Due to its low biodegradability, molasses wastewater is usually defined as high-strength industrial wastewater. The possibility and efficiency of a membrane bioreactor (MBR), containing lab-made membrane modules, in treating molasses containing wastewater was studied. The research was conducted in two phases, namely a first fed-batch operation period followed by a continuous MBR operation for the further duration of the experiment. The results showed that more than 80, 90 and 30% of the chemical oxygen demand, total nitrogen and color, respectively were removed. Also, there was a clear influence of the pore size. Less porous membranes showed a better retention. Scanning electron microscopy and fourier transform infrared analyses revealed no significant differences in organic constituents between the membrane cake layer and the activated sludge.
Membrane bioreactors (MBRs) are being increasingly recognized as an emerging and effective technology for the treatment of high-strength industrial wastewater and complex and recalcitrant compounds. The advantages related to MBR technology are numerous, namely: 1) a total solids retention even at very high concentrations of mixed liquor suspended solids (MLSS), 2) a small footprint, 3) a low excess sludge production, and 3) a considerably better effluent quality compared to conventional activated sludge systems (Visvanathan, 2000). However, the promising prospect of MBR technology is limited by membrane fouling, which leads to an unavoidable frequent membrane cleaning. The combination of an aggressive chemical cleaning and extensive aeration, in order to limit fouling formation, eventually shortens the membrane life time and thus contributes to the high operational cost of MBRs.
Get your grade
or your money back
using our Essay Writing Service!
Molasses containing wastewaters are commonly found in distillery processes of ethyl alcohol production and are characterized by very high levels of Chemical Oxygen Demand (COD), Biochemical Oxygen Demand (BOD) and a high COD/BOD ratio. This type of wastewater contains low and high molecular weight organic substances, inorganic substances and colorants such as melanoidin, phenolics, caramel and melanin (Satyawali and Balakrishnan, 2008a). The dark brown melanoidin appears to be the dominant colorant (2%) and consists of low and high molecular weight polymers and is therefore characterized by a relatively low biodegradability (Martins and Van Boekel, 1999; Guimaraes et al., 1999). By using a conventional anaerobic-aerobic effluent treatment process, only 6-7% of the melanoidins are degraded (Guimaraes et al., 1999; Choo and Lee, 1996; Kang et al., 2002; Zhang et al., 2006; Satyawali and Balakrishnan, 2008b). It is our hypothesis that a more advanced biological treatment occurring in MBRs may be able to enhance the better degradation of these compounds, which is supported by several reports. Choo and Lee (1996) and Chang et al. (2002) described the anaerobic treatment of alcohol-distillery wastewater in a thermophilic external MBR. Enhanced COD removal was achieved with the complete retention of biomass and a high COD removal of > 90%. (Kang et al., 2002). Zhang et al. (2006) reported about the aerobic treatment of simulated distillery wastewater in lower concentration (1000Â mg/LÂ COD). The mean COD and TN removal efficiencies were 94.7% and 84.4%, respectively. Extensive studies reported by Satyawali and Balakrishnan (2008b) reported about the treatment of distillery effluent wastewater containing molasses using an aerobic mesh filter-MBRs. Their results showed that about 41% of the COD was removed in a 245 days of operation. Low molecular weight compounds were successfully degraded, while the high molecular weight compounds comprising the colour imparting melanoidins remained unaffected. The filtration performance could be enhanced by adding powdered activated carbon as this increased the critical flux and prolonged the time between filter cleaning.
There are currently no published data on the treatment of molasses wastewater using an aerobic MBR. Using mesh filters instead of membranes could not achieve a complete retention of suspended solids (SS). On the other hand, by applying the membrane, a complete solid retention, the contact time of the activated sludge and pollutants is prolonged, which facilitates the removal of slowly biodegradable substances. In this study, the performance of a lab-scale aerobic MBR was investigated for treatment of diluted synthetic molasses containing wastewater. The tests were performed in a high-throughput (HT) lab scale MBR (HT-MBR), which is equipped with six lab-made membrane modules containing membranes with different pore sizes (3x2 duplicates). The biodegradation performance and the membrane filtration characteristics e.g., flux profile and membrane fouling as function of time were studied in detail.
