Chapter 3: Materials and methods
3.2 Methods
3.2.7 Excitation-Emission Matrix (EEM) method
Excitation-Emission Matrix (EEM) method was used to characterize the fluorescent substance in the membrane fouling including the cake layer deposited on membrane surface and small particle absorbed into the membrane pore. This method can be used to identify many kinds of protein as well as humic acid-like and fulvic acid –like but polysaccharide-like substances can not be detected by this method. Figure 3.9 shows the location of EEM peak based on excitation-emission wavelength. The specific details about excitation-emission wavelength was also showed in Table 3.5
Figure 3.9 Location of EEM peak based on excitation-emission wavelength Source: Chen et al., 2003
32
Table 3.5 Excitation-emission wavelength of the five regions of EEM
Region Excitation (nm)
Emission
(nm) Description
I 200-250 280-330 Tyrosine-like, Aromatic protein
II 200-250 330-380 Tryptophan-like, Aromatic protein
III 200-250 380-500 Fulvic acid-like
IV 250-400 280-380 Soluble microbial by-product-like
V 250-400 380-500 Humic acid-like
Source: Chen et al., 2003
33
Chapter 4
Results and Discussions 4.1 Critical flux determination
4.1.1
Flux-step tests
Flux-step method was conducted to determine the critical flux of different membranes under different operating conditions, in which the beginning flux at 6 lm-2h-1, a step height of 3 lm-2h-1 and step length of 15 min were selected. All the parameters such as TMP and flux were automatically recorded by the computer system (as introduced in the section 3.1.1).
The flux-step experiments were divided into two stages with different purposes. In order to determine critical flux, the first stage was conducted from the beginning of experiment up to the 120th minute in case of the HPI membrane in MBR-1 with the flux from 6 to 30 lm-2h-1, while for the HPI and HPO membrane in MBR-2, it was from the beginning up to the 210th minute with the flux from 6 to 45 lm-2h-1. The second stage was conversely conducted against the first stage to observe effects of shear-stress (discussed further in section 4.1.3) to the foulants during the flux-step experiments as the descending phases. Figure 4.1~3 illustrate the critical flux determination for the HPI membrane under the activated sludge concentration (MBR-1) of 7,000 - 7,500 mg-MLSS/L, and for HPI and HPO membrane under 6,000 - 6,500 mg-MLSS/L (MBR-2), respectively.
From Figure 4.1, the TMP was almost stable when the flux increased from 6 to 18 lm-2h-1. At the flux of 21 lm-2h-1, a little change of TMP from -1.3 kPa to -1.5 kPa was observed which indicated a growing fouling appeared on the membrane. Fouling mechanism, under low flux, is considered as the absorption of small particle such as solute and colloidal fractions in sludge while flocs deposition is absent under a microscope observation (Chang et al., 2002). The sharp increase of TMP from 24 lm-2h-1 showed a large amount of membrane fouling created and therefore indicated the vicinity of the critical flux. This trend of TMP is similar in Figure 4.2-3 in which the sharp increase of TMP was observed at a flux of 33 lm-2h-1.
34
Comparisons in changes of TMP between HPI membranes operated in MBR-1 and MBR-2 and between HPI and HPO membrane operated in MBR-2 were plotted in Figure 4.4 so that the occurrence of membrane fouling on membranes in flux-step experiments are easily compared.
Figure 4.1 Critical flux determination of HPI membrane in MBR-1
The TMP changes of HPI membranes performed in MBR-1 (green curve) and MBR-2 (red curve) give information about the effects of activated sludge characteristics on membrane fouling. Two TMP curves seem to be linear from the flux of 6 - 18 lm-2h-1. With the consecutive flux step (from 21 lm-2h-1), the TMP line of HPI membrane operated in MBR-1 go dramatically far from the line of HPI membrane operated in MBR-2. The reason for these different trends in TMP changes may be due to the difference of activated sludge concentration. With the concentration 7,000 - 7,500 mg/L, the mixed liquor in MPR-1 would contain a higher amount of colloids, solutes, organic macromolecular such as EPS, SMP or other substances resulting from the cell lysis than that of the 6,000 - 6,500 mg-MLSS/l in MBR-2. Therefore, membranes operating with the higher MLSS concentration are easier to be fouled than those of the lower (Le-Clech et al., 2006).
