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

Available online at www.sciencedirect.com

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

Nitrication performance in a membrane bioreactor treating industrial wastewater

k a, Jan Svojitka b, Ji  Wanner c, Thomas Wintgens b,* s Dvo ra Luka r
2, Centre for Nanomaterials, Advanced Technologies and Innovations, Technical University of Liberec, Studentska 461 17 Liberec 1, Czech Republic b Institute for Ecopreneurship, University of Applied Sciences and Arts Northwestern Switzerland, Gru ndenstrasse 40, CH-4132 Muttenz, Switzerland c Department of Water Technology and Environmental Engineering, Institute of Chemical Technology, Prague, 5, 166 28 Prague 6 e Dejvice, Czech Republic Technicka
a

article info
Article history: Received 7 November 2012 Received in revised form 20 March 2013 Accepted 24 March 2013 Available online 16 April 2013 Keywords: Membrane bioreactor Industrial wastewater Activated sludge Nitrication Adaptation Inhibition

abstract
The inuence of industrial (pharmaceutical and chemical) wastewater composition on membrane bioreactor (MBR) performance was investigated in a pilot-scale installation. The study focussed on nitrication performance, which was evaluated based on inuent and efuent parameters as well as batch nitrication rate tests. The industrial wastewater was pumped into the MBR in a mixture with municipal wastewater at constant ow rate. The loading of the MBR with industrial wastewater was increased stepwise from 0 to 75% share in the mixed inuent to study the adaptation of nitrifying bacteria. Stable nitrication performance was observed until the content of industrial wastewater in the inuent reached 40%, with efuent values of around 0.56 mg L1 NH4-N and 98.3% ammonia removal. Breakdown of nitratation was observed at a 40% industrial wastewater dose and breakdown of nitritation at a 50% dose, respectively. However, after several months of adaptation, both processes recovered. No nitrication was observed when the industrial wastewater share exceeded 50%. Adaptation of nitrifying bacteria in the MBR was also conrmed by results of kinetic tests. The inhibition effect of the concentrated industrial wastewater to the MBR sludge decreased substantially after several months of exposure, while the inhibition of referential activated sludge remained constant. 2013 Published by Elsevier Ltd.

1.

Introduction

Treatment of wastewater originating from pharmaceutical and chemical industry is difcult due to large variations in its volume and composition, high concentration of organic

matter and salts or presence of poorly biodegradable organic compounds or substances (Chang et al., 2008; Chelliapan et al., 2011; Noble, 2006). Besides that, biological treatment of such wastewaters can be negatively affected by the presence of compounds with inhibiting properties towards the activated

Abbreviations: COD, chemical oxygen demand; MWCO, molecular weight cut-off; MLSS, mixed liquor suspended solids; SRT, sludge retention time; HRT, hydraulic retention time; F/M, food to microorganisms ratio; PVDF, polyvinylidene uoride; TMP, transmembrane pressure; AUR, ammonia uptake rate; OUR, oxygen uptake rate; PES, polyethersulfone; IWW, industrial wastewater; ATU, allylthiourea; NT, total nitrogen; PT, total phosphorus; EPS, extracellular polymeric substances; SMP, soluble microbial products. * Corresponding author. Tel.: 49 1775632564. k), jan.svojitka@fhnw.ch (J. Svojitka), jiri.wanner@vscht.cz (J. Wanner), thomas. E-mail addresses: lukas.dvorak@tul.cz (L. Dvo ra wintgens@fhnw.ch, twintgens@yahoo.com (T. Wintgens). 0043-1354/$ e see front matter 2013 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.watres.2013.03.053

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

4413

sludge microorganisms (Li et al., 2004; Xing and Sun, 2009), especially the nitrifying bacteria (Wei et al., 2010). Since European legislation requires nitrogen removal from municipal wastewater before its discharge into water bodies (European Commission, 1991), nitrication (biological oxidation of ammonia nitrogen to nitrate) is a crucial process in biological wastewater treatment. Nitrication bacteria have generally slow growth rate compared to the heterotrophs and they are more sensitive to inhibition. Inhibitory effects of various compounds on nitrication contained in the industrial wastewaters have been observed, including wastewater from pharmaceutical or chemical industry (Dalzell et al., 2002; Sirtori et al., 2009). Problems with biological treatment of industrial wastewaters are often prevented by its (pre-)treatment by physicochemical methods (Ledakowicz and Gonera, 1999), or by cotreatment together with municipal wastewaters (Soupilas et al., 2008) which leads to dilution of the active substance of industrial wastewaters and fading of the concentration peaks. Occasionally, the required treatment efciency is only reached after sufciently long adaptation of activated sludge (De Wever et al., 2007; Lefebvre and Moleta, 2006; Sirtori et al., 2009). Nitrifying bacteria adapted to industrial wastewater represent a cheaper option for the removal of reduced nitrogen species from wastewaters compared to physical or chemical methods (Lay-Son and Drakides, 2008). Moreover, when the activated sludge system is coupled with an anoxic part, efcient total nitrogen removal can be achieved (Rosenberger et al., 2002; Tan et al., 2008). Membrane bioreactor (MBR) technology has recently become an established process for the treatment of municipal and industrial wastewaters. The membrane separation step represents an absolute barrier for the activated sludge microorganisms, which allows the operator to maintain sludge retention time (SRT) independent of the wastewater ow rate or settling properties of the biomass. Higher sludge retention time (above 10 days) (Metcalf and Eddy, 2003) is necessary to prevent the wash-out of nitriers and other slow-growing microorganisms from the system, and to increase the biodiversity of the microorganisms in activated sludge. Additionally, nitriers can benet from the high MLSS concentration in the MBR and resulting low F/M ratios (Liu et al., 2005; Pollice et al., 2008). The operating conditions in an MBR can thus decrease the selection pressure on the nitriers and support their growth especially under stress conditions, such as low temperature or inhibition (Liebif et al., 2001). The microorganisms also become more resistant to changes in operation conditions, e.g. different loading, occurrence of inhibitory or toxic compounds. However, the overall effect of such compounds depends on various factors, their mutual combinations or adaptation of the activated sludge (Grunditz et al., 1998; Pagga et al., 2006). Therefore, the application of MBR technology can be a promising option for the treatment of industrial wastewaters, particularly if nitrication is required (Li et al., 2006; Qin et al., 2007). The main goal of this work was focused on investigating the inuence of industrial (pharmaceutical and chemical) wastewater on the adaptation of activated sludge, particularly related to nitrication capabilities, and overall performance of

