Start-up of the anammox process and membrane fouling analysis in a novel rotating membrane bioreactor

Desalination 311 (2013) 46–53

Contents lists available at SciVerse ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Start-up of the anammox process and membrane fouling analysis in a novel rotating membrane bioreactor Tao Jiang a, Hanmin Zhang a,⁎, Hong Qiang a, Fenglin Yang a, Xiaochen Xu a, Hai Du b a b

Key Laboratory of Industrial Ecology and Environmental Engineering, MOE, School of Environmental Science and Technology, Dalian University of Technology, Dalian 116024, China State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian 116024, China

H I G H L I G H T S ► ► ► ►

A rotating flat-sheet membrane bioreactor was employed to start up anammox process. The anammox process was successfully started up within around 16 days. The hydrodynamic conditions were investigated by particle image velocimetry. After 60 days of operation, the membrane fouling in the novel MBR was very slight.

a r t i c l e

i n f o

Article history: Received 3 August 2012 Received in revised form 24 October 2012 Accepted 26 October 2012 Available online 20 December 2012 Keywords: Anaerobic ammonium oxidation (Anammox) Hydrodynamics Membrane bioreactor (MBR) Membrane fouling Particle image velocimetry (PIV)

a b s t r a c t A rotating flat-sheet membrane bioreactor (RFMBR) was employed to start up anammox process, in comparison with a conventional membrane bioreactor (CMBR). The anammox process was successfully started up within around 16 days in both bioreactors. The particle image velocimetry (PIV) analysis showed a larger velocity gradient and a stronger shear stress on membrane surface in RFMBR than in CMBR. At the end of the experiment, the mean particle size of anammox granules achieved 899 μm in RFMBR, while the value reached 809 μm in CMBR, and the trans-membrane pressure (TMP) reached 4 and 16 kPa in RFMBR and CMBR, respectively. Furthermore, the scanning electron microscope (SEM) observation of the biofilm formed on membranes illustrated that a much thinner biofilm with the thickness of 35 μm was formed in RFMBR, compared to the value of 120 μm in CMBR. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.

1. Introduction Discharge of nitrogen from wastewater into surface water bodies may result in eutrophication, toxicity to aquatic species, as well as emissions of nitrous oxide to atmosphere during the denitrification [1]. The traditional biological nitrogen removal process mainly consists of two sub-steps, i.e. autotrophic nitrification and heterotrophic denitrification. However, this traditional process is costly due to needing supplementation of oxygen for nitrification and external carbon sources for denitrification [2], and is complicated when treating highly concentrated nitrogen wastewaters with low C/N ratio [3]. Anaerobic ammonium oxidation (anammox) process, a newly discovered biochemical pathway that allows coupling between ammonium oxidation with nitrite reduction to nitrogen gas (N2) as the terminal product under anoxic conditions [4,5], provides an attractive alternative for nitrogen removal. In fact, the anammox process has since been successfully employed to treat various ammonium-rich wastewaters [6,7]. Nevertheless, start-up of the anammox process is always a challenge for practical ⁎ Corresponding author. Tel.: +86 411 84706172; fax: +86 411 84708083. E-mail address: [email protected] (H. Zhang).

applications [8], due to anammox bacteria, the bacteria responsible for anammox process, growing at a very slow rate with a doubling time of approximately 11 days [9]. Reactor configuration is one of the factors influencing the anammox start-up process. Various reactors were developed and optimized to enrich anammox bacteria and start up anammox process, such as fixed bed biofilm reactor (FBBR) [7,10–12], sequencing batch reactor (SBR) [9,13–15], rotating biological contactor (RBC) [16–18], fluidized bed reactor [4,7], gas-lift reactor [19], granular sludge bed reactor (GSBR) [20], trickling filter [21], upflow anaerobic sludge blanket (UASB) [22], as well as upflow sludge bed filter (UBF) [23]. It is widely acknowledged that the enrichment of slow-growing microorganisms requires efficient retention of biomass [24]. Nevertheless, in the above-mentioned reactors, start-up of the anammox process was inevitably impeded by a continuous loss of anammox biomass via the effluent, leading to more difficult cultivation of the biomass [9,24]. In recent years, the membrane bioreactor (MBR) becomes a new hotspot for start-up of the anammox process due to its full biomass retention, either immersed MBR (iMBR) [8,24,25] or sidestream MBR (sMBR) [2,26]. For the more compact and energy efficient immersed iMBR [27], a significantly shorter start-up period and higher biomass

0011-9164/$ – see front matter. Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.desal.2012.10.031

