Performance and membrane fouling in an integrated membrane coagulation reactor (IMCR) treating textile wastewater

Chemical Engineering Journal 240 (2014) 82–90

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Performance and membrane fouling in an integrated membrane coagulation reactor (IMCR) treating textile wastewater Jin Li a,⇑, Dan Wang b, Deshuang Yu a, Peiyu Zhang a, Yue Li a a b

School of Chemical and Environmental Engineering, Qingdao University, Qingdao, PR China National Marine Environmental Forecasting Center, State Oceanic Administration, Beijing 100081, PR China

h i g h l i g h t s  The IMCR had good performances in color removal when it was used to treat textile wastewater.  Membrane fouling experienced three phases and a higher PACl dose aggravated the membrane fouling.  The fractal dimension of floc formed with PACl 1 mmol/L was 1.88 while it reached 2.22 with PACl 1.4 mmol/L.  Independent of the membrane MWCO, lower molecular weight (MW) fractions resulted in the irreversible membrane fouling.

a r t i c l e

i n f o

Article history: Received 5 August 2013 Received in revised form 27 October 2013 Accepted 22 November 2013 Available online 1 December 2013 Keywords: Integrated membrane coagulation reactor (IMCR) Textile wastewater Membrane fouling Fractal dimension Particle size

a b s t r a c t An integrated membrane coagulation reactor (IMCR) was employed to treat textile wastewater. Color could be removed almost completely and chemical oxygen demand (COD) removal efficiency could reach 88% with polyaluminum chloride (PACl) dose 1.2 mmol/L, influent pH 5.5 and hydraulic retention time (HRT) 3 h. Membrane fouling experienced three phases and a higher PACl dose aggravated the membrane fouling. The contact angle of new membrane was 52.5 degrees while the fouled ones’ were around 81 degrees. The fractal dimension of floc formed with PACl 1 mmol/L was 1.88 while it reached 2.22 with PACl 1.4 mmol/L. A more permeable filter cake was formed with PACl 1 mmol/L and could alleviate the membrane fouling effectively. Both the trans-membrane pressure (TMP) and TMP increment were greater when the IMCR was equipped with larger molecular weight cutoff (MWCO) membrane. The physically reversible and irreversible filtration resistances (Rf) of the IMCR with membrane MWCO 50, 100 and 150 kDa were 3.2, 9.1 and 31.4 times of there intrinsic membrane filtration resistances, respectively. Membrane was fouled more seriously when the higher MWCO membrane was used during the IMCR process. Independent of the membrane MWCO, lower molecular weight (MW) fractions resulted in the irreversible membrane fouling. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Textile industry is considered as one of the largest water consumers in the world and textile wastewater is one of the most hazardous wastewater for the environment when discharged without proper treatment [1]. Various pollutants such as dyes, degradable organics, surfactants, heavy metals and pH adjusting chemicals could be found in textile wastewater [2]. Furthermore, the composition of wastewater from dyeing and textile processes varied greatly from day to day and even from hour to hour. Dyes were the main pollutants and were considered hazardous to the environment because they affected the nature of the water and reduced the photosynthetic action [3].

⇑ Corresponding author. Tel.: +86 53285955529. E-mail address: [email protected] (J. Li). 1385-8947/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2013.11.058

Efforts had being made to study dye-removal or dye-degradation processes and to develop analytical methodologies for evaluating these processes. In general, chemical or biological treatment was applied for textile wastewater treatment. Coagulation and sedimentation processes were known to be effective in removing colors of insoluble dyes such as disperse ones but they were not the cases for soluble dyes including reactive dyes [4]. The well-known conventional coagulants such as alum, polyaluminum chloride (PACl), FeSO4 and lime were widely used in the textile wastewater treatment. More than 90% of color removal from acid dyes could be achieved by adding activated carbon whose effect was, however, known to be insignificant for base and direct dyes [5]. Biological treatment processes were frequently used to treat textile effluents. These processes were generally efficient for biochemical oxygen demand (BOD5) and suspended solids (SS) removal; however, they were largely ineffective for removing color which was visible even at low concentrations [6]. Although dyes