a pilot-scale membrane bioreactor (MBR). The performance and activated sludge adaptation was evaluated based on inuent and efuent parameters and on the measurement of the respiration activity of the activated sludge. Removal of organic compounds and ltration performance was evaluated as well.

2.
2.1.

Material and methods

Pilot plant

The membrane bioreactor used during the experiments consisted of an activated sludge tank (90 L) coupled with a ZeeWeed-10 ltration unit supplied by GE Water & Process Technologies, providing a total MBR volume of 120 L (ow scheme shown in Fig. 1). The entire MBR was operated under oxic conditions. The ltration unit contained a submerged ultraltration hollow bre membrane module (0.93 m2 active area; 0.04 mm nominal pore size; membrane material PVDF). The MBR was operated in cycles of ltration (300 s, ux 13.8 L m2 h1) and backwash (30 s, ux 18 L m2 h1). The permeate ow was controlled by a reversible gear pump. Mechanical and chemical cleaning of the membrane module was conducted when the TMP exceeded 40 kPa (maximum pressure recommended by the manufacturer was 50 kPa). The chemical cleaning was conducted in the ltration tank and consisted of an oxidation step (NaOCl e 1000 mg L1 active Cl; pH 10.5) and an acidic step (4 g L1 citric acid, pH 3.5). The inow into the activated sludge tank was facilitated by two peristaltic pumps (Ismatec VC-381), which were continuously feeding the municipal and industrial wastewater in the desired ratio. The total inow rate was between 9 and 10 L h1. Recirculation between the activated sludge tank and the ltration tank was provided by a peristaltic pump (ratio between recirculation and permeate ow was approx. 5). Composite samples of mixed inuent and permeate were collected by an automated sampler and kept frozen until analysis. The pilot plant was placed in the pilot hall of the wastewater treatment plant of Basel (Switzerland); its total operation time within this study was 525 days. Since the pilot plant was kept indoors, the temperature uctuation was small and ranged from 20  C to 25  C throughout operation. The concentration of dissolved oxygen in the bioreactor was veried in about weekly intervals by a portable oxygen sensor; all

Fig. 1 e Scheme of the pilot plant.

4414

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

readings were between 3 and 5 mg L1. The pH value in the reactor was measured on-line by the built-in sensor of the pilot plant; its value was oscillating between 6.9 and 8.1 with an average value of 7.4. Sufcient amount of phosphorus in the mixed liquor was ensured by the municipal wastewater in the inuent mixture through most of the experimental period. In order to achieve the desired SRT the excess sludge was discharged regularly (three times per week) and the resulting SRT was calculated as weekly average. Signicant foaming of the bioreactor occurred several times during the pilot plant operation; however, it was always restricted to a few days and never resulted in massive loss of biomass. The basic technological parameters of the MBR during different stages of the experiment are shown in Table 1.

inuent and efuent as well as measurement of its respiration activity. At regular intervals the inuence of the industrial wastewater was compared to a non-adapted population taken from the municipal nitrifying wastewater treatment plant (AUR tests with referential activated sludge).

2.4.

Analytical methods

2.2.

Wastewater

The concentration of chemical oxygen demand (COD), ammonia nitrogen (NH4-N), nitrates (NO3-N) and nitrites (NO2N) in the inuent as well as in the efuent was regularly measured. Analyses of COD were performed using LCK-514 cuvette tests from Hach-Lange; nitrates according to ISO 7890-1:1986 (ISO Standards) and other nitrogen species according to the Standard methods (APHA, 2005). MLSS measurement was carried out gravimetrically.

The inow into the MBR was a mixture of two streams:  Municipal wastewater from the Basel area after mechanical pretreatment (screen, sand trap and primary settling).  Wastewater from industrial (chemical and pharmaceutical) companies in Basel after physico-chemical pretreatment (neutralization, coagulation, occulation, otation). Both wastewater streams were dosed from ow-through pipes with continuous supply of fresh wastewater. The average composition of the two streams is summarized in Table 2.

