and maintain a biofilter. These undesirable backwash control methods
208
mainly resulted from not timely monitoring biofilm growth and filter layer clog condition.
209
In wastewater/water treatment plants, biofilters are normally operated based on operator's
210
experience, and these undesirable backwash control methods often occur, which slow the
211
application and development of biofilter. To date, there is no effective way to on-line
212
monitoring biofilm growth degree and determine filter media clog degree in fundamental
213
and applied research.
214
3.2 Correlation between effluent turbidity and biofilm renewal and removal during
215
normal operation
216
Based on the obtained results during biofilm formation, turbidity might also be used to
217
indicate biofilm growth and removal during filtration and backwash operation. To validate
218
this hypothesis, turbidity and nitrate concentrations in the effluent were monitored online
219
during filtration and backwash processes.
220
3.2.1 Variation in effluent turbidity indicating biofilm growth and filter layer condition
221
during filtration
222
Figure 2a showed typical variations of the effluent turbidity, CODadd, influent nitrate, and
223
effluent nitrate during long-term filtration. With the biodegradation of COD and the
224
reduction of nitrate, biofilm grew gradually and became thicker. Meanwhile, filter layer
225
was gradually blocked. At the beginning of filtration, the effluent nitrate varied with
226
CODadd and the effluent turbidity increase slightly. However, after continuous filtration for
227
about 49 h, since some microcolonies and/or biofilm fragments were sloughed off from the
11
228
surface of biofilm, both the effluent turbidity and nitrate increased significantly. Turb
229
break-1 and Nitrate break-1 points both appeared on the turbidity and ef-nitrate profiles,
230
respectively. Thereafter, the effluent nitrate still varied with CODadd, while the effluent
231
turbidity turned increase to decrease because of biosorption and low nitrate removal amount;
232
whereas, in the 63.25 h, because of higher hydraulic shear force caused by more serious
233
filter layer blockage, the effluent turbidity increased sharply again, and Turb break-2 point
234
appeared on the effluent turbidity. Subsequently, more and more filter layers were blocked
235
due to biofilm growth, and as a result, both the effluent turbidity and nitrate fluctuated
236
significantly. At the same time, both Turb shock and nitrate shock points appeared on the
237
turbidity and effluent nitrate profiles, respectively. Therefore, based on the above results,
238
Turb break points on turbidity profile were suitable for indicating the status of filter layer
239
condition, which can be divided into four phases, including stable, intergraded, clogged and
240
damaged phases.
241
Figure 2b showed relationships between Xbiof and Cfilter. It was found that Cfilter increased
242
with the increase of Xbiof. In each phase, there was a very good linear relationship between
243
Xbiof and Cfilter with R2 higher than 0.99. Yc kept constant in each phase, while, Yc
244
increased with the clogging status of biofilter transferred from one phase to the next.
245
Especially, Yc increased sharply from 0.54 in the clogging phase and to 2.46 in the
246
damaged phase. These results indicate that Yc can be used to determine the status of filter
247
layer clogging degree.
248
Therefore, during normal operation, if backwash cycle was controlled on-line between
249
Turb break-1 point and Turb break-2 point, not only DNBF kept stable operation and
250
avoided filter layer clogging, but also carbon dosage can be well controlled by the on-line
12
251
monitored effluent nitrate. Even when turbidity can not be monitored on-line for a long
252
time, filter layer clogging degree still can be determined by calculating Yc through
253
monitoring the effluent nitrate and turbidity for the final 2-3 h of filtration.