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

in wastewater could be effectively destroyed by advanced chemical oxidation [7], the treatment cost was high. Besides, the incomplete degradation products, for example the benzidine and some carcinogenic aromatic compounds, had great toxicity. Such as phenols inhibited the growth of the aquatic plants and various organisms, benzene has significant toxic effects on human nervous and vascular systems [8,9]. Based on the various hazards, the improvement of the wastewater treatment technology would play an extremely important role in maintaining ecological balance, protecting the environment and human health. Membrane processes could be successfully used not only for producing purified water but also for recycling of specific contaminants in industrial effluents by an efficient separation [10]. However, constituents in feedwater could foul the membranes [11]. Early studies reported that coagulation had been the most successful pretreatment for fouling reduction and made flux improved [12–14]. Taking into account of a substantial reduction of energy depletion, immersed low-pressure hollow fiber membrane processes had gained an unprecedented popularity not only in wastewater treatment but also in drinking water production [15,16]. Under such circumstances, a novel integrated membrane coagulation reactor (IMCR), a combination of coagulation and ultrafiltration (UF), was used in textile wastewater purification. It could achieve easy automation with a low operating pressure and enable excellent liquid/solid separation. In this work, the performance of the IMCR treating textile wastewater was assessed. Besides, the characteristics of membrane fouling with different operating conditions were explored. 2. Materials and methods 2.1. Wastewater characteristics The wastewater used in the experiments was collected from a textile mill in Shandong, China. The wastewater quality was quite complicated because it consisted of various ingredients such as dyes, surfactants, auxiliary chemicals, anti-crease agents and pH adjusting chemicals. Its maximum absorbance wavelength (kmax) was 666 nm. The characteristics of the textile wastewater were presented in Table 1. 2.2. Experimental set-up and procedure The flow chart of the integrated membrane coagulation reactor (IMCR) used in this study was shown in Fig. 1. The reactor included two units: coagulation unit and membrane separation unit. The volumes of the two units were 8 L and 29 L, respectively. The membrane separation unit was equipped with two hollow-fiber ultrafiltration (UF) membrane modules (provided by Korean KMS Company) which were made of polyethylene. The total surface area was 0.97 m2 and the molecular weight cutoff (MWCO) was 80 kDa. Polyaluminum chloride (PACl) was used as coagulant and NaOH or HCl was used for pH adjustment. In the IMCR, aeration and stirring

Table 1 Characteristics of the textile wastewater.

COD BOD5 TSS TKN TP pH Absorbance at kmax Conductivity

Unit

Range

Average

mg/L mg/L mg/L mg/L mg/L – – mS/cm

367–435 43–61 109–147 11–25 3.0–3.8 9.1–10.6 2.7–2.9 61–88

413 51 125 17 3.5 9.8 2.8 75

83

(200 rpm) was continuously carried out, and filtration was intermittently carried out (7 min filtration and 3 min pause) using suction pump. The bubbles pushed the sludge to flow upward between the membrane modules to minimize membrane fouling. However, membrane fouling occurred inevitably during the IMCR process. In order to achieve steady flux, operating pressure had to be enhanced. This aggravated membrane fouling. The membrane fouling of the IMCR could be indexed by an increase of trans-membrane pressure (TMP). When the concentrations of suspended solids (SS) in the membrane separation unit were low, the TMP varied slightly with the increasing SS, whereas when SS increased to a critical value, the TMP increased sharply. As a result, sludge in the IMCR was discharged and the membrane was cleaned by physical and chemical methods. First of all, a sludge cake was flushed out by tap water. Secondly, membrane modules were cleaned chemically by mixed solution of NaClO and NaOH (effective Cl 3000 mg/L, NaOH 500 mg/L). Finally, the modules were dipped into distilled water for 8 h.

2.3. Analytical methods The absorbance at kmax of the liquid was measured and the full wavelength scanning (from 190 nm to 1000 nm) was determined by running a spectrophotometer (HACH DR-5000). Percentage of color removal was calculated by comparing the absorbance values for the treated wastewater to original wastewater. Distilled water served as a reference. Supernatants were withdrawn from the membrane separation unit and were centrifuged at 8000 rpm for 10 min to remove suspended solids from the liquid medium. Sufficient amounts of HgSO4 were added to precipitate chloride ions into HgCl2 in order to avoid chloride ion interfering with COD measurement. The COD contents of the samples were analyzed according to standard methods [17]. All samples were analyzed in triplicate and mean values were reported. The pH measurements were done by using the relevant probe and analyzer (METTLER TOLEDO FE20pH meter). The zeta potentials and floc sizes of particles in raw and coagulated water were evaluated using a zeta meter (Zetasizer, Malvern, UK) and a particle size analyzer (Mastersizer 2000, Malvern, UK). The resistance-in-series model was applied to estimate the filtration resistance and its build-up in a membrane filtration process [18]. According to this model, the permeate flux (J) can be expressed as

J¼

DP

lRt

¼

DP

lðRm þ Rf Þ

ð1Þ

where DP is the trans-membrane pressure, l is the viscosity of permeate, Rt is the total filtration resistance, Rm is the intrinsic membrane filtration resistance, Rf is the physically reversible and irreversible filtration resistance. Scanning electron microscopy (SEM) observations were performed. For SEM observation, the samples were fixed in a 2.5% (v/v) glutaraldehyde solution, dehydrated in grading water–ethanol solutions, dried under vacuum conditions and then sputtercoated with gold before SEM pictures were taken with a JEOL JSM-500LV microscope. Contact angle measurement was employed to determine surface hydrophilicity of the original and fouled membranes according to the reported method [19]. Floc structure was represented as the fractal dimension (dF). The fractal dimension was calculated based on the small-angle static light scattering (SALLS) theory by graphing a log–log plot of intensity (I) vs. the wave number (Q), which was a function of the scattering angle of each detector. In order to precisely calculate the fractal dimension from the log I–log Q plot, Gaussian cutoff