2.5.

Kinetic nitrication rate (AUR) tests

2.3.

Activated sludge adaptation

Activated sludge from a nitrifying municipal wastewater treatment plant was taken as an inoculum for the MBR. This activated sludge was not exposed to any industrial wastewater before its withdrawal and exhibited a high nitrication rate. In later stages of the experiment, samples of activated sludge from the same wastewater treatment plant were used as referential biomass for the AUR tests. The loading of the MBR with industrial wastewater was increased in steps, so that its proportion of the mixed wastewater was 0, 5, 10, 20, 30, 40, 50, 60 and 75% (V/V) respectively. The length of each step was originally around 1 month (for the content 5e40%). Then, based on results of efuent parameters and kinetic tests, the steps were prolonged up to several months. Adaptation of activated sludge to industrial wastewater was evaluated based on the chemical analysis of

The nitrication rate was measured respirometrically from oxygen consumption. Nitrication batch tests were carried out in a respirometric cell (0.5 L) under laboratory temperature (20e24  C) and continuous mixing. At the beginning of a test, industrial or municipal wastewater was mixed with the activated sludge sample from the MBR in a 1:1 volumetric ratio and placed into the respirometric cell together with 50 mg L1 NH4-N. The oxygen concentration was automatically recorded and kept between 2.5 mg L1 and 5.0 mg L1 by the control unit of the respirometer. The overall volumetric respiration rate (OUR) of the activated sludge sample was calculated from the decreasing concentration of oxygen in the respirometric cell between aeration phases. In order to distinguish the OUR attributed to nitrication bacteria from the heterotrophic respiration, 5.0 mg L1 of allythiourea (selective inhibitor of nitrication activity) was added into the respirometric cell during the test. The immediate drop of respiration activity was calculated from the trend-lines of OUR before and after the addition of ATU. This methodology was adapted from Ginestet et al. (1998). For recalculation of OUR to AUR the factor of 4.57, which represents the theoretical consumption of oxygen for oxidation of ammonia nitrogen to nitrate nitrogen, was used. One activated sludge sample collected weekly (in longer intervals at later stages of the experiment) was always subjected to two parallel AUR tests: one with the addition of municipal wastewater, the other with industrial wastewater.

Table 1 e Technological parameters of the MBR (mean value). Parameter 0%

Period length [d] MLSS [g L1] SRT [d] HRT [h] F/M ratio [g COD$g1 MLSS$d1] F/M ratio [mg NH4-N$g1 MLSS$d1] 60 6.2 e 14 0.09 11.3

Share of industrial wastewater in the MBR inuent 5%

26 7.2 51 14 0.07 7.35

10%
28 9.1 38 14 0.13 5.75

20%
28 8.5 38 14 0.12 6.44

30%
35 9.2 34 14 0.13 5.30

40%
35 8.9 38 14 0.21 7.07

50%
194 10.3 33 14 0.18 7.10

60%
80 9.0 30 14 0.22 8.83

75%
39 7.8 30 14 0.23 10.2

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

4415

Table 2 e Average composition of the municipal and industrial wastewater (mg LL1) including standard deviation; minimum and maximum are shown in parentheses. Wastewater/parameter
Municipal wastewater Industrial wastewater

COD
363 172 (150; 1240) 1898 532 (680; 3000)

NT
38.8 9.4 (13.7; 72.0) 66.8 23.1 (30.7; 140)

NH4-N
24.0 5.6 (5.6; 38.2) 29.7 13.3 (9.5; 86.2)

PT
6.7 3.5 (2.6; 26.7) 0.66 0.72 (0.13; 5.1)

The difference between the two AURs could be attributed to the inhibition effect of the industrial wastewater. The same methodology was applied on samples of activated sludge from the MBR as well as for referential activated sludge.

2.6.

Filtration tests

Filterability tests of activated sludge from the MBR were carried out regularly (once per 3 weeks) in the ltration cell (Amicon 8400, 400 mL, membrane area 42 cm2) under laboratory temperature, continuous mixing of 250 min1 and TMP of 1 bar. Three different membranes produced by MicrodynNadir were used for the ltration tests. Their basic specications are shown in Table 3.

load was observed to around 0.21 g COD$g1 MLSS$d1, which caused a rapid increase in suspended solids in the reactor. The second drop of the inuent load started around day 440 (inuent COD 1540 mg L1) when the decline from 0.266 g COD$g1 MLSS$d1 to 0.110 g COD$g1 MLSS$d1 (inuent COD 760 mg L1; day 452) was observed with a similar effect on the sludge concentration in the reactor as during the rst drop in loading. However, in this case the loading remained low until the end of the experiment. The large variations of the composition of industrial wastewater resulted from changes in production in the connected industrial companies. Besides COD variations, changes in colour, odour and foaming propensity were observed. The latter drop was also accompanied by low biodegradability of the inuent substrate; removal efciency of COD was on average 62% in this stage in comparison with 83% reached during the rst drop (see Fig. 6).

3.
3.1.