T. Jiang et al. / Desalination 311 (2013) 46–53

purity was obtained in previous study compared to other reactor configurations. But membrane fouling was still an issue in these investigations. For example, in the experiment of Wang et al. [8], membrane pressure increased rapidly in the first 2 weeks and the module was chemically cleaned on day 43. Van de Star et al. [24] replaced membrane module every 10 days to prevent biofilm growth on the membrane surface. In the research of Trigo et al. [25] that employed a membrane sequencing batch reactor (MSBR), a backwashing period of 3 min was set in a 6 h cycle to minimize fouling and the permeation time was only 18 min in a cycle. Recently, shear-enhanced membrane filtration using a moving membrane module to mitigate membrane fouling has attracted much attention. Zuo et al. [28] introduced a new bioreactor named submerged rotating membrane bioreactor (SRMBR), the membrane module of which comprised several rotatable round flat-sheets and a hollow rotating axis. The flat-sheet plates are parallel to each other, and rotate around the hollow rotating axis. The equilibrium permeate flux rose with the increase in rotary speed of membrane plates, proving that rotation of membrane module could enhance shear forces on membrane surface and mitigate membrane fouling. But the SRMBR also had some drawbacks, such as weak turbulence created by the membrane module, as well as difficult washing and replacement of the membrane material. In this research, a rotating flat-sheet membrane bioreactor (RFMBR) was proposed and used to start up anammox process, in comparison with a conventional membrane bioreactor (CMBR). The membrane fouling was analyzed through the trans-membrane pressure (TMP) rise and biofilm formation on membranes. The particle image velocimetry (PIV), a widely used technique that can provide velocity field in fluids [29], was employed to investigate the hydrodynamic conditions in the reactors. In addition, the morphology of anammox granules and relative abundance of anammox bacteria in total bacteria was also analyzed by scanning electron microscope (SEM) and florescence in situ hybridization (FISH), respectively. 2. Materials and methods 2.1. Experimental set-up The RFMBR and CMBR are the same as those in our previous study [30]. The membrane module of RFMBR is composed of 9 flat-sheets and 2 plates. The diameter of each plate is 160 mm, and the effective height and width of each flat-sheet is 153 and 39 mm, respectively. The two sides of each flat-sheet are covered with polyvinylidene difluoride (PVDF) membrane with an average pore size of 0.2 μm. The flat-sheets are vertically and symmetrically placed on the edge of the plates. The angle between plane of each flat-sheet and its corresponded radius is fixed at 30°. The hub of the bottom plate is connected with gears that are driven by an adjustable speed electromotor. Two flat-sheet membrane modules are vertically and symmetrically installed in the two sides of the CMBR. The membrane material is the same PVDF used in RFMBR. An outlet connected with external outlet pipe is opened on the bottom of each module. The reactor is equipped with a mechanical stirrer driven by an electromotor in the middle. The total effective filtration of both reactors is 0.09 m 2. 2.2. Operational strategy Each reactor was inoculated with anammox activated sludge from an upflow anaerobic sludge blanket (UASB) which had been operating for more than 2 years in our laboratory. After inoculation, the initial SS in each reactor was 2232 mg/L. The MBRs were continuously fed with the same synthetic wastewater (medium composition shown in Table 1) by peristaltic pumps and were operated in the mode of constant flux for 3 stages: the flux was 10, 6 and 8.5 L/(m2h) for stage I (days 0–8), II (days 8–37) and III (days 37–61), corresponded to a HRT of 14.4, 24 and 17 h, respectively. The rotational speed of the

47

Table 1 Medium composition. Medium composition

Concentration (mg/L)

Medium composition

Concentration (mg/L)

(NH4)2SO4 KH2PO4 Trace solution I Trace solution composition I EDTA Trace solution composition II KCl CaCl2·2H2O

550 50 1 mL/L Concentration (g/L) 10 Concentration (g/L) 1.4 1.4

NaNO2 KHCO3 Trace solution II Trace solution composition I FeSO4·7H2O Trace solution composition II NaCl FeSO4·7H2O

300 500 1 mL/L Concentration (g/L) 18 Concentration (g/L) 1 1

membrane module in RFMBR was moderately set to 20 rpm according to our previous study [30], when the membrane fouling rate was relatively low; whereas the stirrer in CMBR worked at a speed of 60 rpm in order to provide enough forces to make the biomass suspended. The synthetic wastewater was replaced every day to avoid the changes in feed composition and was purged with pure nitrogen gas during the preparation process to remove the oxygen (influent DOb 0.1 mg/L). The temperature in the MBRs was maintained at about 33 °C. The sludge retention time (SRT) was infinite since there was no sludge waste from the reactors during the whole experimental trial except for sampling. According to the previous reported ratio of nitrite consumption to ammonia consumption in anammox reaction (1.32) [9], the medium concentrations of (NH4)2SO4 and NaNO2 were initially set to about 150 and 200 mg N/L, respectively, to maintain the ratio of ammonia to nitrite at 1:1.33, and the N-loading rate was changed by varying the HRT. 2.3. Analytical methods 2.3.1. Chemical analysis The concentrations of ammonium, nitrite, nitrate, suspended solids (SS) and volatile suspended solids (VSS) were determined according to standard methods for the examination of water and wastewater described in detail by American Public Health Association [31]. DO and pH were measured by a DO meter (YSI55/12FT, USA) and a pH meter (Sartorius PB-10, Germany), respectively. The particle size was obtained with a laser particle size analysis system (Mastersizer 2000, Malvern, UK). The extraction of bound extracellular polymeric substances (EPS), normalized as the sum of protein (PN) and polysaccharide (PS), was performed based on a cation ion exchange resin (Dowex-Na form) method [32]. For quantitative analysis of proteins, the modification of Lowry method described by Frolund et al. [33] was used with bovine serum albumin as standard. The anthrone method modified by Raunkjer et al. [34] was employed for the quantification of polysaccharides with glucose as standard. 2.3.2. Fluorescence in situ hybridization (FISH) analysis FISH analysis was used to investigate the proportion of anammox bacteria to background bacteria. Paraformaldehyde cell fixation and FISH analysis were performed according to the standard hybridization protocol [35,36], using oligonucleotide probes for eubacterium (EUB338 plus) and anammox bacteria (AMX820) [21,37]. The hybridization was performed on 4% (w/v) paraformaldehyde-fixed sludge samples. A Leica TCS-SP2 confocal scanning laser microscope (CSLM) (Leica, Germany) was employed to acquire images. 2.3.3. Morphological observation The granule morphology was analyzed by a camera (Canon EOS 550D, Japan), a light microscope (Olympus CX21, Japan), as well as the scanning electron microscope (SEM, JEOL JSM-5600LV, Japan). The surface and section morphology of the biofilm on fouled membranes