84

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

3

100

2.5 Absorbance

function was used to cover Guinier regime data. The detailed relevant information had been discussed in the previous work [20]. A Waters liquid chromatography system consisting of the following components was used for the Molecular Weight (MW) analysis: Waters 2487 Dual k Absorbance Detector, Waters 1525 pump system. Separation was performed with a Shodex KW 802.5 column (Shoko Co., Japan). The mobile phase, Milli Q water buffered with 5 mM phosphate to pH 6.8, and 0.01 M NaCl, was filtered through a 0.22 lm membrane, and then degassed for 30 min. The flow rate was 0.8 mL/min and the injection volume was 200 lL. The detailed information had been discussed in the early reports [21,22].

80

2 60 1.5 40 1 20

0.5 0

0.2

0.4

0.6 0.8 1.0 1.2 PACl dose (mmol/L)

1.4

1.6

Decolorization efficiency (%)

Fig. 1. Schematic diagram of the IMCR treating textile wastewater (1-feed tank; 2-coagulant tank; 3, 4, 11-pump; 5, 14-liquid flowmeter; 6-stirrer; 7-integrated membrane coagulation reactor; 8-air diffuser; 9-membrane module; 10-manometer; 12-air flowmeter; 13-air compressor; 15, 16-valve).

0

3. Results and discussion 450

100

400 80

COD (mg/L)

350

3.1.1. Decolorization and COD removal with PACl dose Coagulant dosage was an important factor during the membrane coagulation process, and the decolorization and COD removal with different PACl doses were presented in the Fig. 2. The influent pH value was fixed at 7.5 and the hydraulic retention time (HRT) was 2 h. When the PACl dose was 0.2 mmol/L, the decolorization by coagulation and by the IMCR were 29% and 50%, respectively. Both increased with the PACl dose. With the PACl dose of 1.2 mmol/L, the decolorization achieved by coagulation was 70% while the color removed by the IMCR was 82%. Then they decreased when the PACl was added further. When the PACl dose was 1.6 mmol/L, the decolorization by coagulation and by the IMCR were 59% and 74%, respectively. When it came to the COD removal, similar results were achieved. Both the COD removal by coagulation and by the IMCR increased when the PACl dose was no more than 1.2 mmol/L. They could peak at 47% and 68%, respectively. The effluent COD was 138 mg/L. However, the COD removal got worse with the increasing PACl dose. The zeta potential of textile wastewater was negative when the pH value was 7.5 (data shown in Section 3.1.2). Increasing PACl dosage neutralized the negative charge and favored coagulation process. As a result, optimal decolorizaton and COD removal efficiencies could be achieved with the PACl dose of 1.2 mmol/L. However, when the PACl dosage was more than 1.2 mmol/L, the pollutants in the wastewater would be positively charged due to excessive aluminum ions. They restabilized again and made the color and COD removal decreased. Compared to decolorization, membrane played a more important role in COD removal. COD removal by membrane could account for about one third of the total COD removal. Residual organics could be rejected both by membrane and floc cake which affiliated on the membrane surface.

300 60

250 200

40

150 100

20

50 0

0.2

0.4

0.6 0.8 1.0 1.2 PACl dose (mmol/L)

1.4

1.6

COD removal efficiency (%)

3.1. Performance of the IMCR with different operation conditions

0

Fig. 2. Decolorization and COD removal with different PACl doses.

3.1.2. Decolorization and COD removal with influent pH value The influent pH value affected the reactor performance as well and the color and COD removal with the pH values were presented in the Fig. 3. The PACl dose was fixed at 1.2 mmol/L and the HRT was 2 h. When the influent pH was 2.5, the color removal by coagulation and by the IMCR were 29% and 56%, respectively. Both peaked when the pH was 5.5. The decolorization efficiency by coagulation was 78% while the color removal by the IMCR was 91%. Then they decreased with the growing pH value. However, when the influent pH value was 13, they increased slightly. The COD removal efficiency by coagulation was 11% while the COD removed by the IMCR was 31% with the influent pH of 2.5. When the influent pH value was 5.5, the optimal COD removal was achieved. The COD removal efficiency by the reactor was 79%, and the effluent COD value was around 90 mg/L. As well as color removal, the COD removal efficiency decreased with the raising pH and grew slightly when the influent pH value was 13.