Results and discussion

Sludge concentration 3.2. Nitrogen removal

The MBR was inoculated with a relatively low sludge concentration (2.5 g L1). In order to achieve the target concentration of 8e11 g L1 no or little excess sludge was discharged in the initial stages of the operation (see SRT values in Table 1). After 100 days of operation, the MLSS concentration reached 10 g L1 and the SRT was shortened to around 35 days. This SRT was then kept almost constant for the rest of the operation; the MLSS concentration in the bioreactor (Fig. 2) was thus varying due to changes in inuent load. A sharp decrease of suspended solids was observed around day 400 and later around day 470 of the experiment due to the drop of inuent load. The rst drop of inuent load was from 0.146 g COD$g1 MLSS$d1 (inuent COD 1120 mg L1; day 389) to 0.053 g COD$g1 MLSS$d1 (inuent COD 220 mg L1; day 399), with an average inuent load in this period (60% industrial wastewater) of 0.120 g COD$g1 MLSS$d1. The low loading resulted in low sludge production, which did not match the amount of suspended solids removed as excess sludge (aiming at a constant SRT value). After that, a stable increase in inuent

The evolution of nitrogen species concentration throughout the MBR operation is displayed in Fig. 3. Shortly after the startup of the MBR, an elevated concentration of nitrites was detected in the efuent due to the transition state of the biomass after its transfer to the new conditions. After equilibration a period of stable oxidation of ammonium to nitrate followed. During that period the increase in the content of industrial wastewater in the MBR inuent had almost no inuence on the concentration of nitrogen species in the efuent from the MBR. The average efuent NH4-N concentration in this interval (0e30% industrial wastewater in the inuent) was <0.5 mg L1. Very good water quality in terms of

Table 3 e Basic specications of membranes used for lterability test. Type

PM MV020 (microltration) PM MP005 (microltration) PM UV150 (ultraltration)

Material
PVDF PES PVDF

MWCO
0.2 mm 0.05 mm 150 kDa

Water ux at 0.7 bar (L m2 h1)

>500 >200 >200

Fig. 2 e Mixed liquor suspended solids in the MBR.

4416

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

Fig. 3 e Nitrogen species in the efuent depending on the proportion of industrial wastewater in the MBR inuent; grey area represents the proportion of industrial wastewater in the inuent into the MBR.

Fig. 5 e Referential sludge e nitrication rates (AUR) measured with and without addition of industrial wastewater during the respirometric test.

ammonia nitrogen concentration (on average 1.01 mg L1, removal efciency 97.9%) was also achieved by Radjenovic et al. (2007) in a laboratory-scale MBR treating a mixture of municipal, hospital and industrial wastewater with prevailing municipal load. Changes in the distribution of N species in the efuent appeared when the content of industrial wastewater reached 40%. Up to 20 mg NO2-N L1 was detected in the efuent at this stage; however the ammonium concentration remained low

(<3 mg L1). The oxidation of nitrite to nitrate proceeded only with approx. 50% efciency. This observation indicated the inhibition of nitrite oxidation (second step of nitrication). The decrease of nitrication ability at this point cannot be explained by overloading of the sludge, since the nitrogen load (F/M) was comparable to that during the interval with 20% industrial wastewater and organic loading was only slightly higher (0.16 g COD$g MLSS1$d1 vs. 0.14 g COD g1$MLSS1$d1) and still in the range suitable for nitrifying systems. It is thus likely that nitrication was inhibited by the

Fig. 4 e MBR sludge e nitrication rates (AUR) measured with and without addition of industrial wastewater during the respirometric test.

Fig. 6 e COD concentration in the efuent and COD removal efciency depending on proportion of industrial wastewater in the MBR inuent. Grey area represents the proportion of industrial wastewater in the inuent into the MBR.

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

4417

industrial wastewater. Strong inhibition by undiluted pharmaceutical efuent towards nitrication was also observed by Dalzell et al. (2002) in different rapid toxicity bioassays. The ndings are moreover in agreement with numerous references that the inhibition starts with the second step of nitrication, which is usually described as more sensitive to different factors compared to the rst step (Ciudad et al., 2006; Dinc er and Kargi, 1999; Kim et al., 2011; Mota et al., 2005). With further increase of the content of industrial wastewater to 50%, ammonium became the dominant nitrogenous species in the MBR efuent. The additional load of NH4-N in this period increased by approx. 10% compared to the previous period with 40% industrial wastewater and does not correspond to the 20-fold increase of ammonium concentration in the efuent. This increase can be attributed to a higher dose of industrial wastewater whose concentration became sufcient to inhibit both steps of nitrication. The oxidation of ammonium recovered after 25 days (NO2-N in efuent 15e25 mg L1, NO3-N 1e5 mg L1), however, the nitrate production was only restored after another 120 days. Most of the phase with 50% industrial load was characterized by highly varying concentrations indicating an unstable nitrication process. Nevertheless, at the end of this phase nitrifying organisms were able to adapt to the wastewater composition and ammonium was almost completely converted into nitrate again. However, with a 60% share of industrial wastewater nitrication collapsed and ammonium prevailed in the efuent. With 75% the efuent concentration of NH4-N was even higher than in the inuent due to the hydrolysis of organic nitrogen compounds from the inuent, which could not be assimilated in new biomass anymore. The respirometric tests were carried out in regular intervals to monitor the nitrication activity. The nitrication rate of each activated sludge sample taken from the MBR was measured both with (black labels in Fig. 4) and without (grey) the addition of industrial wastewater. The difference between the two rates can be attributed to the inhibition by the industrial wastewater. The nitrication rates (AUR) in the diagram show the following trends: (1) Decreasing maximum nitrication rate (i.e. without industrial wastewater) with a local minimum at 50% industrial wastewater in the inow (day 235) and (2) Increasing nitrication rate with time in the presence of industrial wastewater. These observations indicate that the inhibition effect gradually decreased during the adaptation process. Before the start-up of the adaptation process, the inhibition effect of industrial wastewater was 95% (related to AUR with municipal wastewater only) while after 213 days of adaptation and stepped increase of industrial wastewater in the MBR inuent it was only 23%. After subsequent increase of the proportion of industrial wastewater in the inuent to 50%, the inhibition effect was not observed anymore; in some tests the AURs were even higher in the test with 100% of industrial wastewater than in the test with 0% of industrial wastewater. The overall nitrication activity was decreasing with higher industrial wastewater load and reached its minimum