Regarding to the operation of HPI and HPO membrane in MBR-2, the trends of two TMP curves (red and blue) were quite similar when the flux increased 6 to 27 lm-2h-1.
35
In the next two consecutive flux steps, the TMP of HPI membrane was slightly higher than that of HPO membrane; however, became lower after. In general, this phenomenon means that TMP of the HPI membrane seems to be more sustainable than HPO membrane in the short-term experiments of flux-step method.
Although the changes of TMP can be seen visibly, it is not easy to exactly determine the critical flux value for each case by just observing the TMP changes from these figures. The method to determine critical flux will be described further in section 4.1.2. The flux-step method performance is the first stage of this study, by which critical flux can be found and used for the next tests.
Figure 4.2 Critical flux determination of HPI membrane in MBR-2
0 30 60 90 120 150 180 210 240 270 300 330 360 390 -5.0
-4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5
-1.0 -0.5
0.0 0
3 6 9 12 15 18 21 24 27 30 33 36 39 42 TMP (kPa) 45
Flux (Flux (Lm-2h-1)
TMP (kPa) Flux (Lm-2 h-1 )
Time (min)
36
HPI membrane operated in MBR-1 HPI membrane operated in MBR-2 HPO membrane operated in MBR-2
TMP (kPa)
Flux (Lm-2h-1)
Figure 4.3 Critical flux determination of HPO membrane in MBR-2
Figure 4.4 Comparison of TMP changes between membranes during flux-step trials Time (min)
37
4.1.2 Critical and sub-critical flux determination
In this study, permeability and fouling rate used in critical flux finding were based on an assumption of Le-Clech (2003b). Permeability indicates a quality of membrane that allows the clean water pass through membrane. A decline of permeability shows the appearance of membrane fouling preventing the efficient performance of MBRs. The decrease of membrane permeability is corresponding with the increase of membrane fouling rate.
Figure 4.5 Permeability and fouling rate of HPI membrane in MBR-1
A relationship between permeability and fouling rate of HPI membrane in MBR-1, in MBR-2 and HPO membrane in MBR-2 was illustrated in Figures 4.5~7, respectively.
In Figure 4.5, permeability increases from 21.2 to 27.7 lm-2h-1kPa-1 with the increase of flux from 6 to 15 lm-2h-1 (within 3 consecutive flux-steps). Results of unchanged fouling rate (blue curve) showed that membrane fouling in this stage was insignificant. Via flux-step increments from 18 to 30 lm-2h-1, the fouling rate then raised together with the reduction of membrane permeability. Figures 4.6~7 show the relationship between permeability and fouling rate of HPI and HPO membrane operated in the same reactor (MBR-2). The permeability of these two cases seems to be more stable than the one in Figure 4.5. In particular, membrane permeability of HPI membrane reached up to 24.7 lm
-38
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 10
12 14 16 18 20 22 24 26 28 30
0.000 0.005 0.010 0.015 0.020 0.025 0.030 K (Lm-2h-1kPa-1)
dTMP/dt (kPa.min-1)
K (Lm-2 h-1 kPa-1 ) dTMP/dt (kPa.min-1 )
Flux (Lm-2h-1)
2h-1kPa-1 after 6 consecutive flux-steps operation, and then it gradually descended. For HPO membrane, the permeability was quite stable for 7 consecutive flux-steps operation before going down.
Figure 4.6 Permeability and fouling rate of HPI membrane in MBR-2
The specific data of permeability values and fouling rates are given in Table 4.1.
Equation 3.1 was used to calculate the fouling rate and permeability was identified by Equation 3.2~3. In order to determine the exact values of critical fluxes, the initial permeability values were determined. They are 21.2, 21.8 and 21.4 lm-2h-1kPa-1 for HPI membrane operated in MBR-1, in MBR-2 and HPO membrane operated in MBR-2. Next, 90% of initial permeability values (K0) were calculated, as given in Table 4.2, which were used to calculate the sub-critical fluxes.