48

T. Jiang et al. / Desalination 311 (2013) 46–53

was also observed by the SEM. The preparation of samples used for SEM observation was as followed. The samples were first washed with phosphate-buffer solution (PBS, pH 7.5) and then were fixed with 2.5% glutaraldehyde solution for 12 h overnight, followed by dehydration using a series of tertiary butanol–water mixture (50, 70, 80, 90, 95, 100, 100 vol.% of tertiary butanol) with order, each mixture for 15 min. Further, the samples were dried and coated with a thin gold layer for observation. 2.3.4. Characterization of membrane fouling Membrane fouling was characterized by trans-membrane pressure (TMP) that was measured with pressure sensors (AOB 131, Shanghai Aobo Automation Equipment Co., Ltd., China). The pressure sensors were connected to a 32-channel data acquisition system (PISO-813, ICP-DAS) linked with a personal computer via a PCI interface. The system collected TMP data every 30 min. 2.4. PIV measurement system and instrumentation We employed the same PIV system and experimental method used in our previous study in this research [30]. Pine pollen was utilized as the tracer particle and clean water as the experimental liquid. The particle relaxation time is in the order of 10−8 s, much shorter than the time interval between successive exposures. The flat-sheets of RFMBR and membrane modules of CMBR were replaced by the one-piece glass in the experiments. For RFMBR, the laser light sheet was parallel to the rotary plates, while for CMBR, the light sheet was fixed near the internal surface of one membrane module. The sizes of images captured by the camera for RFMBR and CMBR were 192.0 mm× 143.0 mm and 112.0 mm× 83.4 mm, respectively. The rotating directions of both membrane module and stirrer were anticlockwise viewing from the top. 3. Results and discussion

by reducing permeate flux. In this stage, the ammonium and nitrite removal rates continued to increase, and the removal efficiency of ammonium and nitrite reached about 90% on about day 16 in both RFMBR and CMBR, indicating a successful start-up of the anammox process. The MBRs in our study exhibited a superior performance on the aspect of start-up time, compared with other reactor configurations as well as MBRs in other researches (Table 2). In stage III from day 37, the ammonium and nitrite loading rates were elevated to ca. 208 and 273 mg/(Ld), respectively. In this stage, the ammonium and nitrite removal efficiency remained at a very high level (both with an average of more than 92%), showing an excellent anammox activity. In the initial period of the operation, the NH4+-N and NO2−-N removal rates were much higher in CMBR than in RFMBR. This may be due to the anammox biomass not adapting to the intense hydrodynamic conditions in RFMBR at the beginning of the operation. As the experiment progressed, the difference in removal rate decreased gradually and the nitrite removal rate and ammonium removal rate in RFMBR began to exceed those in CMBR from about day 25 and day 50, respectively. But during the late stage of the process, the removal rates in CMBR can sometimes be higher than those in RFMBR, and the removal rates were very close in both reactors, showing that the hydrodynamic conditions may have a slight impact on the nitrogen removal in the stable stage. The average ratio of nitrite consumption to ammonia consumption was 1.24:1 and 1.31:1 in CMBR and RFMBR, respectively, close to the value of 1.32:1 reported by Strous et al. [9]. From Fig. 1B, it can be observed that the pH in effluent was higher than that in influent in the two reactors due to the acidity consumption in the anammox reaction, similar to the previous studies [38,39]. Nitrate production is considered as an important indicator of anammox process, since the oxidation of nitrite to nitrate through nitrate reductase is vital to yield energy for the growth of anammox bacteria [40]. The high initial nitrate concentration may be due to the extracellular oxidation of nitrite to nitrate by the residual oxygen in the reactors in the beginning and as the experiment progressed, the nitrate production was nearly consistent with the anammox performance.