after the share of industrial wastewater in the MBR inuent was increased to 50%. The nitrication rate recovered while the inuent ratio was kept constant at 50%. However, further increase to 60% caused a breakdown of nitrication and no nitrication activity was measurable by the respirometric tests anymore. These observations correspond to the analyses of the nitrogen species in MBR inuent and efuent from the MBR discussed before. The AUR tests with referential activated sludge were carried out in similar intervals as with the samples from the MBR pilot plant (Fig. 5). In contrast to the MBR sludge, the inhibition effect to the samples of referential activated sludge did not change substantially during the whole experimental period and ranged from 70 to 92%. The result of both AUR tests and efuent analyses conrmed that the increasing dose of industrial wastewater leads to inhibition of nitrication activity of the biomass. The maximum nitrication rate of the sludge was reduced and was no more measurable with more than 50% industrial wastewater in the inuent. On the other hand, unless a certain threshold is exceeded, the nitriers could become more resistant to high loadings with the industrial wastewater, which can be attributed to their adaptation to the tested wastewater.

3.3.

Chemical oxygen demand

The MBR performance regarding COD removal is illustrated in Fig. 6. Since the industrial wastewater is generally characterized by higher COD content and higher variations compared to the municipal wastewater, these were gradually becoming more pronounced in the mixed inuent as well. The average COD concentration in the inuent rose from 370 mg L1 (municipal wastewater only) up to 1100 mg L1 (60% industrial wastewater). However, the impact on F/M ratio was less dramatic, since the suspended solids concentration in the MBR increased as well (see Fig. 2). The presence of industrial wastewater also led to the increase of COD concentration in the MBR efuent due to both higher COD load and decreasing removal efciency. The removal dropped substantially (below 50%) when the content of industrial wastewater reached 75% and the efuent concentration of COD rose up to 620 mg L1. Non-ideal phosphorus supply during the last 14 days of the pilot plant operation (day 511e525) and resulting limitations of the biological processes cannot be excluded. At this point the concentration of total phosphorus in the inuent decreased to 1.7 mg L1. Considering the volume of excess sludge extracted and a typical phosphorus content in the biomass of 0.2 mg P g1 VSS (Ramdani et al., 2012), approximately 1.9 mg P L1 would be needed to cover the losses of phosphorus with excess sludge. In the rest of the interval with 75% industrial wastewater the phosphorus content in the inuent exceeded the needs for biomass synthesis by 30e50%. The MBR operation was only marginally inuenced by the presence of industrial wastewater when its content was kept below 40%: the efuent COD was below 100 mg L1 (corresponding to >80% removal) and with no signicant variations. Remarkably low COD removal efciency was achieved at the beginning of MBR

4418

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

operation, probably as a consequence of non-stable conditions following the seeding of the MBR. Comparable studies dealing with the treatment of industrial wastewaters in MBRs usually report higher COD removal efciencies than observed in this study during periods with 50e75% of industrial wastewater in the inuent. Chang et al. (2008) operated a pilot-scale MBR for treatment of wastewater consisting of pharmaceutical wastewater and septic tank efuent for a period of 140 days. These wastewaters were characterized by largely uctuating ow and composition; the average removal of COD exceeded 95%. High COD removal efciencies in membrane bioreactors treating different types of industrial wastewaters were also observed by other authors. For instance, the reported COD removal efciency reached 93e98% in the case of oil contaminated wastewater (Scholz and Fuchs, 2000); approx. 90% with pharmaceutical wastewater (Radjenovic et al., 2007), 97% with wastewaters similar to the streams from wineries or 86% from a tannery factory (Artigaa et al., 2005). But the actual degradability is highly dependent on the wastewater composition and current results indicate that parts of efuents received by the pilot plant operated in the present study were poorly bio-available. Biodegradability of the bulk organic content of the wastewater can potentially be further increased by long biomass adaptation or pretreatment implementation, e.g. using the photoFenton method (Sirtori et al., 2009). These authors reached a more than 30% higher degradability when photo-Fenton treatment was applied for pretreatment of industrial pharmaceutical wastewater.