Materials and Methods
Sludge inoculum and synthetic wastewater
The sludge, used to inoculate the lab-scale MBR, was obtained from a pilot-scale MBR treating molasses wastewater (Waterleau, Wespelaar, Belgium). The sludge was initially acclimatized using a fed-batch operation period until an MLSS level of 8 g/L was obtained. The feed solution, further referred to as synthetic wastewater, was prepared by diluting a molasses stock solution to a final concentration of 2.25 ml/L molasses. The molasses stock was stored at 4Â°C and a fresh feeding solution was prepared twice a week.
Always on Time
Marked to Standard
2.2 Membrane and module preparation
The used flat sheet membranes were prepared from polysulfone (PSF) (BASF-Ultrason)/N-Methyl-2-pyrrolidone (NMP) (ACROS) solutions (8, 12 and 16%) by phase inversion (Vankelecom et al., 2004). Each polymer concentration was prepared in duplicate. The polymer solutions were casted on a polypropylene support (Novatexx 2471, kindly supplied by Freudenberg, Germany). The characteristics of the membranes used in this study are presented in Table 1. The nominal pore size and thickness of the membranes were determined using scanning electron microscopy (SEM) (Philips SEM XL30 FEG with Adax dx-4i system).The SEM images of the membrane surface were analyzed using image processing software (ImageJ) to obtain the nominal pore size. To obtain the thickness, SEM images of the cross section of membrane were visually observed and measured. The clean water permeability (CWP) was measured in vacuum filtration mode and the critical flux (CF) was measured according to flux-step method described by Le Clech et al. (2003).
Table 1 Characteristics of the used flat sheet polymer membranes
PSF Membrane (%)
Pore size (Î¼m)
Clean water permeability (CWP) (L/m2/h)
2.3 Experimental Set-up
The schematic diagram of the lab-scale MBR is shown in Fig. 1. The HT-MBR system has a working volume of 18.6 liter and is equipped with six 0.016 m2 flat-sheet membrane modules, positioned vertically in the bioreactor. The filtration was performed at a fixed flux of 7.6 L/m2/h using a multichannel peristaltic pump (Watson Marlow) to ensure similar mixed liquor properties. The filtration cycle was controlled using an Intelli-Cycle timer (Hidro Masta, Australia) and was performed in an eight minutes permeation and two minutes relaxation cycle without back-flush. The aeration system was positioned underneath the membrane modules to provide a distributed air-flow rate of 0.6Â m3/h. The system was kept at a temperature of 21Â°C. A feed pump, steered by the level controller, was installed in order to maintain a steady biomass volume. The reactor operated at a hydraulic retention time (HRT) of 18-20 h and a mixed liquor suspended solid (MLSS) concentration of 8-12 g/L. This concentration range was assured by partially withdrawing sludge from the bioreactor. An additional air diffuser was installed to provide fine bubbles for an efficient oxygen transfer to keep the dissolved oxygen level acceptable. Each membrane module was cleaned on a weekly basis starting from the first membrane cleaning on the third day of operation. The foulant layer was scraped and collected for FT-IR spectroscopy to determine the fouling components. Physical cleaning was applied by flushing tap water to clean the membrane surface from any visible foulant. No chemical cleaning was applied throughout the experiment.
Fig. 1 Schematic diagram of the HT-MBR system.
2.3 Analytical methods
The bioactivity of the activated sludge was monitored by measuring the dissolved oxygen (DO), pH, temperature, floc and MLSS on a daily basis. The dissolved oxygen, and temperature were measured using HACH HG30d equipment, the pH was measured by means of a WTW pH330i, the floc structure was analyzed using inverted light microscopy (OLYMPUS CKX-41), and the MLSS concentration was determined according to a standard method (APHA, 1992). The COD, color, turbidity and total nitrogen (TN) in the feed and in each of the six permeates were measured once a week according to the standard method (APHA, 1992).