254
3.2.2 Effluent turbidity indicating the biofilm removal degree during backwash
255
Figure 3 showed the variations in the effluent turbidity and nitrate concentration during
256
filtration and backwash processes. During the filtration, when the Turb break-1 point lasted
257
for 30 min, the biofilter transformed from the stable phase to the intergraded phase. Then
258
the filtration was stopped and backwash was performed. During the air backwash, turbidity
259
increased significantly. Hereafter, during air + water backwash, most biofilm was detached
260
and carried out by backwash water. When the biofilm almost was completely detached,
261
turbidity reached a peak and then decreased; turbidity apex points appeared on the turbidity
262
profile. Thereafter, during water backwash, backwash water carried the residual detached
263
biofilm out of filter layer. Meanwhile, turbidity decreased sharply, followed by a gradual
264
decrease. After that, turbidity kept constant, which indicated that the detached biofilm was
265
completely removed by backwash water. Among three phases of backwash, air + water
266
backwash was the most intensive and decisive phase to keep the equilibrium between the
267
excessive biofilm growth and biofilm removal, in which, more than half of the detached
268
biofilm was removed. Furthermore, during air + water backwash, Turb apex indicated that
269
the loose biofilm was almost completely detached. If air + water backwash was
270
continuously performed, the tight formation of biofilm would be removed, which would
271
lead to the recovery of nitrate removal efficiency and biofilm renewal in the next filtration
272
operation. Therefore, Turb apex points and Turb platform states appeared on turbidity
13
273
profiles can be used to accurately determine air+water backwash time and water backwash
274
time, respectively.
275
Variations of turbidity during backwash well fitted normal distribution (R2=0.96313,
276
Figure 3). Backwash process was very similar to chromatographic separation, in which, air
277
and water, the mobile phase, carried the detached biofilm out of the filter media. Air mainly
278
enhanced filter media scour, while water flushed and carried the detached biofilm out of
279
biofilter. Hydrodynamic shear force was an effective tool to control of biofilm structures
280
(Morgenroth and Wilderer, 2000). However, in an engineering sense, as there is no
281
effective and simple method to measure the biofilm growth, the biomass removal, and the
282
strength of hydrodynamic shear force, some biofilters cannot be kept under stable condition
283
(Snowball, 2006). Furthermore, some filters had to remove filter media due to the extensive
284
clog (Figure S1). The results clearly showed that turb peak height, turbidity increase rate
285
and decrease rate indicated the strength of hydraulic shear force caused by air+water
286
velocity, air velocity and water velocity, respectively. Furthermore, turb peak area indicated
287
the removed biofilm amount. Therefore, biofilm removal during backwash process can be
288
well optimized using turbidity as control parameter.
289
3.3 Real-time control of the biofilm growth and renewal in DNBF using the effluent
290
turbidity
291
Preliminary studies found that the effluent turbidity could effectively indicate the biofilm
292
growth, renewal and removal in the DBNF, which could be used to real-time control
293
filtration and backwash for DNBF. Therefore, in Phase 3, the DNBF was operated for a
294
long period of time using the effluent turbidity to control backwash. In this test, two
14
295
widely-used carbon sources, sodium acetate and methanol, were used in DNBF,
296
respectively.
297
3.3.1 Filter performance
298
Figure 4 and Figure 5 showed the variations in the effluent nitrate and turbidity during
299
filtrations and backwashes with real-time control using sodium acetate and methanol as
300
carbon sources, respectively. During the filtrations, turbidity gradually increased and nitrate
301
gradually decreased, which indicated that biofilm gradually grew and filter layer were in
302
the stable stage. After Turb break-1 point appeared on turbidity profiles for 1.5-2.5 h
303
(extend time), the filtration was stopped. As backwash cycles were well controlled between
304
Turb break-1 point and Turb break-2 point, filter layer were well controlled at the
305
intergraded stages with Yc lower than 0.27. Turbidity and Yc during filtration in DNBF
306
using sodium acetate as carbon source were little higher than that using methanol. During
307
backwash, air+water backwash time and water backwash time was controlled at Turb apex
308
points and Turb platform points or turbidity lower than 2.5 NTU points. Each backwash
309
did not affect nitrate removal efficiency in the following filtration, which indicated
310
backwashes were well controlled without excessive backwash.