Fig. 8 e Permeability during laboratory ltration tests with samples from different stages of the experiment; each sample was ltered through three different membranes e data taken after the rst 10 min of testing.

3.4. Filtration performance in the MBR and in lab-scale ltration tests

Permeability was calculated from automatic measurement of TMP and ow in the pilot plant. Gradual deterioration of the ltration performance was observed with increasing share of the industrial wastewater in the inuent (Fig. 9). The fouling

rate was continuously increasing and chemical cleaning of the membrane was necessary in approx. 30 day intervals when the industrial wastewater share reached 60 and 75%, respectively (Fig. 7). The ltration tests of the activated sludge taken from the MBR were carried out regularly to assess the ltration properties under lab conditions. The aim of ltration tests was to determine whether the changing proportion of industrial wastewater affects the mixed liquor lterability and thus membrane permeability. The lterability of activated sludge is inuenced by the character of suspended solids in the mixed liquor, e.g. oc size or presence of extracellular polymeric

Fig. 7 e Permeability measured in the MBR during the experiment. The points of chemical cleaning can be recognized from the permeability rise.

Fig. 9 e Decrease of permeability during laboratory ltration tests with two different sludge samples: 20% IWW in the inuent (thick lines) and 40% IWW (thin lines).

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

4419

substances (EPS) and soluble microbial products (SMP). Several studies proposed that EPS and SMP are the key factors responsible for fouling (Drews, 2010; Meng et al., 2009). Bai and Leow (2002) concluded that the oc size can signicantly affect the permeability, especially if oc diameter becomes comparable with the membrane pore size. However, neither EPS and SMP concentration nor oc size were measured because they were not fundamental in this study. Since the oc size and EPS/SMP production are inuenced by the presence of toxins (Henriques and Love, 2007) or F/M ratio (Wu and Lee, 2011), it was expected that the increasing share of industrial wastewater in the MBR inuent can alter the activated sludge characteristics signicantly. The results of ltration tests are summarized in Fig. 8, which compares the permeability reached with each activated sludge sample and membrane type in the rst 10 min of ltration. The permeabilities are decreasing with higher content of industrial wastewater, although this trend is only apparent starting from a 20% share. This observation corresponds to the permeability decrease in the MBR pilot plant. The permeability seems to be independent of the suspended solids in the bioreactor, since the concentration of 9e12 g L1 SS was reached in the 10% period and maintained until the end of operation. In fact, the sludge lterability in this interval was very similar to that without industrial wastewater (after a temporal deterioration during the period with 5% industrial wastewater). The lowest permeability (for all membranes) was observed for the sample taken during the highest tested share of industrial wastewater e 50%. Filtration tests with samples from periods with 60 and 75% industrial wastewater were not conducted. The negative effect of increasing industrial wastewater content in the MBR inuent on sludge lterability was probably caused by changes in activated sludge structure, which were observed by microscopic analyses and included disintegration of activated sludge ocs and common presence of non-settling fragments and zoogloea (data not shown). The time course of permeability during the test indicates the fouling of the tested membrane by the particular sludge sample. Examples representing two observed permeability patterns are shown in Fig. 9: with samples from the period with 20% and 40% industrial wastewater in the MBR inuent, respectively. Differences in the permeability observed in the period with 20% and 40% industrial wastewater can be explained by changes in activated sludge morphology. Various factors including changes in toxic compound concentration or ionic strength were shown to cause de-occulation of activated sludge (Morgan-Sagastume and Allen, 2004); increasing content of industrial wastewater in the reactor can thus decrease the activated sludge oc size. Very small ocs were observed in an activated sludge reactor treating dye wastewaters by Sponza (2002). According to Meng et al. (2006) and Bai and Leow (2002), the ltration layer formed particularly by colloids, nonsettling fragments having diameter <50 mm and small activated sludge particles is very compact resulting in low permeability. In other words, increasing oc size of activated sludge causes also its higher porosity and the ltration layer is then more permeable (Clauss et al., 1998). In general faster fouling was observed for membranes with more open pores when ltrating samples from periods with 0e20% industrial wastewater in the inuent. This was probably caused by high

initial ux which caused fast transport of fouling material towards the membrane surface. In later stages when the low lterability of the activated sludge led to small uxes the performance of the three different membrane sheets was almost equal, indicating a cake layer controlled mass transfer.

4.

Conclusion

A pilot-scale MBR was operated for 525 days under increasing loading by wastewater from chemical and pharmaceutical industry. The system exhibited stable and efcient performance with up to 30% industrial wastewater in the inuent. Higher loading led to deterioration of both COD and nitrogen removal. Inhibition of nitrite oxidation and subsequently ammonium oxidation was observed when the share of industrial wastewater in the inuent reached 40 and 50%, respectively. However, nitrication ability was recovered after several weeks (40%) or months (50%) of lag-phase. No nitrication activity was measured when the share of industrial wastewater in the MBR inuent exceeded 50%. The results of the respirometric tests corresponded to the trends of nitrogen species observed in the MBR efuent. Comparison of nitrication rate with and without industrial wastewater showed both inhibition (reduced maximum nitrication rate) and adaptation effects (ability to nitrify in the presence of the industrial wastewater) in the activated sludge exposed to the industrial wastewater. Filtration tests showed lower lterability of activated sludge with higher content of industrial wastewater. Higher share of industrial wastewater also accelerated the fouling of the membrane in the pilot plant. In future studies the possibility to lower the content of potentially nitrication inhibiting compounds through predenitrication could be tested.