3.1. Nitrogen removal performance of the reactors During stage I, in which the NH4+-N and NO2−-N loading rate was 261 and 342 mg/(Ld) (average of the experimental values), respectively, the initial NH4+-N and NO2−-N removal rate of both RFMBR and CMBR was low, with a gradual increase with time, as shown in Fig. 1A. This may be due to the anammox biomass not adapting to the new environment in the beginning, including a too high nitrogen loading rate (NLR) in stage I. On day 8, the ammonium and nitrite loading rates were adjusted to ca. 156 and 245 mg/(Ld) respectively

3.2. Fluid velocity fields The velocity fields were measured in horizontal plane for RFMBR and vertical plane for CMBR. For RFMBR, the horizontal plane was measured because it could directly reflect the overall fluid velocity situation, near the membrane surface as well as in the mixed liquor bulk, in one velocity map. For different horizontal planes in RFMBR, the velocity fields should be similar, but for vertical planes with different areas, the velocity fields would be strongly different and a mass of vertical

Fig. 1. Change in removal rates of NH4+-N and NO2−-N (A), and change in NO3−-N concentration and pH (B) during the period of operation in CMBR and RFMBR.

T. Jiang et al. / Desalination 311 (2013) 46–53 Table 2 Overview of the start-up time of anammox process in different reactor set-ups. Reactor Seed sludge

Inlet

Start-up time

Reference

MBR

Synthetic mediuma

2 months

[8]

Synthetic mediuma Anaerobic digester effluents Synthetic mediuma

>300 days [12] 35 days [14] 101 days

[15]

Synthetic mediuma

105 days

[20]

Synthetic mediuma Synthetic mediuma Synthetic mediuma

31 days >80 days 16 days

[23] [25] This study

FBBR SBR SBR GSBR UBF MSBR MBR

Aerobic activated sludge Denitrifying sludge Anammox sludge Aerobic activated sludge Nitrifying activated sludge Activated sludge Anammox sludge Anammox sludge

49

consequently, the shear stress related to velocity gradient in RFMBR was larger than that in CMBR. 3.3. Characteristics of the anammox sludge

a

The Medium compositions of the synthetic medium are not the same but are similar.

planes must be measured to display an overall fluid velocity situation. Furthermore, due to the module configuration of RFMBR, the velocity fields in vertical planes are difficult to be obtained. For CMBR, the velocity fields near the membrane module were measured to reflect the hydrodynamic conditions near the membrane surface directly and in the mixed liquor bulk indirectly. So through comparing the velocity fields in horizontal plane for RFMBR and vertical plane for CMBR, we can investigate the fluid situations near the membrane surface as well as in the mixed liquor bulk in the reactors, and thus to study the impact of hydrodynamics on the membrane fouling and the graduation of anammox biomass. The time-averaged velocity profiles shown in Fig. 2 are average of 50 instantaneous velocity profiles. It can be observed from Fig. 2A that the maximal fluid velocity in RFMBR appeared near the rotating plates; when moving to the inner part, the fluid velocity decreased gradually. Moreover, the decrease in fluid velocity from rotating plates to inner part showed a clear gradient. Since the stirrer in CMBR rotated anticlockwise viewing from the top, we can see from Fig. 2B that the direction of flow at the bottom of velocity field picture was from left to right, and when moving to the upper part the flow changed to two distinct parts because of the lift force induced by the stirrer. For the left and right part, the direction of the flow was upper right and downward, respectively, so a weak re-circulation flow zone was formed. As shown in Fig. 2, the magnitude of the fluid velocity in RFMBR was much larger than that in CMBR (p b 0.001). It can also be observed that the gradient of the velocity was much more obvious for RFMBR compared to CMBR;

3.3.1. Mean particle size The mean particle size of the anammox sludge in the two MBRs during the operation was investigated (Fig. 3). The mean particle size of the seed sludge was 366±44 μm and as the experiment went along, the mean particle size reached 780±20 and 820±10 μm on day 48 in CMBR and RFMBR, respectively, and reached 809±7 and 899±20 μm, respectively, on day 61, indicating a faster granulation process in RFMBR than in CMBR. 3.3.2. Morphological observation The morphology of anammox granules in the two reactors also had differences with each other (Fig. 4). The photographs and light micrographs of the granules showed that the granules in RFMBR had a darker red color compared to those in CMBR (Fig. 4A, B, C and D). The SEM photos of the granular sludge in RFMBR showed a high degree of compactness (Fig. 4F and H); while the granules in CMBR were relatively loose (Fig. 4E and G). From the exterior view, each microelement of the granules in RFMBR was tightly integrated with other portions and there was little interspace among them, which is in favor of the granular sludge joining tightly and existing stably; while the granular surface for CMBR was rough and heterogeneous with irregular shapes like lobate extrusions and cavities. The structure of granules in RFMBR exhibited sphericity, and this structure may have been formed by the intense shear forces caused by rotation of the membrane module. Contrastively, the granule structure for CMBR was elliptical due to the relatively weak shear stress. 3.3.3. EPS analysis It is well known that the EPS plays an important role in the evolution from flocs to granules, since it is helpful for forming multilevel porous structures in which macroflocs are composed of smaller aggregates or microcolonies because of the cohesion of microorganisms [41]. The EPS amount was larger in CMBR than in RFMBR during the operation (Fig. 5), with 49 ±2 mg/g VSS in CMBR and 27±4 mg/g VSS in RFMBR on day 48, and 22 ±5 mg/g VSS in CMBR and 16 ± 5 mg/g VSS in RFMBR on day 61. Previous studies have proved that augmenting aeration and lengthening sludge retention time can increase EPS amount [42], but little is known about the relations between hydrodynamics and EPS amount for anammox bacteria. As shown in this study, the hydrodynamic conditions can influence the EPS secretion of anammox