85

Absorbance

2.5

80

2 60 1.5 40 1 20

0.5

COD (mg/L)

0

2.5

450 400 350 300 250 200 150 100 50 0

4.0

5.5

7.0 8.5 Influent pH

10.0

11.5

0

13.0

100 80 60 40 20 2.5

4.0

5.5

7.0 8.5 Influent pH

10.0

11.5

13.0

COD removal efficiency (%)

100

3

Decolorization efficiency (% )

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

0

values were presented in the Fig. 4. The PACl dose was fixed at 1.2 mmol/L and the influent pH was 5.5. In general, decolorization was performed better with the longer HRT. Color removal by the IMCR was 41% with the HRT of 0.5 h while it could reach 98% with the HRT of 3 h. However, decolorization efficiency increased slightly when the HRT was lengthened further. Similar results were achieved in the COD removal. When the HRT was 0.5 h, the COD removal by coagulation was only 6%. Effluent COD was around 300 mg/L. Coagulation was not performed well due to the short contact time between pollutants and coagulant. COD removal efficiency increased with the growing HRT. When the HRT was 3 h, the total COD removal arrived at 88% and the effluent COD was around 50 mg/L. When the HRT was 4 h, effluent quality was improved slightly. Take operation cost into account, the HRT of 3 h could be regarded as the optimal value. Compared to COD removal, decolorization was performed better in the IMCR. Integrate absorbance in the whole region, which allowed assessing the decolorization of the reactor. With the HRT of 3 h, the full wavelength scanning (from 190 nm to 1000 nm) of the influent, supernatant and effluent were determined by running the spectrophotometer. The results were shown in the Fig. 5. It could be seen that the curve shapes of influent and supernatant appeared to be completely identical. The supernatant did not show any characteristic peak within the scanning wavelength. It indicated that the molecular structures of the dyes were not changed through the coagulation. The dyes were only transferred from the wastewater into the flocs which could be separated more easily. Compared to supernatant, the peak at the wavelength of 622 disappeared in the effluent. The dyes which had characteristic peak at this wavelength could be removed completely by the membrane

30

3

0 -10

2.5

-20

-40

Influent Mixed liquor

-50 -60 2.5

4

5.5

7

8.5

10

11.5

13

1

0

Fig. 3. Decolorization, COD removal and zeta potential with different pH values.

3.1.3. Decolorization and COD removal with HRT HRT was a key parameter in IMCR process, not only because system performance but because reactor volume was associated with it. Color and COD removal efficiencies with different HRT

1.5

0.5

pH value

0.5

1.0

1.5

2.0 2.5 HRT (h)

3.0

3.5

4.0

100

450 400

80

350 COD (mg/L)

In order to explore the performance mechanism, we investigated the charge characteristics of the influent and mixed liquor further. In general, the pollutants in the wastewater were negatively charged and the zeta potentials of the influent were negative with the low pH values. The flocs in the mixed liquor possessed fewer charges due to the charge neutralization of the PACl, and the zeta potential of the mixed liquor moved towards positively. When the pH value was 5.5, the zeta potential of the mixed liquor was near the zero. As a result, the optimal color and COD removal could be attributed to charge neutralization mechanism. When the pH value was 13, the zeta potential of the influent became positive. However, the formation of AlðOHÞ 4 dominated the hydrolysis products. Charge neutralization effect could occur again. As a result, both the color and COD removal efficiencies grew slightly with the pH of 13.

2

300 60

250 200

40

150 100

20

50 0

0.5

1.0

1.5

2.0 2.5 HRT (h)

3.0

3.5

4.0

0

Fig. 4. Decolorization and COD removal with different HRTs.

COD removal efficiency (%)

-30

100 90 80 70 60 50 40 30 20 10 0

Decolorization efficiency (%)

10

Absorbance

Zeta potential (mV)

20

86

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

3

Influent Supernatant Effluent

2.5

Absorbance

2 1.5 1 0.5 0 -0.5 190

390

590

790

990

Wavelength (nm) Fig. 5. Spectrogram of the influent, supernatant and effluent with the HRT of 3 h.

rejection. Besides, membrane had more potential to remove characteristic peaks and reduce peak area within the ultraviolet range (less than 400 nm).

3.2. Characteristics of membrane fouling during the IMCR process Membrane fouling was tested with different PACl doses. The pH value was fixed at 5.5 and the HRT was 3 h. The variations of the TMP during the process were shown in the Fig. 6. In general, the membrane fouling could be grouped into three phases based on the TMP variation rates. In the first phase, the membrane fouling was formed quickly after the contact between the membrane and flocs. Then the TMP got into steady state, i.e., Phase II. At last, the membrane fouling worsened and the TMP increased sharply. Phase I occurred in the first 6 days of the operation process. The initial TMP values were similar; however, the TMP of PACl 1.4 mmol/L arrived at 25 kPa with 6 day’s operation. The mean TMP increases of the PACl 1 mmol/L, PACl 1.2 mmol/L and PACl 1.4 mmol/L were 2.37 kPa/d, 2.75 kPa/d and 3.18 kPa/d. All of the TMP values varied little in the Phase II, and the mean TMP increases were below 0.25 kPa/d. During the Phase III, the TMP of PACl 1.4 mmol/L grew fastest and peaked at 57.2 kPa on the 25th day. The mean TMP increases of the PACl 1 mmol/L, PACl 1.2 mmol/L and PACl 1.4 mmol/L were 2.81 kPa/d, 3.41 kPa/d and 4.79 kPa/d. All were greater than they were in Phase I. It could indicate that the membrane fouling in Phase III was more severe than that in Phase I. Besides, higher PACl dose aggravated the membrane fouling. The optimum condition of coagulation for