The evolution of the TMP was monitored on a daily basis at the end of permeation and relaxation period, by means of a manual vacuum gauge reading. More detailed TMP observations were done each minute during a first filtration cycle following every physical cleaning. A part of the sludge was analyzed to compare the functional groups among the cake layer and mixed liquor. The foulant cake layer and the sludge were dried at room temperature. Each sample was mixed with potassium bromide 10:90 w/w (Fluka), powdered and mechanically pressured to form the thin film for FT-IR (NICOLET 6700). At the end of the experiment, the membrane microstructure of both the fouled and cleaned membrane was observed using SEM. The membranes were coated with gold prior to analysis.
Results and Discussion
Acclimatization and sludge characteristic
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
During the acclimatization period the sludge inoculum was fed using fed-batch regime. After two months, the MLSS concentration increased from 2 till 8 g/L. The bioreactor was fed using synthetic molasses wastewater at a food to microorganism (F/M) ratio of 0.25. Consequently, the feed concentration increase followed the increase in MLSS concentration. At the end of fed-batch operation the molasses concentration was 2.25 ml/L and was kept stable for the whole continuous operation. The tests of COD removal during the fed-batch operation showed a variation of 50-75%. The immediate COD removal of the molasses wastewater was expected as the activated sludge inoculum was taken from the pilot MBR that treat similar wastewater only differing in the molasses concentration.
Fig. 2. Activated sludge micrographs taken at 100x magnification: (a) Sludge inoculum, (b) sludge after fed-batch operation, (c and d) sludge during the continues operation.
The microscope analyses of the sludge flocs (Fig. 2) showed that the sludge inoculum was mainly contained a high concentration of filamentous bacteria in addition to the numerous small flocs (<5 Î¼m). However, after fed-batch operation, the overall condition of the sludge increased and the concentration of filamentous bacteria decreased, while the floc sizes significantly increased (60-100Î¼m). During the rest of the experiment, the floc size of the mixed liquor measured 15-50Âµm. The reduction of the floc size was also found in other reports of submerged MBRs and is attributed to the hydrodynamic stress (Zhang et al., 1997; Song et al., 2003).
Operational stability and Biological Performance
The monitored operational MBR parameters are shown in Fig. 3. There appeared to be no significant variation in pH (8.1-8.8), temperature (17.4-22.8Â°C) and DO (6.4-8.5 mg/L). The continuous operation period started at an MLSS concentration of 8 g/L (= day zero). This concentration increased to a level of 11-12 g/L after day 20 and remained stable for the further duration of the experiment (day 20-38), which was an indication of the low biomass yield in an MBR (Farizoglu et al., 2007; Khor et al., 2006). The slight difference in the temperature during the operation was due to the significantly difference in ambient air that was pumped to the system.
Fig. 3. Profile of the operational parameters
Table 2. Summary of the biological performance of the lab-scale MBR
80.9 + 5.1
Color in mg/L(Pt-Co)**
*) The samples tested were the accumulated permeate of each membrane
**) The value given is the mean + standard deviation value of the different membrane permeates
During operation, the effluent quality was periodically monitored and the results are summarized in Table 2. The monitoring was divided into three periods of 0-23, 24-30 and 31-38 days. The result also shows that the average removal of COD during the operation ranged between 80 and 85%. The reported COD removal in MBR studies are normally very high (>99%), especially in case of municipal wastewater, containing a low influent COD (<600 mg/L) (Visvanathan et al., 2000). The relatively low COD removal in this study could possibly be attributed to the high feed concentration and the presence of difficult biodegradable, recalcitrant organics and growth inhibiting substances in the molasses (Pant and Adholeya, 2000; Valderrama etal., 2002). An even lower biodegradation (<41%) was found by Satyawali and Balakrishnan (2009) for the treatment of molasses based distillery wastewater. The high TN removal (> 90%) indicated that the growth of slow-growing microorganisms, such as nitrifying bacteria, was facilitated by long SRT and complete retention of the biomass (Wang et al., 2005). An almost complete SS retention was obtained during MBR operation as expected. It was obvious that the formation of a cake layer on the membrane surface controlled the SS rejection by forming a dynamic membrane (Visvanathan et al., 2000; He et al., 2005; Meng et al., 2009). The active layer is formed on the membrane surface by deposition of substances present in the mixed liquor.