Fig. 2. Time-averaged velocity fields in RFMBR at the speed of 20 rpm (A) and CMBR at the speed of 60 rpm (B). The unit of velocity legend is m/s. Note the different scales in A and B.

50

T. Jiang et al. / Desalination 311 (2013) 46–53

Fig. 3. Change in mean particle size of the anammox sludge in CMBR and RFMBR.

biomass, and specifically intense shear stress can inhibit the EPS secretion. The reason may be due to the intense shear stress making the anammox granules more compact (see Section 3.3.2), inhibiting the EPS secretion of the inner anammox biomass. Past studies generally considered that because of the high content of negatively charged amino acids, protein was more involved than polysaccharide in electrostatic bonds with multivalent cations, a critical factor in stabilizing the structure of aggregates, and therefore, a high PN/PS is favorable for aggregate structure stability [43,44]. On day 48, PN/PS was 10.1 and 28.4 in CMBR and RFMBR, respectively, and the ratio was 6.6 and 14.7, respectively, on day 61, implying that the granules had a more stable structure in RFMBR than in CMBR. The reason may be that due to the strong shear stress in RFMBR, the granules in the reactor must possess a more stable structure in order to adapt to the intense hydrodynamic conditions. 3.3.4. FISH analysis Sludge samples were analyzed by FISH technique at the end of the experiment to investigate the relative abundance of anammox bacteria in total bacteria. AMX820 probe was used to target anammox bacteria, and EUB338 plus probe was employed to target total eubacteria. FISH analysis showed that the percentage of anammox bacteria in the two reactors had no significant difference, both accounting for around 90% of total bacteria (Fig. 6), implying that hydrodynamic conditions had a very slight impact on the relative abundance of anammox bacteria. 3.4. Membrane fouling analysis 3.4.1. TMP profiles In fact, except for adjusting the nitrogen loading rate, operating the reactors under different constant flux was to tentatively determine the critical flux of the reactors, and then we can operate them slightly above the critical flux to investigate the fouling characteristics of the reactors. Analyzing the data presented in Fig. 7, we can conclude that the critical flux of RFMBR and CMBR, although not the same, were both above 6 L/(m2h) and below 10 L/(m2h), since the TMP remained almost unchanged under 6 L/(m2h) and increased rapidly under 10 L/(m2h). During stage III, of which the flux was 8.5 L/(m2h), for CMBR, the TMP increased with a linear trend and reached around 16 kPa at the end of the experiment; while for RFMBR, the TMP reached about 4 kPa in the end. So it can be concluded that the TMP for CMRB increased much more rapidly than that for RFMBR, indicating a more severe fouling for CMBR than for RFMBR. It is widely acknowledged that the solid particle size in the mixed liquid is an important factor affecting membrane permeability. Lots of previous research reported that the membrane fouling mitigated with

Fig. 4. Photographs (A, B) and light micrographs (C, D) of anammox granules on day 48 in CMBR (A, C) and RFMBR (B, D); and SEM images of anammox granules on day 61 in CMBR (E, G) and RFMBR (F, H).

the increase in mean particle size [27]. As shown in Section 3.3.1 and this section, the membrane fouling rate in RFMBR was much lower than that in CMBR which possessed smaller mean particle size. This may be attributed to the deposition of particles on the membrane surface; the larger the particles, the more easily they were impeded from depositing on the membrane surface by shear forces induced by rotation of membrane module/stirrer. EPS also has a significant effect on membrane fouling, and many previous findings showed that membrane fouling was negatively associated with EPS [45,46], which is consistent with the result in this study. EPS influences membrane fouling in two ways: 1) due to the heterogeneous nature of EPS, it could build a highly hydrated gel matrix in which microbial cells are embedded and could thus assist to

T. Jiang et al. / Desalination 311 (2013) 46–53

EPS/mg/g VSS or PN/PS

50 45

CMBR, PN/PS

40

RFMBR, PN/PS

35

CMBR, EPS

30

RFMBR, EPS

PN

25 20 15

PN PN

10

PS 5

PN

PS PS

PS

51

indicating that the mechanical forces are helpful for diminution of the biofilm. It is well known that the filtration resistances in MBRs included membrane resistance, cake (biofilm) resistance, as well as blocking and irremovable fouling resistance. Lee et al. [47] also reported that for suspended microbial culture systems the proportion was about 12%, 80% and 8% for membrane resistance, cake resistance, and blocking and irremovable fouling resistance, respectively, which means that the cake resistance possessed the main proportion of the filtration resistances. Accordingly, removal of the biofilm on membranes is quite favorable for the mitigation of membrane fouling, especially during operation of the MBR, since the blocking resistance is usually rid by chemical cleaning, which is unpractical during the operation. We observed that the surface of biofilm in RFMBR was a bit more compact than that in CMBR (Fig. 8C and D, crack was produced because of the preparation process of samples for SEM observation), due to the stronger mechanical forces, but the difference was not so obvious.