311
3.3.2 Biofilm structure and diversity of denitrifying bacteria
312
Biofilm structure plays an important role in both overall performance of filter. Figure S9
313
showed SEM images of the biofilm structure when sodium acetate (SA-biofilm) and
314
methanol (M-biofilm) were used as carbon sources. SA-biofilm and M-biofilm were both
315
composed by organisms and the biofilm matrix. The compact SA-biofilm with massive
15
316
distinct porous and channels completely covered the surface of filter media, indicating that
317
SA-biofilm was not only continuous and uneven, but also complicate and thick. Some
318
microcolonies with loose structure attached on the biofilm matrix. However, M-biofilm was
319
relatively simple and has clear structure. The distribution of M-biofilm was not only
320
discontinuous and uneven, but also thinner than SA-biofilm. Clearly, the bottom layer of
321
M-biofilm was single and regular shape bacillus, which was coated and covered by EPS.
322
SA-biofilm contained a relatively higher EPS than M-biofilm, further indicating SA-
323
biofilm was thicker than M-biofilm. The cell cluster with loose structure was also attached
324
to the biofilm matrix. The phylogenetic analysis of nirK gene sequences showed that nirK-
325
encoding denitrifiers in M-biofilm and SA-biofilm had high biodiversity (Figure 6);
326
whereas, nirK-encoding denitrifiers in M-biofilm were much more different from that in
327
SA-biofilm. Mesorhizobium and Bradyrhizobium were dominant denitrifier species in SA-
328
biofilm, while, Hyphomicrobium and Paracoccus were key denitrifying populations in M-
329
biofilm.
330
The differences of biofilm structure and diversity of nirK-encoding denitrifiers between
331
M-biofilm and SA-biofilm directly resulted in different variations of turbidity profiles using
332
sodium acetate and methanol as carbon sources. Although Turb break-1 points were both
333
appeared on turbidity profiles whether using sodium acetate or methanol as carbon sources
334
during filtration, because thick SA-biofilm has relatively more and larger microcolonies
335
with loose structure than M-biofilm, at the final stage of filtration, turbidity in the effluent
336
of DNBF when using sodium acetate as carbon source was fluctuated, and a higher value
337
than that using methanol as carbon source was observed. During the following backwash
16
338
process, turb-apex points also appeared on turbidity profiles using both sodium acetate and
339
methanol as carbon sources. Furthermore, because difference biomass growth rate caused
340
by different denitrifying communities using sodium acetate and methanol as carbon sources
341
(Gómez et al., 2000), turbidity at turb-apex point was relatively higher when sodium acetate
342
was used. Thus, because real-time control strategy via turbidity variation well indicated
343
biofilm growth and removal degree, no matter what carbon source was used, characteristic
344
points were appeared on turbidity profiles during filtration and backwash, which can be
345
used to control the filter layer and prevent excessive backwash.
346
In biofilm technologies for treating wastewater, biofilm is too thick or matured, which
347
would cause bulk biofilm slough out of carrier in the moving bed process (Huang et al.,
348
2014) or biofilter clog in the fixed bed process. However, biofilm is too thin or immature,
349
which would cause the decrease of treatment efficiency. Thus, it is difficult to control of
350
biofilm growth in biofilm system. This study suggests that the proposed real-time control
351
method can not only indicate biofilm formation and growth degree, but also well control
352
and optimize biofilm removal and renewal. More importantly, this real-time control method
353
via on-line turbidity monitoring might also be used in other fields, such as, the moving-bed
354
process, rotating biological contactor, granular system and water pipe system, which should
355
be investigated further.
356 357 358
4. Conclusions
The main conclusions obtained in the pilot-scale DNBF (600m3/d):
17
359
Biofilm formation, growth and removal can be well detected and controlled using on-
360
line monitored turbidity during biofilm cultivation, filtration and backwash processes
361
in DNBF.