Fig. 6. TMP variations with different PACl doses during the IMCR process.

conventional treatment systems was not necessarily applicable to membrane-based treatment systems. Surface and cross-section morphologies of the new and fouled membrane were analyzed by employing scanning electron microscopy. The results were presented in the Fig. 7. In terms of the new membrane, there were lots of micro pores in both the surface and cross section. However, the membrane fouling occurred instantly upon the contact between membrane and mixed liquid. During the Phase I, due to the suction effect, membrane surface was covered by the sludge cake of dyes and flocs quickly. The membrane filtration resistance increased and the TMP grew. However, the membrane fouling mainly occurred on the surface of the membrane and it could be removed easily by the physical cleaning (data not shown). The sludge cake could also fall out from the membrane surface with the shear provided by the aeration. When the dynamic equilibrium was achieved between cake falling out and attachment, the membrane filtration resistance kept steady and the TMP varied little. During the IMCR process, Phase II was the key stage to maintain long-term and steady-state operation. In the third phase, almost all the membrane surface was covered by the sludge cake and many small flocs were sucked into the membrane pores. The micro pores of the membrane were blocked severely and the filtration resistance grew sharply. The irreversible membrane fouling occurred in the stage and the foulants could not be removed effectively by the physical cleaning only. Early studies indicated that a more hydrophilic membrane could provide a better wettability, possess a higher permeate flux and result in a lower membrane fouling potential [23,24]. Contact angle measurement was employed to analyze the hydrophilicity/ hydrophobicity of the membrane. The results were presented in Fig. 8. The contact angle value of new membrane was 52.5 degrees, while the values of the fouled membranes were 81.2 (with PACl 1 mmol/L), 81.8 (with PACl 1.2 mmol/L) and 80.9 (with PACl 1.4 mmol/L) degrees. In general, higher contact angles indicated a more hydrophobic surface [25]. Compared to new membrane, the fouled membranes were more hydrophobic. However, it was affected slightly by the different PACl doses. Increasing membrane hydrophobicity suggested that the resistance to membrane fouling was much worse and it was not applicable for the long-term operation during the IMCR process. Huang et al. reported that changing mutual affinities of contaminants or their affinities to membrane surfaces could impact membrane filtration and alleviate membrane fouling [13]. When it came to IMCR process, the floc affinity to membrane surface might be changed through adding oxidant or using novel coagulant. If its affinity to membrane decreased, the membrane could became more hydrophilic even with long-term operation. Further study was needed. Our findings also agreed with the previous report that dosage of coagulant was influential on the degree of membrane fouling and high dosage of coagulant frequently caused more severe fouling [26]. In order to explore the phenomena, fractal dimension (dF) was analyzed for floc structure. The results were presented in the Fig. 9. It indicated that the fractal dimension increased with the enhancing coagulant dose. The dF value of flocs formed with PACl 1 mmol/L was 1.88 while it reached 2.22 with PACl 1.4 mmol/L. Floc geometry was commonly described as fractal, implying that aggregates were self-similar and scale invariant. The dF was defined in three dimension ranges 1–3, and gave information on floc structure. The dF value close from 1 characterized open and highly branched structure, whereas it close from 3 revealed very compact and spherical aggregates. Floc particles formed with the PACl dose of 1 mmol/L were more porous and less compact than those formed with PACl 1.4 mmol/L. As a result, a more permeable filter cake was formed with PACl 1 mmol/L. This could alleviate the membrane fouling effectively. Early study reported that underdosed coagulant could lead to good results in terms of membrane

87

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

a

b

c

d

Fig. 7. SEM images of the new membrane and the fouled membrane with 25 day’s operation ((a) surface of the new membrane; (b) surface of the fouled membrane; (c and e) cross section of the new membrane; (d and f) cross section of the fouled membrane).

a

b

c

d

Fig. 8. Contact angle images of the new membrane and the fouled membrane with 25 day’s operation ((a) new membrane; (b) fouled membrane with PACl 1 mmol/L; (c) fouled membrane with PACl 1.2 mmol/L; (d) fouled membrane with PACl 1.4 mmol/L).

recovery by hydraulic cleaning [14]. When the IMCR was employed to treat wastewater, not only pollutant removal but also membrane fouling should be taken into consideration. In order to achieve long-term and steady-state operation, it might be deserved to sacrifice a higher pollutant removal performance by reducing coagulant dose for alleviating the membrane fouling. 3.3. Effect of different membrane MWCOs on membrane fouling Membrane is the significant factor to IMCR. Not only pollutant removal but operation stability is related to it. In this work, the IMCR was equipped with three sets of membrane modules in sequence. There MWCOs were 50, 100 and 150 kDa, respectively.