0 48

61

4. Conclusions

Time/days Fig. 5. Change in EPS in CFMBR and RFMBR.

form a significant barrier to permeate flow in membrane processes [27]; and 2) EPS could have a great effect on other sludge characteristics, such as sludge volume index (SVI), hydrophobicity, surface charge, and sludge viscosity; thus, the change in EPS may result in a more significant change in sludge characteristics, resulting in a notable influence on membrane fouling. 3.4.2. SEM observation of biofilm on membranes The biofilm formed on fouling membranes in the MBRs on day 61, i.e. at the end of the experiment, was observed by SEM (Fig. 8). It can be seen that the thickness of biofilm in RFMBR was much thinner than that in CMBR, and corresponded to 35 and 120 μm, respectively,

MBR is a suitable and promising reactor to start up anammox process because of its full biomass retention. But the membrane fouling is still a challenge in the practical application. In this study, a rotating flat-sheet membrane bioreactor (RFMBR) was designed from a fouling reduction point of view. The RFMBR was used to start up anammox process, in comparison with a conventional membrane bioreactor (CMBR). More specific outcomes of this study were as followed: (1) The anammox process was successfully started up within about 16 days in RFMBR and CMBR. The nitrogen removal rate was higher in CMBR than in RFMBR in the initial period of the operation, but as the experiment went along, the difference decreased gradually and the nitrogen removal rates were close in the late stage of the process. (2) The anammox granules possessed a more compact structure as well as secreted less EPS in RFMBR than in CMBR. Furthermore,

Fig. 6. FISH micrographs of sludge samples from CMBR (A, B, C) and RFMBR (D, E, F) on day 61: green color indicates all bacteria hybridized with EUB338 plus probe, and blue color indicates anammox bacteria hybridized with AMX820 probe. C and F are synthesized images of A and B, and D and E, respectively. Bar: 150 μm.

52

T. Jiang et al. / Desalination 311 (2013) 46–53

Fig. 7. TMP profiles during 61 days of operation in CMBR and RFMBR: stage I: 10 L/(m2h), stage II: 6 L/(m2h), stage III: 8.5 L/(m2h).

the final mean particle size of granules reached 899 μm in RFMBR, higher than the value of 809 μm in CMBR. (3) PIV analysis showed a larger velocity gradient and a stronger shear stress on membrane surface in RFMBR at 20 rpm than in CMBR at 60 rpm. (4) The final TMP reached 4 and 16 kPa in RFMBR and CMBR, respectively. The thickness of biofilm formed on membrane was 35 μm in RFMBR, much thinner than the value of 120 μm in CMBR.

Acknowledgments This work was financially supported by the Fundamental Research Funds for the Central Universities and the Specialized Research Fund for the Doctoral Program of Higher Education of China (No. 20110041110002). The authors would also like to express their sincere appreciation to the Editor and Reviewers for their helpful suggestions and comments.

Fig. 8. SEM images of biofilm on the PVDF membranes at the end of operation in CMBR (A, C) and RFMBR (B, D): (A, B) cross section, (C, D) surface. Note the different scales in A and B.