362
The status of filter layer condition can be divided into stable, intergraded, clogged and
363
damaged phases based on Turb break points on turbidity profile. Yc can be used to
364
determine filter layer condition in DNBF.
365
Although biofilm structure and nirK-coding denitrifying communities using different
366
carbon sources were much more different, DNBF was still successfully and stably
367
optimized and real-time controlled via on-line turbidity.
368
Acknowledgements
369
This research was supported by National Science and Technology Major Projects on
370
Water pollution Control and Treatment (2011ZX07316-001 and 2013ZX07314-001) and
371
National Natural Science Foundation of China (51508561). The authors are grateful to
372
Jianmin Wang and Guoqiang Liu from Department of Civil, Architectural and
373
Environmental Engineering, Missouri University of Science and Technology, USA, for
374
providing language help and giving helpful advices.
375 376
Supplementary materials
377
Filter media pictures of the serious clogged DNBF in wastewater reclaimed plant (figure
378
S1), parameters, materials and schematic diagram of the pilot-scale DNBF (Table S1 and
18
379
Figure S2), correlation between biofilm formation and effluent turbidity during start-up of
380
lab-scale DNBF(Figure S3-4), backwash operation during the entire experiments (Table
381
S2), effects of low and higher backwash strength on water quality and biofilter
382
operation(Figure S5-8), and biofilm structure in DNBF using sodium acetate and methanol
383
as carbon sources(Figure S9)
384
References:
385
1. Amburgey, J.E., Amirtharajah, A., 2005. Strategic filter backwashing techniques and
386
resulting particle passage. Journal of environmental engineering-asce, 131, 535-547.
387
2. APHA, 1998. Standard Methods for Examination of Water and Wastewater. American
388
Public Health Association, Washington, DC.
389
3. Busscher, H.J., Norde, W., Sharma, P.K., van der Mei, H.C., 2010. Interfacial re-
390
arrangement in initial microbial adhesion to surfaces. Current Opinion in Colloid &
391
Interface Science, 15, 510-517.
392
4.
Corona, F., Mulas, M., Haimi, H., Sundell, L., Heinonen, M., Vahala, R., 2013.
393
Monitoring nitrate concentrations in the denitrifying post-filtration unit of a municipal
394
wastewater treatment plant. Journal of Process Control, 23, 158-170.
395
5. Gao, J., Luo, X., Wu, G., Li, T., Peng, Y., 2014. Abundance and diversity based on
396
amoA genes of ammonia-oxidizing archaea and bacteria in ten wastewater treatment
397
systems. Applied Microbiology and Biotechnology, 98, 3339-3354.
398
6. Giaouris, E., Heir, E., Hébraud, M., Chorianopoulos, N., Langsrud, S., Møretrø, T.,
399
Habimana, O., Desvaux, M., Renier, S., Nychas, G., 2014. Attachment and biofilm
400
formation by foodborne bacteria in meat processing environments: Causes, implications,
19
401
role of bacterial interactions and control by alternative novel methods. Meat Science
402
Advancing Beef Safety through Research and Innovation: Prosafebeef, 97, 298-309.
403
7. Gómez, M.A., González-López, J., Hontoria-Garcı́ a, E., 2000. Influence of carbon
404
source on nitrate removal of contaminated groundwater in a denitrifying submerged
405
filter. Journal of Hazardous Materials, 80, 69-80.
406 407
8. Hallin, S., Lindgren, P.E., 1999. PCR detection of genes encoding nitrile reductase in denitrifying bacteria. Applied and Environmental Microbiology, 65, 1652-1657.
408
9. Huang, H., Ren, H., Ding, L., Geng, J., Xu, K., Zhang, Y., 2014. Aging biofilm from a
409
full-scale moving bed biofilm reactor:Characterization and enzymatic treatment study.
410
Bioresource Technology, 154, 122-130.