39 Flux (Lm-2h-1)
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
K (Lm-2 h-1 kPa-1 )
10 12 14 16 18 20 22 24 26 28 30
dTMP/dt (kPa.min-1 )
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040 K (Lm-2h-1kPa-1)
dTMP/dt (kPa.min-1)
Figure 4.7 Permeability and fouling rate of HPO membrane in MBR-2
40
Table 4.1 Permeability and fouling rate of HPI and HPO membranes in step-flux experiments
HPI operated in MBR-1 HPI operated in MBR-2 HPO operated in MBR-2 Flux (lm-2h-1) Permeability (*) K0 : initial permeability
41
HPI membrane operated in MBR-2 HPO membrane operated in MBR-2 HPI membrane operated in MBR-1
TMP (kPa)
Flux (Lm-2h-1)
∆TMP
Table 4.2 Critical flux determination
Membrane MLSS
4.1.3 Hysteresis loop for the short-term experiment
Hysteresis loop was observed at three above step-flux experiments to assess the stability and reversibility of fouling. In the other word, the effects of shear-stress in MBRs against the formation of fouling were considered.
Figure 4.8 Hysteresis loop for the short-term tests
42
The results in Figure 4.8 point out that the TMP values of all corresponding fluxes obtained from the ascending phase were higher than in the descending phase. This indicates to the occurrence of membrane fouling because the increase in fouling rate was observed as the increase of TMP with a proportional relation. Equation 2.1 can be summarized as following:
∆P
( ∆P
4.1
The viscosity of solution (µ) in the equation was constant and flux (J) was the same in operation. Equation 4.1 clearly depicts the proportional relationship between total resistances (including membrane fouling) with TMP. The uneven of TMP (∆ TMP) between the same flux of hysteresis loop conducted in the ascending phase and descending phase was considered as the amount of fouling formed on the membrane surface. In Figure 4.8, at flux of 18 lm-2h-1, we can obviously see the difference about membrane fouling among three hysteresis curves. The ∆TMP of HPI membrane operated in 2 (red curve) is equal to 5.1% of the ∆TMP of HPI membrane operated in MBR-1 (green curve). While for the ∆TMP of HPO membrane operated in MBR-2 (blue curve), it is 43.1%. Moreover, at the initial point as well as the end point of this experiment (J = 6 lm-2h-1), it is 19.3% and 47.4% of the ∆TMP of HPI membrane operated in MBR-1 for HPI and HPO membrane in MBR-2, respectively.
Regarding to the removal of fouling by shear-stress, it can be assessed by the disparity of ∆TMP between the flux of 6 and 18 lm-2h-1 (other certain flux could be selected for this assessment excluding the initial flux of 6 lm-2h-1) in an experiment. In case of HPI membrane operated in MBR-1, the ∆TMP at 6 lm-2h-1 was equal to 26.38% of the ∆TMP at 18 lm-2h-1. That meant the fouling removal by shear stress was 73.62%.
Moreover, this removal was 45.2% for HPO membrane operated in MBR-2 and 0% for HPI membrane operated in MBR-2. In the later case of 0% of fouling removal, this can be explained due to the small amount of fouling absorbed in the inside of the membrane pore causing the uselessness of shear stress.
In conclusion, with the same membrane material, operation in a reactor with higher sludge concentration would cause a higher propensity of membrane fouling. HPO membrane seems to be easier fouled than HPI membrane in the short-term experiment.
43
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 -45
-40
-35
-30
-25
-20
-15
-10
-5
0
HPO membrane operated under 80% critical flux in MBR-2 HPI membrane operated under 60% critical flux in MBR-1 HPI membrane operated under 80% critical flux in MBR-2 HPI membrane operated under 60% critical flux in MBR-2
TMP (kPa)
Time (day)
4.2 TMP change in sub-critical flux operation
Long-term experiments were conducted to observe the occurrence of membrane fouling under sub-critical flux operation. Figure 4.9 shows the variation of TMP under various sub-critical fluxes. In theoretical point of view, the membrane fouling would not occur under sub-critical flux (Field et al., 1995). But in practice, membrane fouling always occurs even operated under sub-critical flux.
Figure 4.9 Variation of TMP under various sub-critical fluxes operation
In Figure 4.9, the imposed flux was controlled with a given sub-critical flux as calculated in Table 4.2. It is obvious that membrane fouling occurred in the first day of operation. The TMP of HPO membrane operated under 80% critical flux in MBR-2 exhibited the fastest jump up to approximately 40 kPa after 11-days operation. It is 16-days and 18-16-days operation for HPI membrane operated under 80% and 60% of critical flux in MBR-2, respectively. HPI membrane operated under 60% critical flux in MBR-1 shows the longest operation time, approximately 36 days.
The reason for the fast jump of the red color curve might be due to the hydrophobic interaction between fouling and membrane material. Green and blue color
44
show a comparison between the same HPI membranes operated under the same operation condition but the flux was different at 60% critical flux and 80% critical flux, respectively.