T. Jiang et al. / Desalination 311 (2013) 46–53

References [1] S. Philips, H.J. Laanbroek, W. Verstraete, Origin, causes and effects of increased nitrite concentrations in aquatic environments, Rev. Environ. Sci. Biotechnol. 1 (2002) 115–141. [2] S.Q. Ni, P.H. Lee, A. Fessehaie, B.Y. Gao, S. Sung, Enrichment and biofilm formation of anammox bacteria in a non-woven membrane reactor, Bioresour. Technol. 101 (2010) 1792–1799. [3] N. Chamchoi, S. Nitisorvut, J.E. Schmidt, Inactivation of ANAMMOX communities under concurrent operation of anaerobic ammonium oxidation (ANAMMOX) and denitrification, Bioresour. Technol. 99 (2008) 3331–3336. [4] A. Mulder, A.A. Graaf, L.A. Robertson, J.G. Kuenen, Anaerobic ammonium oxidation discovered in a denitrifying fluidized bed reactor, FEMS Microbiol. Ecol. 16 (1995) 177–184. [5] A.A. Van de Graaf, A. Mulder, P. Debruijn, M.S.M. Jetten, L.A. Robertson, J.G. Kuenen, Anaerobic oxidation of ammonium is a biologically mediated process, Appl. Environ. Microbiol. 61 (1995) 1246–1251. [6] A. Joss, D. Salzgeber, J. Eugster, R. Konig, K. Rottermann, S. Burger, P. Fabijan, S. Leumann, J. Mohn, H. Siegrist, Full-scale nitrogen removal from digester liquid with partial nitritation and anammox in one SBR, Environ. Sci. Technol. 43 (2009) 5301–5306. [7] M. Strous, E. Van Gerven, P. Zheng, J.G. Kuenen, M.S.M. Jetten, Ammonium removal from concentrated waste streams with the anaerobic ammonium oxidation (anammox) process in different reactor configurations, Water Res. 31 (1997) 1955–1962. [8] T. Wang, H.M. Zhang, F.L. Yang, S.T. Liu, Z.M. Fu, H.H. Chen, Start-up of the anammox process from the conventional activated sludge in a membrane bioreactor, Bioresour. Technol. 100 (2009) 2501–2506. [9] M. Strous, J.J. Heijnen, J.G. Kuenen, M.S.M. Jetten, The sequencing batch reactor as a powerful tool for the study of slowly growing anaerobic ammonium-oxidizing microorganisms, Appl. Microbiol. Biotechnol. 50 (1998) 589–596. [10] C. Fux, V. Marchesi, I. Brunner, H. Siegrist, Anaerobic ammonium oxidation of ammonium-rich waste streams in fixed-bed reactors, Water Sci. Technol. 49 (2004) 77–82. [11] T. Kindaichi, I. Tsushima, Y. Ogasawara, M. Shimokawa, N. Ozaki, H. Satoh, S. Okabe, In situ activity and spatial organization of anaerobic ammoniumoxidizing (anammox) bacteria in biofilms, Appl. Environ. Microbiol. 73 (2007) 4931–4939. [12] I. Tsushima, Y. Ogasawara, T. Kindaichi, H. Satoh, S. Okabe, Development of high-rate anaerobic ammonium-oxidizing (anammox) biofilm reactors, Water Res. 41 (2007) 1623–1634. [13] A. Dapena-Mora, J.L. Campos, A. Mosquera-Corral, M.S.M. Jetten, R. Mendez, Stability of the ANAMMOX process in a gas-lift reactor and a SBR, J. Biotechnol. 110 (2004) 159–170. [14] J.R. Vazquez-Padin, M.J. Pozo, M. Jarpa, M. Figueroa, A. Franco, A. Mosquera-Corral, J.L. Campos, R. Mendez, Treatment of anaerobic sludge digester effluents by the CANON process in an air pulsing SBR, J. Hazard. Mater. 166 (2009) 336–341. [15] T. Wang, H.M. Zhang, D.W. Gao, F.L. Yang, G.Y. Zhang, Comparison between MBR and SBR on anammox start-up process from the conventional activated sludge, Bioresour. Technol. 122 (2012) 78–82. [16] K. Egli, U. Fanger, P.J.J. Alvarez, H. Siegrist, J.R. Van der Meer, A.J.B. Zehnder, Enrichment and characterization of an anammox bacterium from a rotating biological contactor treating ammonium-rich leachate, Arch. Microbiol. 175 (2001) 198–207. [17] S.T. Liu, F.L. Yang, Y. Xue, Z. Gong, H.H. Chen, T. Wang, Z.C. Su, Evaluation of oxygen adaptation and identification of functional bacteria composition for anammox consortium in non-woven biological rotating contactor, Bioresour. Technol. 99 (2008) 8273–8279. [18] K. Pynaert, B.F. Smets, S. Wyffels, D. Beheydt, S.D. Siciliano, W. Verstraete, Characterization of an autotrophic nitrogen-removing biofilm from a highly loaded lab-scale rotating biological contactor, Appl. Environ. Microbiol. 69 (2003) 3626–3635. [19] A.O. Sliekers, K.A. Third, W.R. Abma, J.G. Kuenen, M.S.M. Jetten, CANON and anammox in a gas-lift reactor, FEMS Microbiol. Lett. 218 (2003) 339–344. [20] P. Zheng, F.M. Lin, B.L. Hu, J.S. Chen, Start-up of anaerobic ammonia oxidation bioreactor with nitrifying activated sludge, J. Environ. Sci. (China) 16 (2004) 13–16. [21] M. Schmid, U. Twachtmann, M. Klein, M. Strous, S. Juretschko, M. Jetten, J.W. Metzger, K.H. Schleifer, M. Wagner, Molecular evidence for genus level diversity of bacteria capable of catalyzing anaerobic ammonium oxidation, Syst. Appl. Microbiol. 23 (2000) 93–106.