411 412 413 414 415 416
10. Ji, B., Wang, H., Yang, K., 2014. Nitrate and COD removal in an upflow biofilter under an aerobic atmosphere. Bioresource Technology, 158, 156-160. 12. Leverenz, H.L., Tchobanoglous, G., Darby, J.L., 2009. Clogging in intermittently dosed sand filters used for wastewater treatment. Water Research, 43, 695-705. 13. Liu, H., Fang, H.H.P., 2002. Extraction of extracellular polymeric substances (EPS) of sludges. Journal of Biotechnology, 95, 249-256.
417
14. Mathieu, L., Bertrand, I., Abe, Y., Angel, E., Block, J.C., Skali-Lami, S., Francius, G.,
418
2014. Drinking water biofilm cohesiveness changes under chlorination or
419
hydrodynamic stress. Water Research, 55, 175-184.
420
15. Métadier, M., Bertrand-Krajewski, J.L., 2012. The use of long-term on-line turbidity
421
measurements for the calculation of urban stormwater pollutant concentrations, loads,
422
pollutographs and intra-event fluxes. Water Research, 46, 6836-6856.
20
423
16. Miura, Y., Watanabe, Y., Okabe, S., 2006. Membrane Biofouling in Pilot-Scale
424
Membrane Bioreactors (MBRs) Treating Municipal Wastewater: Impact of Biofilm
425
Formation. Environmental Science & Technology, 41, 632-638.
426 427
17. Morgenroth, E., Wilderer, P.A., 2000. Influence of detachment mechanisms on competition in biofilms. Water Research, 34, 417-426.
428
18. Pang, C.M., Hong, P., Guo, H., Liu, W., 2005. Biofilm Formation Characteristics of
429
Bacterial Isolates Retrieved from a Reverse Osmosis Membrane. Environmental
430
Science & Technology, 39, 7541-7550.
431
19. Pires, L., Sachsenheimer, K., Kleintschek, T., Waldbaur, A., Schwartz, T., Rapp, B.E.,
432
2013. Online monitoring of biofilm growth and activity using a combined multi-
433
channel impedimetric and amperometric sensor. Biosensors & Bioelectronics, 47, 157-
434
163.
435 436
20. Reyes-Romero, D.F., Behrmann, O., Dame, G., Urban, G.A., 2014. Dynamic thermal sensor for biofilm monitoring. Sensors and Actuators A: Physical, 213, 43-51.
437
21. Saber, N., Ju, Y., Hsu, H., Lee, S., 2013. A feasibility study on the application of
438
microwaves for online biofilm monitoring in the pipelines. International Journal of
439
Pressure Vessels and Piping, 111, 99-105.
440
22. Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B.,
441
Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B.,
442
Thallinger, G.G., Van Horn, D.J., Weber, C.F., 2009. Introducing mothur: Open-Source,
443
Platform-Independent, Community-Supported Software for Describing and Comparing
444
Microbial Communities. Applied and Environmental Microbiology, 75, 7537-7541.
21
445 446
23. Simões, M., Simões, L.C., Vieira, M.J., 2010. A review of current and emergent biofilm control strategies. LWT - Food Science and Technology, 43, 573-583.
447
24. Slavik, I., Jehmlich, A., Uhl, W., 2013. Impact of backwashing procedures on deep bed
448
filtration productivity in drinking water treatment. Water Research, 47, 6348-6357.
449
25. Snowball, M., 2006. Water treatment: Reducing backwash with air scouring. Filtration
450
& Separation, 43, 39-40.
451
26. Tamura, K., Stecher, G., Peterson, D., Filipski, A., Kumar, S., 2013. MEGA6:
452
Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and
453
Evolution, 30, 2725-2729.
454
27. Wang, H., Ji, G., Bai, X., 2015. Enhanced long-term ammonium removal and its ranked
455
contribution of microbial genes associated with nitrogen cycling in a lab-scale
456
multimedia biofilter. Bioresource Technology, 196, 57-64.