Acknowledgement
Sciex-NMSch (Scientic Exchange Programme between Switzerland and the New Member States of the EU) project code CZ 0908002 is gratefully acknowledged. ProRheno AG is kindly acknowledged for kind cooperation and support. The paper was supported in part by the Project Development of Research Teams of R&D Projects at the Technical University of Liberec CZ.1.07/2.3.00/30.0024.

references

APHAEaton, A.D., Clesceri, L.S., Rise, E.W., Greenberg, A.E., 2005. Standard Methods for the Examination of Water and Wastewater, twenty-rst ed. APHA, Washington. ndeza, R., Artigaa, P., Ficarab, E., Malpeib, F., Garridoa, J.M., Me 2005. Treatment of two industrial wastewaters in a submerged membrane bioreactor. Desalination 179 (1e3), 161e169. Bai, R., Leow, H.F., 2002. Microltration of activated sludge wastewater e the effect of system operation parameter. Separation and Purication Technology 29 (2), 189e198.

4420

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

Chang, Ch.-Y., Chang, J.-S., Vigneswaran, S., Kandasamy, J., 2008. Pharmaceutical wastewater treatment by membrane bioreactor process e a case study in southern Taiwan. Desalination 234 (1e3), 393e401. Chelliapan, S., Wilby, T., Yuzir, A., Sallis, P.J., 2011. Inuence of organic loading on the performance and microbial community structure of an anaerobic stage reactor treating pharmaceutical wastewater. Desalination 271 (1e3), 257e264. Ciudad, G., Werner, A., Bornhardt, C., Mun oz, C., Antileo, C., 2006. Differential kinetics of ammonia- and nitrite-oxidizing bacteria: a simple kinetic study based on oxygen afnity and proton release during nitrication. Process Biochemistry 41 (8), 1764e1772. Clauss, F., Helaine, D., Balavoine, C., Bidault, A., 1998. Improving activated sludge oc structure and aggregation for enhanced settling and thickening performance. Water Science and Technology 38 (8e9), 35e44. Dalzell, D.J.B., Alte, S., Aspichueta, E., Sota, A., Etxebarria, J., Gutierrez, M., Hoffmann, C.C., Sales, D., Obst, U., Christo, N., 2002. A comparison of ve rapid direct toxicity assessment methods to determine toxicity of pollutants to activated sludge. Chemosphere 47 (5), 535e545. De Wever, H., Weiss, S., Reemstsma, T., Vereecken, J., Mu ller, J., Knepper, T., Ro rden, O., Gonzalez, S., Barcelo, D., Hernando, M.D., 2007. Comparison of sulfonated and other micropollutants removal in membrane bioreactor and conventional wastewater treatment. Water Research 41 (4), 935e945. Dinc er, A.R., Kargi, F., 1999. Salt inhibition of nitrication and denitrication in saline wastewater. Environmental Technology 20 (11), 1147e1153. Drews, A., 2010. Membrane fouling in membrane bioreactors e characterisation, contradictions, cause and cures. Journal of Membrane Science 363 (1e2), 1e28. European Commission, 1991. Council Directive 91/271/EEC of 21 May 1991 Concerning Urban Waste-water Treatment. Ginestet, P., Audic, J.M., Urbain, V., Block, J.C., 1998. Estimation of nitrifying bacterial activities by measuring oxygen uptake in the presence of the metabolic inhibitors allylthiourea and azide. Applied and Environmental Microbiology 64 (6), 2266e2268. Grunditz, C., Gumaelius, L., Dalhammer, G., 1998. Comparison of inhibition assays using nitrogen removing bacteria: application to industrial wastewater. Water Research 32 (10), 2995e3000. Henriques, I.D.S., Love, N.G., 2007. The role of extracellular polymeric substances in the toxicity response of activated sludge bacteria to chemical toxins. Water Research 41 (18), 4177e4185. ISO 7890-1:1986. Water Quality e Determination of Nitrate e Part 1: 2,6-Dimethylphenol Spectrometric Method. Kim, Y.M., Cho, H.U., Lee, D.S., Park, D., Park, J.M., 2011. Comparative study of free cyanide inhibition on nitrication and denitrication in batch and continuous ow systems. Desalination 279 (1e3), 439e444. Lay-Son, M., Drakides, Ch, 2008. New approach to optimize operational conditions for the biological treatment of a highstrength thiocyanate and ammonium waste: pH as key factor. Water Research 42 (3), 774e780. Ledakowicz, S., Gonera, M., 1999. Optimisation of oxidants dose for combined chemical and biological treatment of textile wastewater. Water Research 33 (11), 2511e2516. Lefebvre, O., Moleta, R., 2006. Treatment of organic pollution in industrial saline wastewater: a literature review. Water Research 40 (20), 3671e3682. Li, S., Li, X., Wang, D., 2004. Membrane (RO-UF) ltration for antibiotic wastewater treatment and recovery of antibiotics. Separation and Purication Technology 34 (1e3), 109e114.