The PACl dose was fixed at 1.2 mmol/L. The pH value and HRT were 5.5 and 3 h, respectively. There was little difference in the pollutant removal (data not shown). Color could be removed almost completely independent of the membrane MWCO. However, the TMP and Rf varied greatly during the IMCR process with different membrane MWCOs. The results were presented in Fig. 10. In general, the IMCR with all MWCOs experienced three phases. However, the time of each phase varied. During the IMCR with membrane MWCO 50 kDa, the initial TMP was 5 kPa. It grew sharply within the first 4 days’ operation. The mean TMP increase was 2 kPa/d. The Phase I in the IMCR with membrane MWCO 100 kDa was 6 d and its mean TMP increase was around 2.7 kPa/d. When it came to the IMCR with MWCO 150 kDa, the time of Phase I

88

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

5

PACl 1 mmol/L dF 1.88 PACl 1.2 mmol/L dF 2.05 PACl 1.4 mmol/L dF 2.22

4 3

Log I

2 1 0 -1 -2 -3 -6

-5

-4

-3

-2

-1

Log Q Fig. 9. Fractal dimension of flocs formed with different PACl doses during the IMCR process.

tively. Cake resistance dominated the Rf and it could be physically reversible. Reversible membrane fouling occurred in this stage. During the Phase III, the Rf of the IMCR with membrane MWCO 50 kDa, 100 kDa and 150 kDa were 2.95  1011, 5.92  1011 and 10.99  1011 m1, respectively, and all of the mean Rf increment were greater than they were in Phase I. Unlike Phase I, physically irreversible filtration resistance was predominant in the Rf and irreversible membrane fouling mainly occurred in Phase III. After 25 days’ operation, the distributions of the IMCR filtration resistances with different MWCOs were analyzed. Fig. 11 showed that the intrinsic membrane filtration resistance of the IMCR with membrane MWCO 50, 100 and 150 kDa were 0.9  1011, 0.65  1011 and 0.35  1011 m1, respectively. The Rf values of the IMCR with membrane MWCO 50, 100 and 150 kDa were 3.2, 9.1 and 31.4 times of there intrinsic membrane filtration resistances, respectively. When the membrane with higher MWCO was adopted during the IMCR process, its intrinsic membrane filtration resistance was fewer. However, it could be fouled more seriously. The dominant filtration resistances consisted of cake resistance and physically irreversibly filtration resistance. The intrinsic membrane filtration resistance was no longer significant. In order to explore the difference among the IMCR with different MWCOs, particle size distribution of the coagulated suspension formed in membrane separation unit was analyzed. Fig. 12 showed that the coagulated flocs around 0.18 lm dominated the suspended particles. Early study reported that particles near 0.2 lm in diameter produce severe fouling [18]. However, in this work, the membrane fouling occurred in the IMCR with MWCO 50 kDa was not very serious and relatively long time for steady-state operation was achieved. The nominal pore size of the membrane MWCO 50 kDa was about 0.05 lm. The size of coagulated flocs was so large that they might hardly deposit in the membrane pores, which was assumed on the basis of the steady-state flux for the filtration. The nominal pore size of the membrane MWCO 150 kDa was around 0.15 lm. It was found that the floc sizes were just comparable to the size of micro pores. The particles could plug themselves into the micro pores and cause physically irreversible fouling. During the IMCR process, it should take both nominal pore size of the membrane and floc size into consideration. Membrane fouling could not be attributed to any single factor. It seemed to be quite important to avoid the formation of flocs with the sizes that were close to those of membrane pores. 3.4. MW distribution of membrane eluent through chemical cleaning One of the critical problems encountered during the membrane process was irreversible membrane fouling [27]. All fouled mem-

12

was much longer. The TMP increased greatly within the first 9 days’ operation. Both the lasting time and mean TMP increase were the most. The time of Phase II for IMCR with 50 kDa, 100 kDa and 150 kDa were 17, 13 and 8 d, respectively. Then all of the TMPs increased sharply and peaked at 24, 36 and 46 kPa, respectively. Both the TMP and TMP increment were greater when the IMCR was equipped with larger membrane MWCO. The initial filtration resistance Rf of the IMCR with different membrane MWCOs were similar. However, in Phase I, it increased fastest and peaked at 3.7  1011 m1 during the IMCR with membrane MWCO 150 kDa. The mean Rf increment of the IMCR with membrane MWCO 50 kDa, 100 kDa and 150 kDa in the first phase were 0.09  1011, 0.22  1011 and 0.41  1011 m1 d1, respec-

Resistance (10 11 m -1 )

Fig. 10. TMP and Rf variations during the IMCR process with different MWCOs.