457
28. Wang, Y., Guo, G., Wang, H., Stephenson, T., Guo, J., Ye, L., 2013. Long-term impact
458
of anaerobic reaction time on the performance and granular characteristics of granular
459
denitrifying biological phosphorus removal systems. Water Research, 47, 5326-5337.
460
29. Yang, G., Feng, L., Wang, S., Zhou, J., Guo, C., Xia, T., Sun, W., Jiang, Y., Sun, X.,
461
Cao, L., Xu, X., Zhu, L., 2015. Potential risk and control strategy of biofilm
462
pretreatment process treating raw water. Bioresource Technology, 198, 456-463.
463
22
Filtration
COD/(mg/L)
100
DNBFin
DNBFcarbon
DNBFout
(A)
75 50 25 0
0
1
2
3
4
5
6
(B)
20 10 3 0
(C)
Turbidity/(NTU)
Pressure
10 8
2
6
1
4 2
0 0
1
2
3
4
5
6
Pressure/(kpa)
NO3--N/(mg/L)
30
0
Time/(days)
Figure 1. Variation of COD, NO3--N, turbidity and pressure concentrations during the biofilm cultivation stage of DNBF
Figure 2. Typical variations of the effluent turbidity, CODadd, influent nitrate, and effluent nitrate during long-term filtration (a) and relationships between Xbiof and Cfilter (b) in DNBF
Backwash
Filtration Filtration Air Air+Water
Water
900 800 700
500
Y=12.58 + 897.5*exp(-2*((X-2047)/6.3)^2)
Turb Apex
600
9
500 400
6
300 200
400
3
2036 2038 2040 2042 2044 2046 2048 2050 2052 2054 2056 2058 2060 2062 2064
0
30 25 20
Turb Platform Nitrate Break
100
300 200
15
0
10
Time(min)
100
Turbidity Break
0 0
35
6
12
18 Time(h)
-
600
12
Turb Fit Curve
-
Turb(NTU)
700 Turb(NTU)
Adj.R-Square=0.96313
15
40
24
30
Figure 3. Typical variations of turbidity during filtration and backwash in DNBF
NO3 -N(mg/L)
800
-
NO3 -N
NO3 -N(mg/L)
900
5 0 36
NO3--NOUT
NO3--NIN
1000
Yc=0.26
Yc=0.18
Yc=0.25
Turbidity
35 Yc=0.21
Yc=0.17
Yc=0.20
30 800 25 20 20 15 10
Turb Break-1
Turb Break-1 Turb Break-1
Turb Break-1
Turb Break-1
Turb Break-1
NO3--N(mg/L)
Turbidity(NTU)
600
15 10
5
5
0
0 0
1
2
3
4 5 6 Time (Days)
7
8
9
Figure 4. Variations in the effluent nitrate and turbidity during filtrations and backwashes with real-time control using sodium acetate as carbon sources
-
Yc=0.21
Yc=0.13
35
Yc=0.11 Yc=0.12
Yc=0.11 Yc=0.13
800
30
600
25
400
20 15
15
-
Turbidity(NTU)
1000
Turbidity
NO3 -NOUT
NO3 -N(mg/L)
-
NO3 -NIN
Turb
Turb Break-1
10
Turb Break-1
Turb
Turb
Break-1
Break-1
Break-1
Turb Break-1
10
5
5
0
0 11
0
1
2
3
4
5
6
7
8
9
10
Time(days) Figure 5. Variations in the effluent nitrate and turbidity during filtrations and backwashes with real-time control using methanol as carbon source
OTU5 (9/27) OTU3 (1/29) OTU5 (9/27) Bradyrhizobium sp. D189(AB480442.1) Bradyrhizobium sp. D203a(AB480454.1) OTU3 (1/27) OTU16 (1/27) Bradyrhizobium japonicum USDA 110 DNA(BA000040.2) Bradyrhizobium sp. BTAi1(CP000494.1) OTU10 (1/27) Bosea sp. D257c(AB480483.1) Bosea sp.