Li, H., Yang, M., Zhang, Y., Yu, T., Kamagata, Y., 2006. Nitrication performance and microbial community dynamics in a submerged membrane bioreactor with complete sludge retention. Journal of Biotechnology 123 (1), 60e70. Liebif, T., Wagner, M., Bjerrum, L., Denecke, M., 2001. Nitrication performance and nitrier community composition of a chemostat and a membrane-assisted bioreactor for the nitrication of sludge reject water. Bioprocess and Biosystems Engineering 24, 203e210. Liu, R., Huang, X., Xi, J.Y., Qian, Y., 2005. Microbial behaviour in a membrane bioreactor with complete sludge retention. Process Biochemistry 40 (10), 3165e3170. Meng, F., Zhang, H., Yang, F., Zhang, S., Li, Y., Zhang, X., 2006. Identication of activated sludge properties affecting membrane fouling in submerged membrane bioreactors. Separation and Purication Technology 51 (1), 95e103. Meng, F., Chae, S.-R., Drews, A., Kraume, M., Shin, H.-S., Yang, F., 2009. Recent advances in membrane bioreactors (MBRs): membrane fouling and membrane material. Water Research 43 (6), 1489e1512. Metcalf and Eddy, Inc., 2003. Wastewater Engineering: Treatment and Reuse. McGraw-Hill, New York. Morgan-Sagastume, F., Allen, D.G., 2004. Impact of stresses or transient conditions on deocculation in engineered systems. In: Droppo, I.G., Leppard, G.G., Liss, S.N., Milligan, T.G. (Eds.), Flocculation in Natural and Engineered Environmental Systems. CRC Press, Boca Raton, FL. Mota, C., Head, M.A., Ridenoure, J.A., Cheng, J.J., Reyes, F.L., 2005. Effects of aeration cycles on nitrifying bacterial populations and nitrogen removal in intermittently aerated reactors. Applied and Environmental Microbiology 71 (12), 8565e8572. Noble, J., 2006. GE ZeeWeed MBR technology for pharmaceutical wastewater treatment. Membrane Technology 2006 (9), 7e9. Pagga, U., Bachner, J., Stotmann, U., 2006. Inhibition of nitrication in laboratory tests and model wastewater treatment plants. Chemosphere 65 (1), 1e8. Pollice, A., Laera, G., Saturno, D., Giordano, C., 2008. Effects of sludge retention time on the performance of a membrane bioreactor treating municipal sewage. Journal of Membrane Science 317 (1e2), 65e70. Qin, J.J., Wai, M.N., Tao, G., Kekre, K.A., Seah, H., 2007. Membrane bioreactor study for reclamation of mixed sewage mostly from industrial sources. Separation and Purication Technology 53 (3), 296e300. , D., 2007. Analysis of Radjenovic, J., Petrovic, M., Barcelo pharmaceuticals in wastewater and removal using a membrane bioreactor. Analytical Bioanalytical Chemistry 387 (4), 1365e1377. le ris, S., Houweling, D., Ramdani, A., Dold, P., Gadbois, A., De Comeau, Y., 2012. Characterization of the heterotrophic biomass and the endogenous residue of activated sludge. Water Research 46 (3), 653e668. Rosenberger, S., Kruger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36 (2), 413e420. Scholz, W., Fuchs, W., 2000. Treatment of oil contaminated wastewater in a membrane bioreactor. Water Research 34 (14), 3621e3629. Sirtori, C., Zapata, A., Oller, I., Gernjak, W., Agu era, A., Malato, S., 2009. Decontamination industrial pharmaceutical wastewater by combining solar photo-Fenton and biological treatment. Water Research 43 (3), 661e668. Soupilas, A., Papadimitriou, C.A., Samaras, P., Gudulas, K., Petridis, D., 2008. Monitoring of industrial efuent ecotoxicity in the greater Thessaloniki area. Desalination 224 (1e3), 261e270.

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

4421

Sponza, D.T., 2002. Extracellular polymer substances and physicochemical properties of ocs in steady and unsteadystate activated sludge systems. Process Biochemistry 37 (9), 983e998. Tan, T.W., Ng, H.Y., Ong, S.L., 2008. Effect of mean cell residence time on the performance and microbial diversity of predenitrication submerged membrane bioreactors. Chemosphere 70 (3), 387e396. Wei, X., Wang, Z., Fan, F., Wang, J., Wang, S., 2010. Advanced treatment of a complex pharmaceutical wastewater by

nanoltration: membrane foulant identication and cleaning. Desalination 251 (1e3), 167e175. Wu, S.C., Lee, C.M., 2011. Correlation between fouling propensity of soluble extracellular polymeric substances and sludge metabolic activity altered by different starvation conditions. Bioresource Technology 102 (9), 5375e5380. Xing, Z.P., Sun, D.Z., 2009. Treatment of antibiotic fermentation wastewater by combined polyferric sulfate coagulation, Fenton and sedimentation process. Journal of Hazardous Materials 168 (2e3), 1264e1268.