10 8 6 4 2 0 IMCR with

IMCR with

IMCR with

membrane MWCO membrane MWCO membrane MWCO 50 kDa

100 kDa

150 kDa

Fig. 11. Distributions of the IMCR filtration resistances with different MWCOs.

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

89

4. Conclusions

10 9

Volume (%)

8 7 6 5 4 3 2 1 0 0.01

0.1

1

10

100

1000

10000

Partical size (µm) Fig. 12. Particle size distribution of the coagulated suspension formed in membrane separation unit.

branes could be cleaned effectively by physical and chemical methods, and the permeate flux could almost recovered completely (data not shown). In order to explore the characteristics of foulants which devoted to the irreversible membrane fouling, the MW distributions of different membrane eluents through chemical cleaning were analyzed. Fig. 13 indicated that all eluents contained foulants whose MW varied between 5000 and 11,800 Da. Besides, foulants with MW around 65,000 Da were detected in the elluents from IMCR with membrane MWCO 100 and 150 kDa. However, there abundances were lower than those with less MW. Early study reported that in terms of MW, a higher MW fraction of organics could cause faster fouling during UF than a lower MW fraction [28,29]. However, Carroll et al. [30] indicated that lower MW fractions had a higher impact on membrane fouling than higher MW ones. In this work, independent of the MWCO of membrane, lower MW fractions resulted in the irreversible membrane fouling. Besides, the MWs of foulants were much less than the membrane MWCO. The bigger MW dyes could form a gel or concentration polarization layer on top of the membrane and prevent the subsequent pollutants with the MWs near the membrane pores from entering them. However, the dyes with smaller MW could easily enter membrane pores and cause pore blocking and adsorption type fouling. The larger the membrane pores, the more susceptible it was to irreversible membrane fouling.

Fig. 13. MW distributions of different membrane eluents through chemical cleaning.

Treating textile wastewater by employing an IMCR, color could be removed almost completely and COD removal efficiency could reach 88% with the optimal operation conditions. Membrane fouling experienced three phases and a higher PACl dose aggravated the membrane fouling. The optimum condition of coagulation for conventional treatment systems was not necessarily applicable to membrane-based treatment systems. Compared to new membrane, the fouled membranes were more hydrophobic. The dF value of flocs formed with PACl 1 mmol/L was 1.88 while it reached 2.22 with PACl 1.4 mmol/L. Fractal dimension increased with the enhancing coagulant dose. A more permeable filter cake was formed with PACl 1 mmol/L and could alleviate the membrane fouling effectively. Both the TMP and TMP increment were greater when the IMCR was equipped with larger MWCO membrane. The Rf values of the IMCR with membrane MWCO 50, 100 and 150 kDa were 3.2, 9.1 and 31.4 times of there intrinsic membrane filtration resistances, respectively. Membrane was fouled more seriously when the higher MWCO membrane was adopted during the IMCR process. Independent of the membrane MWCO, lower MW fractions resulted in the irreversible membrane fouling. The larger the membrane pores, the more susceptible it was to irreversible membrane fouling. Acknowledgements The work was supported by the National Natural Scientific Foundation (Nos. 41206106 and 51278258). The authors would like to thank the editor and the anonymous reviewers for their editing and review. References [1] S. Wijetunga, X. Li, C. Jian, Effect of organic load on decolourization of textile wastewater containing acid dyes in upflow anaerobic sludge blanket reactor, J. Hazard. Mater. 177 (2010) 792–798. [2] S. Wijetunga, X. Li, R. Wenquan, J. Chen, Evaluation of the efficacy of upflow anaerobic sludge blanket reactor in removal of color and reduction of COD in real textile wastewater, Bioresour. Technol. 99 (2008) 3692–3699. [3] Q. Wang, Z. Luan, N. Wei, J. Li, C. Liu, The color removal of dye wastewater by magnesium chloride/red mud (MRM) from aqueous solution, J. Hazard. Mater. 170 (2009) 690–698. [4] T. Robinson, G. McMullan, R. Marchant, P. Nigam, Remediation of dyes in textile effluent: a critical review on current treatment technologies with a proposed alternative, Bioresour. Technol. 77 (2001) 247–255. [5] C.H. Shin, J.S. Bae, A stability study of an advanced co-treatment system for dye wastewater reuse, J. Ind. Eng. Chem. 18 (2012) 775–779. [6] N.K. Lazaridis, T.D. Karapantsios, D. Georgantas, Kinetic analysis for the removal of a reactive dye from aqueous solution onto hydrotalcite by adsorption, Water Res. 37 (2003) 3023–3033. [7] S.F. Kang, C.H. Liao, M.C. Chen, Pre-oxidation and coagulation of textile wastewater by the Fenton process, Chemosphere 46 (2002) 923–928. [8] N. Othman, S.N. Zailani, N. Mili, Recovery of synthetic dye from simulated wastewater using emulsion liquid membrane process containing tri-dodecyl amine as a mobile carrier, J. Hazard. Mater. 198 (2011) 103–112. [9] A.A. Hassan, Simple physical treatment for the reuse of wastewater from textile industry in the Middle East, J. Environ. Eng. Sci. 6 (2007) 115–122. [10] I. Vergili, Y. Kaya, U. Sen, Z. Beril, C. Aydiner, Techno-economic analysis of textile dye bath wastewater treatment by integrated membrane processes under the zero liquid discharge approach, Resour. Conserv. Recycl. 58 (2012) 25–35. [11] K.J. Howe, A. Marwah, K. Chiu, S.S. Adham, Effect of coagulation on the size of MF and UF membrane foulants, Environ. Sci. Technol. 40 (2006) 7908–7913. [12] L. Fan, T. Nguyen, F.A. Roddick, J.L. Harris, Low-pressure membrane filtration of secondary effluent in water reuse: pre-treatment for fouling reduction, J. Memb. Sci. 320 (2008) 135–142. [13] H. Huang, K. Schwab, J.G. Jacangelo, Pretreatment for low pressure membranes in water treatment: a review, Environ. Sci. Technol. 43 (2009) 3011–3019. [14] E. Barbot, S. Moustier, J.Y. Bottero, P. Moulin, Coagulation and ultrafiltration: understanding of the key parameters of the hybrid process, J. Memb. Sci. 325 (2008) 520–527. [15] H. Huang, N.H. Lee, T. Young, A. Gary, J.C. Lozier, J.G. Jacangelo, Natural organic matter fouling of low-pressure, hollow-fiber membranes: effects of NOM source and hydrodynamic conditions, Water Res. 41 (2007) 3823–3832.