As a result, the higher the flux controlled, the faster the TMP jumped. Figure 4.10 shows the membrane after long-term sub-critical flux operation.
(a) (b)
(c) (d)
Figure 4.10 Flat-sheet membranes after long-term sub-critical fluxes operation (a) HPO membrane operated under 80% of critical flux in MBR-2
(b) HPI membrane operated under 80% of critical flux in MBR-2 (c) HPI membrane operated under 60% of critical flux in MBR-1 (d) HPI membrane operated under 60% of critical flux in MBR-2
45
4.3 Membrane performance under different operation condition
The change of MLSS in activated sludge and TOC in permeate were monitored every day to observe the stability of the system. If any trouble occurred in the system, a suitable adjustment should be needed for keeping the MBRs operating under given conditions. Figures 4.11~14 present a fluctuation of MLSS, TOC in permeate and TOC removal from the beginning to the end of operation. The operational conditions seem to be stable and meet the given operation condition shown in scope of study (Figure 1.1). The average MLSS, TOC in permeate and TOC removal were shown in Table 4.3, in which the average MLSS for MBR-1 was 7371 ± 287 mg/l, permeate TOC was 1.80 ± 0.51 mg/l and for TOC removal, it was approximately 98.88%. For MBR-2, all operational case are recorded in Table 4.3
Figure 4.11 Operational conditions of HPI membrane with 60% of critical flux operated in MBR-1
46
Figure 4.12 Operational conditions of HPO membrane with 80% of critical flux operated in MBR-2
Figure 4.13 Operational conditions of HPI membrane with 80% of critical flux operated in MBR-2
47
Figure 4.14 Operational conditions of HPI membrane with 60% of critical flux operated in MBR-2
Table 4.3 The average MLSS, TOC in permeate and TOC removal Membrane Bioreactor Percentage of
critical flux (%)
48
HPO with 80% critical flux in MBR-2 HPI with 80% critical flux in MBR-2 HPI with 60% critical flux in MBR-1 HPI with 60% critical flux in MBR-2 HPI membrane
4.4 Analysis of membrane fouling
4.4.1 Qualitative analysis of foulant4.4.1.1 Detection of EPS compositions by FTIR analysis
The FTIR spectra of membrane fouling occurred on HPI and HPO membrane operated under different operational conditions are illustrated in Figure 4.15. The functional groups of foulants were detected at some specific peaks for four cases as introduced above. The peak at wave number of 1035 cm-1 shows the presence of polysaccharides (Grube et al., 2006). While at wave number of two peaks near 1639 and 1531 cm-1, that are assigned to the existing of Amide-I and Amide-II represented to the presence of protein in membrane foulants (Wang et al., 2008; Kimura et al., 2005). N-H stretching and C-H stretching was observed at two peaks with the wave number of 3284 and 2920 cm-1, respectively (Zurasilam et al., 2006). As a result, these FTIR experiments have demonstrated to the presence of proteins and polysaccharides in the membrane foulants on membrane surface.
Figure 4.15 FTIR spectra of fouling on membrane surface
49
4.4.1.2 Protein composition of foulant
The EEM fluorescence spectra of membrane fouling in cake layer and inside of membrane pore operated by four imposed sub-critical fluxes were illustrated in Figures 4.16~19. EEM spectra show information about the fouling compositions by observing the distribution region of it. There are five regions in EEM spectra numbered from Region I to V as clearly introduced in Figure 3.9 and Table 3.5. In the results, it was also numbered for a straightforward assessment about the fouling compositions. All the results with the main peaks just located on Region I and IV. Region I indicate to the tyrosine-like aromatic protein (Chen et al, 2003; Wang et al, 2009), while Region IV refers to soluble microbial by-product-like protein. That meant protein and protein-like products always existed in the membrane foulants for all operational case.
Table 4.4 shows the peaks location and intensity of membrane fouling as protein.
Regarding to the fouling (was assumed as protein including tyrosine-like protein and soluble microbial protein) in cake layer, the fouling formed on HPI membrane surface operated with 60% critical flux in MBR-1 were higher than that in MBR-2 based on the peaks intensity. Proteins on HPI membrane operated with 80% of critical flux in MBR-2 presented the higher concentration of proteins than that of 60% critical flux and it also higher than proteins on HPI membrane
In case of membrane pore inside, fouling on HPI membrane operated with 60% of critical flux in MBR-1were lower than that in MBR-2. Operated with 80% of critical flux in MBR-2, the proteins in fouling formed on HPI membrane were higher than that operated with 60% of critical flux and also higher than the protein in fouling deposited on HPO membrane surface.