53

[22] Y.H. Ahn, I.S. Hwang, K.S. Min, ANAMMOX and partial denitritation in anaerobic nitrogen removal from piggery waste, Water Sci. Technol. 49 (2004) 145–153. [23] R.C. Jin, P. Zheng, A.H. Hu, Q. Mahmood, B.L. Hu, G. Jilani, Performance comparison of two anammox reactors: SBR and UBF, Chem. Eng. J. 138 (2008) 224–230. [24] W.R.L. Van der Star, A.I. Miclea, U. Van Dongen, G. Muyzer, C. Picioreanu, M.C.M. Van Loosdrecht, The membrane bioreactor: a novel tool to grow anammox bacteria as free cells, Biotechnol. Bioeng. 101 (2008) 286–294. [25] C. Trigo, J.L. Campos, J.M. Garrido, R. Mendez, Start-up of the anammox process in a membrane bioreactor, J. Biotechnol. 126 (2006) 475–487. [26] Y. Tao, D.W. Gao, Y. Fu, W.M. Wu, N.Q. Ren, Impact of reactor configuration on anammox process start-up: MBR versus SBR, Bioresour. Technol. 104 (2012) 73–80. [27] S. Judd, The MBR Book: Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment, Elsevier, Oxford, 2006. [28] D.Y. Zuo, H.J. Li, H.T. Liu, G.P. Wu, A study on submerged rotating MBR for wastewater treatment and membrane cleaning, Korean J. Chem. Eng. 27 (2010) 881–885. [29] L. Martinelli, C. Guigui, A. Line, Characterisation of hydrodynamics induced by air injection related to membrane fouling behavior, Desalination 250 (2010) 587–591. [30] T. Jiang, H.M. Zhang, D.W. Gao, F.L. Yang, H. Du, Relationships between mechanicallyinduced hydrodynamics and membrane fouling in a novel rotating membrane bioreactor, Desalin. Water Treat., in press. [31] APHA, Standard Methods for the Examination of Water and Wastewater, 19th ed., 1995. (New York, USA). [32] B. Frolund, R. Palmgren, K. Keiding, P.H. Nielsen, Extraction of extracellular polymers from activated sludge using a cation exchange resin, Water Res. 30 (1996) 1749–1758. [33] B. Frolund, T. Griebe, P.H. Nielsen, Enzymatic-activity in the activated-sludge floc matrix, Appl. Microbiol. Biotechnol. 43 (1995) 755–761. [34] K. Raunkjer, T. Hvitved-Jacobsen, P.H. Nielsen, Measurement of pools of protein, carbohydrate and lipid in domestic wastewater, Water Res. 28 (1994) 251–262. [35] R.I. Amann, In situ identification of microorganisms by whole cell hybridization with rRNA-targeted nucleic acid probes, in: A.D.L. Akkermans, J.D. Van Elsas, F.J. De Bruijn (Eds.), Molecular Microbial Ecology Manual, Kluwer Academic Publishers, Dordrecht, The Netherlands, 1995, pp. 1–15. [36] R.I. Amann, B.J. Binder, R.J. Olson, S.W. Chisholm, R. Devereux, D.A. Stahl, Combination of 16s rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations, Appl. Environ. Microbiol. 56 (1990) 1919–1925. [37] H. Daims, A. Bruhl, R. Amann, K.-H. Schleifer, M. Wagner, The domain-specific probe EUB338 is insufficient for the detection of all bacteria: development and evaluation of a more comprehensive probe set, Syst. Appl. Microbiol. 22 (1999) 434–444. [38] C.J. Tang, P. Zheng, Q. Mahmood, J.W. Chen, Start-up and inhibition analysis of the anammox process seeded with anaerobic granular sludge, J. Ind. Microbiol. Biotechnol. 36 (2009) 1093–1100. [39] L. Zhang, J.C. Yang, Y.G. Ma, Z.G. Li, T. Fujii, W.J. Zhang, N. Takashi, K. Furukawa, Treatment capability of an up-flow anammox column reactor using polyethylene sponge strips as biomass carrier, J. Biosci. Bioeng. 110 (2010) 72–78. [40] H.S. Li, S.Q. Zhou, W.H. Ma, G.T. Huang, B. Xu, Fast start-up of ANAMMOX reactor: operational strategy and some characteristics as indicators of reactor performance, Desalination 286 (2012) 436–441. [41] D.S. Henriques, N.G. Love, The role of extracellular polymeric substances in the toxicity response of activated sludge bacteria to chemical toxins, Water Res. 41 (2007) 4177–4185. [42] G.Y. Long, P.T. Zhu, Y. Shen, M.P. Tong, Influence of extracellular polymeric substances (eps) on deposition kinetics of bacteria, Environ. Sci. Technol. 43 (2009) 2308–2314. [43] C.S. Laspidou, B.E. Rittmann, A unified theory for extracellular polymeric substances, soluble microbial products, and active and inert biomass, Water Res. 36 (2002) 2711–2720. [44] S. Qiao, Y. Kawakubo, Y.J. Cheng, T. Nishiyama, T. Fujii, K. Furukawa, Identification of bacteria coexisting with anammox bacteria in an upflow column type reactor, Biodegradation 20 (2009) 117–124. [45] B.D. Cho, A.G. Fane, Fouling transients in nominally sub-critical flux operation of a membrane bioreactor, J. Membr. Sci. 209 (2002) 391–403. [46] S. Rosenberger, M. Kraume, Filterability of activated sludge in membrane bioreactors, Desalination 151 (2003) 195–200. [47] J. Lee, W.Y. Ahn, C.H. Lee, Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor, Water Res. 35 (2001) 2435–2445.