Membrane Filtration and Fouling
The TMP profile at a typical first filtration cycle after a physical cleaning is shown in Fig. 4 for each of the 3 membrane modules. It is clear that even after cleaning, there is a significant TMP increase for PSF 16 and 12% membranes during the first filtration cycle. As shown in Table 1, PSF 16% membranes had the smallest pores and lowest CWP, followed by PSF 12 and 8%. The high TMP obtained in this experiment can be explained by the high membrane resistance and corresponding operational flux, which causes fast blockage of the pores by mixed liquor components due to polarization concentration. To compensate for the membrane resistance, a significant pressure increase is required, especially for low CWP membranes. When running the membrane with different porosity, the lower porosity membrane has higher permeate flux that leads to higher convective flow through each membrane pore. It leads to a higher foulant concentration at the membrane surface due to concentration polarization and higher amounts of foulants get attached onto the membrane and pore wall in shorter periods of time. This causes a fast cake/foulant layer formation and narrows the effective pore size of the membrane (Zhang et al., 2006). A fast TMP build-up on the PSF 16% membrane throughout the operation indicated the formation of membrane fouling.
Fig. 4. The typical TMP profile during the first cycle of operation after a physical cleaning.
The TMP evolution of the membranes during long-term operation is shown in Fig 5. The highest TMP increase was shown by the 16% PSF membrane (380-800 mbar) followed by 12 (150-600 mbar) and 8% (0-10 mbar). The presented data are the mean + standard deviation of each identical membrane and were taken after the eight minutes of filtration time, immediately before relaxation started. From the long term observations, it was clear that each membrane pair resulted in a different TMP evolution pattern. The immediate and exponential TMP jump was shown for 16% PSF membranes initially 380 mbar to about 800 mbar in three days of operation. This initial high TMP and TMP jump pattern was also observed after each physical cleaning followed by a nearly constant TMP until the next cleaning. The TMP jump is most probably caused by the fact that the operational flux was below the critical flux of the membrane which lead immediate local critical flux to be achieved. On low surface porosity of the membrane, in the short period of time, the open surface is reduced due to pore blocking. It makes the available open pore to operate above the set flux and may achieve that the local critical flux in a very short period of time, which causes the abrupt change in TMP (Ognier et al., 2004). The constant TMP occurred due to pump limitation, the stagnant TMP was compensated by a decrease of filtration flux.
Fig. 5. The TMP profile during operation
For 12% PSF membranes, the TMP increased in two phases, namely a first slow increase followed by a TMP jump before physical cleaning. During the slow TMP increase: different phenomenon's might have occurred like pore blocking, biopolymer deposition, bacterial attachment and biofilm formation probably occurred (Zhang et al., 2006). During filtration, the effective filtration area slowly decreased due to irreversible membrane fouling. This promoted the increase of the local flux on effective pore and eventually reaching the critical flux that lead the TMP jump. The TMP jump was characterized by a high hydraulic resistance from strong deposition of suspended matter on the membrane surface. This mechanism was also found by Ognier et al., (2004) and Cho and Fane (2002).
This explanation is supported by the observation of a foulant layer on the membrane surface before physical cleaning. An uneven distribution of the fouling layer was found which qualitatively showed the tendency to cover the membrane edges near the potting wall, as shown in Fig. 6. This foulant layer was very compact and completely blocked the covered surface, leading to the reduction of the effective filtration area. However, this non-distributed foulant is very difficult to be prevented by considering the difficulty of providing the similar surface shear (Zhang et al, 2006).