Looking back to see the Figure 4.9, HPO membrane operated under 80% of critical flux in MBR-2 was fouled quicker than HPI membrane illustrated by the quick jump of TMP. But in this section, the fouling (as protein) of HPO membrane presented the lower concentration not only in cake layer but also in membrane pore inside than HPI membrane.
That meant proteins didn’t play a role in the TMP increasing of HPO compared to HPI membrane.
In contrary, with the difference of imposed flux operation, HPI membrane with 80% of critical flux was fouled quicker than HPI membrane with 60% of critical flux
50
corresponding with the protein concentration in EEM test. The protein concentration of HPI with 80% of critical flux was higher than that of HPI with 60% of critical flux in two cases: in cake layer and inside of membrane pore.
In case of HPI membrane operated under 60% of critical flux but in different MBR, the results show that, the protein concentration in cake layer of membrane in MBR-1 was relatively higher than in MBR-2, but in membrane pore inside, it was conversely. This reveals that the proteins inside membrane pores effect to the quicker TMP increasing of HPI membrane in MBR-2.
51
Table 4.4 Peaks location and intensity of membrane fouling compositions
Region I Region IV
Membrane Bioreactor Percentage of critical flux (%)
Ex/Em Intensity Ex/Em Intensity
In cake
52
300.00 350.00 400.00 450.00 500.00
459.36
300.00 350.00 400.00 450.00 500.00
963.14
Figure 4.16 Fluorescent EEM of membrane fouling on HPI membrane operated under 60% critical flux in MBR-1. (a) In cake layer; (b) Inside of membrane pore
Emission (nm)
53
300.00 350.00 400.00 450.00 500.00
691.61
Figure 4.17 Fluorescent EEM of membrane fouling on HPI membrane operated under 80% critical flux in MBR-2. (a) In cake layer; (b) Inside of membrane pore
300 400 500
300.00 350.00 400.00 450.00 500.00
963.20
54
300.00 350.00 400.00 450.00 500.00
493.76
300.00 350.00 400.00 450.00 500.00
823.45
Figure 4.18 Fluorescent EEM of membrane fouling on HPO membrane operated under 80% critical flux in MBR-2. (a) In cake layer; (b) Inside of membrane pore
Emission (nm)
55
Excitation (nm) Excitation (nm)
300 400 500
300.00 350.00 400.00 450.00 500.00
666.10
Figure 4.19 Fluorescent EEM of membrane fouling on HPI membrane operated under 60% critical flux in MBR-2. (a) In cake layer; (b) Inside of membrane pore
300 400 500
300.00 350.00 400.00 450.00 500.00
423.98
56
MBR1 HPI 60%
MBR2 HPI 60%
MBR2 HPI 80%
MBR2 HPO 80%
Soluble polysaccharide concentration ( mg/l as glucose)
0 50 100 150 200 250 300 350
In cake layer
Inside of membrane pore
4.4.2 Characterization of fouling composition
Membrane fouling plays an important core role in MBR system effecting to wastewater treating efficiency. As introduced in Section 2.1.3, membrane fouling includes three basic types: adsorption, pore blocking and cake layer (Hong et al., 2005). In which, cake layer contributes 80% of resistance (Lee et al., 2001). Yet, cake layer compositions were concerned to compare the fouling behavior of four sub-critical fluxes operation.
Besides, the fouling in the inner membrane pores was also studied. The extent of this section was to characterize the main fouling compositions represented as EPS and SMP.
In which, EPS includes soluble EPS and bound EPS. SMP was considered as soluble EPS constituted by soluble polysaccharides and soluble proteins. Cell-bound polysaccharides and cell-bound proteins were identified as the compositions of bound EPS.
Figure 4.20 Soluble polysaccharides in cake layer and in membrane pore
Figure 4.20 shows the soluble polysaccharides in cake layer and inside of membrane pore of four operational cases in this study. That were HPI and HPO
Figure 4.20 shows the soluble polysaccharides in cake layer and inside of membrane pore of four operational cases in this study. That were HPI and HPO