(HQ916684.1) 100 OTU11 (1/27) 100 Rhodanobacter sp. D206a(AB480456.1) Ochrobactrum sp. clone 1-40(GU136454.1) 100 Ochrobactrum sp. clone 2-80(GU136458.1) OTU15 (1/27) OTU12 (1/27) OTU1 (2/27) OTU9 (1/27) OTU6 (1/27) Mesorhizobium opportunistum WSM2075(CP002279.1) Mesorhizobium sp. GSM-373(FN600572.1) OTU9 (1/29) OTU2 (1/27) OTU14 (1/27) Chelativorans sp. BNC1(CP000390.1) OTU13 (1/27) 69 Paracoccus sp. R-26823(AM230847.1) 63 Paracoccus sp. R-28242(AM230886.1) 100 OTU5 (1/29) Paracoccus sp. R-26824(AM230857.1) OTU14 (1/29) OTU1 (1/29) OTU11 (1/29) OTU4 (10/29) 100 Rhodobacter sphaeroides forma sp.(AJ224908.1) Rhodobacter sphaeroides(U62291.1) OTU4 (3/27) Rhizobium gallicum(CP006880.1) OTU20 (1/27) Rhizobium sp. PY13(DQ096645.1) Rhizobium sp. R-24663(AM230832.1) Hyphomicrobium zavarzinii IFAM ZV-622(AJ224902.1) Hyphomicrobium nitrativorans NL23(CP006912.1) OTU14 (1/29) OTU2 (2/29) OTU7 (3/29) OTU13 (2/29) OTU8 (2/29) OTU6 (1/29) OTU10 (2/29)
100 47 46 49 67 68 49 27
90
42
13
81 13
74 25
4 15
100 46
19 29 23
24
75
100 39
92
16
100 52 60
7
100 69 67
100 62
75 67
96 55 47
Bradyrhizobium
Bosea Rhodanobacter Ochrobactrum
Mesorhizobium
Chelativorans
Paracoccus
Rhodobacter
Rhizobium
Hyphomicrobium
0.05
Figure. 6 NJ phylogenetic tree of the nirK -containing bacteria using methanol and sodium acetate as carbon sources. Sequences of nirK genes using methanol and sodium acetate as carbon sources are shown with "▲ OTU" and "♦OTU", respectively.
Table 1. Experimental procedure and aims Phases Duration Operation and main Parameters Aims (days) 1st 6/15* CS: sodium acetate Determine correlation between biofilm formation and turbidity during the FV:4.76m/h; FV:2.38m/h* start-up of the pilot-scale and lab-scale DNBF nd 2 9-120 CS: sodium acetate; FV: 4.76 m/h (40) Fixed time backwash cycle(48h) Effects of backwash on DNBF operation and establish backwash strength (38) Long-term filtration: Three times Determine the possibility of using turbidity as control parameter to indicate (Each time: filtration 5-6 days and biofilm growth and filter layer condition during long-term filtration. recovery 5days) Establish the methods of determining filter layer clogging degree. (33) Backwash cycle: Turb Break-1 point The correlation between the variations of the effluent turbidity and backwash (FV: 4.76-8.79m/h) rd 3 121-210 FV=8.79m/h Establish the real-time control approach of biofilm growth, removal and (30) CS: sodium acetate;; Real-time control renewal in DNBF via on-line monitoring turbidity. CS: methanol; Test the stability of the real-time control approach using two typical carbon (30) recovery stage sources. (30) Real-time control Analyze biofilm structure and diversity of denitrifying bacteria *in the lab-scale DNBF; CS—Carbon source; Real-time control—backwash was controlled via turbidity;
1
464 465
Optimizing and control of biofilm was studied in biofilter (600 m3/d).
466
Biofilm formation, growth and removal can be controlled using turbidity.
467
The filter layer status can be indicated by Turb break points on turbidity profile.
468
Filter layer clogging coefficient can be used to determine filter layer condition.
469 470
23