90

J. Li et al. / Chemical Engineering Journal 240 (2014) 82–90

[16] X.D. Wang, L. Wang, Y. Liu, W.S. Duan, Ozonation pretreatment for ultrafiltration of the secondary effluent, J. Memb. Sci. 287 (2007) 187–191. [17] APHA, Standard Methods for the Examination of Water and Wastewater, American Public Health Association, Washington, DC, 1999. [18] J.D. Lee, S.H. Lee, M.H. Jo, P.K. Park, C.H. Lee, J.W. Kwak, Effect of coagulation conditions on membrane filtration characteristics in coagulation– microfiltration process for water treatment, Environ. Sci. Technol. 34 (2000) 3780–3788. [19] Y. Baek, J. Kang, P. Theato, J. Yoon, Measuring hydrophilicity of RO membranes by contact angles via sessile drop and captive bubble method: a comparative study, Desalination 303 (2012) 23–28. [20] C.M. Sorensen, Light scattering by fractal aggregates: a review, Aerosol Sci. Technol. 35 (2001) 648–687. [21] C.W.K. Chow, R. Fabris, J. van Leeuwen, D. Wang, M. Drikas, Assessing natural organic matter treatability using high performance size exclusion chromatography, Environ. Sci. Technol. 42 (2008) 6683–6689. [22] D. Wang, L. Xing, J. Xie, C.W.K. Chow, Z. Xu, Y. Zhao, M. Drikas, Application of advanced characterization techniques to assess DOM treatability of micropolluted and un-polluted drinking source waters in China, Chemosphere 81 (2010) 39–45. [23] A. Schafer, L. Nghiem, T. Waite, Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis, Environ. Sci. Technol. 37 (2003) 182–188.

[24] D. Rana, T. Matsuura, Surface modifications for antifouling membranes, Chem. Rev. 110 (2010) 2448–2471. [25] C.Y. Tang, Y.N. Kwon, J.O. Leckie, Effect of membrane chemistry and coating layer on physiochemical properties of thin film composite polyamide RO and NF membranes: II. Membrane physiochemical properties and their dependence on polyamide and coating layers, Desalination 242 (2009) 168– 182. [26] K. Kimura, T. Maeda, H. Yamamura, Y. Watanabe, Irreversible membrane fouling in microfiltration membranes filtering coagulated surface water, J. Memb. Sci. 320 (2008) 356–362. [27] C.W. Li, Y.S. Chen, Fouling of UF membrane by humic substance: effects of molecular weight and powder-activated carbon (PAC) pre-treatment, Desalination 170 (2004) 59–67. [28] C.F. Lin, T.Y. Lin, O.J. Hao, Effects of humic substance characteristics on UF performance, Water Res. 34 (2000) 1097–1106. [29] W. Yuan, A.L. Zydney, Humic acid fouling during microfiltration, J. Memb. Sci. 157 (1999) 1–12. [30] T. Carroll, S. King, S.R. Gray, B.A. Bolto, N.A. Booker, The fouling of microfiltration membranes by NOM after coagulation treatment, Water Res. 34 (2000) 2861–2868.