Fig. 6. Membrane module: (a) Uneven foulant distribution on the membrane surface, (b) Cleaned membrane
A high SS removal was surprisingly found for both damaged membranes (PSF 8%) even though a very small part of the membrane became detached from its support. It was decided to keep on using the filtration membrane to maintain the HRT of the reactor. During all time of operation, very small TMP increase (10 mbar) was detected on these membranes. The small detachment on membrane active layer was not significant to affect the filtration performance. High SS rejection was also found in both membranes representing the dynamic membrane formation on the detached part. The low TMP increment can be explained due to very low operational flux. The one week operation was relatively short and was not enough to obtain significant and irreversible fouling that could be detected from the TMP evolution. For comparison J. Zhang et al., (2009) found no TMP jump for flux of 10 L/m2/h for 280 hours of operation. Ognier et al., (2004), Cho and Fane (2002), and Yu et al., (2003) found the vey slow TMP increase and TMP jump occurring after 550 relatively, 450, and 300 hours of operation.
The obtained COD removals are presented in Fig. 7. The samples were taken from accumulated permeates from each membrane. Fig. 7 also shows the different rejection for different membranes. The rejection showed a reverse trend to permeability. The membranes had the capability to retain not only the suspended solid and biomass but also soluble substance from the mixed liquor. Rejection of solute could be attributed to two reasons: (1) by biofilm developed on the membrane surface, and (2) to the dynamic layer formed by the foulants or biomasses that act as a filter (Choi and Ng, 2008). In this case, the biodegradation of biofilm was not dominant. The significant difference on solute rejection showed even on the initial stage of filtration before the biofilm exist. The dynamic layer formation that reduces the effective pore size to form the cake later is more obvious.
Fig. 7. Color and COD removal of membrane bioreactor
The further analysis of color intensity of feed and permeates is given in Fig 8. The significant differences in permeate color intensity fits with the COD rejection pattern. This indicated that the rejected solute may originate from non-degraded colorant substances. The colorant substances in molasses have a molecular weight (MW) distribution between 5 and 40 kDa. When comparing to the nominal pore size of the membranes, these substances much lower size. This confirms the formation of a dynamic layer which in further extent enables the membrane to reject high MW substances associated to the cake layer formation as discussed earlier.
Fig. 8. Color intensity of feed and permeates
Membrane fouling and autopsy
Every week during operation, (except for the first operation, 4 days), a physical cleaning of each membrane was performed by flushing the membrane surface with tap water. The foulant cake layer was scraped and used for autopsy. In order to examine the membrane foulant formed on the membrane surface, the TMP recovery was calculated. It was assumed that only the cake layer was removed since no back flushing was applied. Among most fouling analysis, the cake layer is the main contributor of membrane fouling with more than 80% of the total resistance (Park et al., 2005; Hwang et al., 2007). The autopsy of cake layer becomes very interesting because it mainly contribute for membrane fouling.
220.127.116.11 FT-IR Analysis
FT-IR was used to characterize the major organic foulants (Maruyama, 2001). The spectra of activated sludge and foulant from each of the six different membranes are presented in Fig. 9. The spectra showed a broad adsorption peak at 3419 cm-1, which is attributed to the stretching of O-H bonds in hydroxyl functional groups, and peaks at 2932 and 2880,which is due to the stretching of C-H bonds representing the alkyl group (Kumar et al., 2006 and Wang et al., 2008). Two additional peaks (1666Â cmâˆ’1 and 1564Â cmâˆ’1) were also observed in the spectra and are unique to the protein secondary structure: Amide I, C=O stretching and Amide II, mostly N-H bending. It indicates that proteins were one of the foulant components. The peaks at 1252Â cmâˆ’1 and 1443Â cmâˆ’1 might be due to the presence of C-O (ester) bonds and N-N=O group, respectively. In addition, a broad peak at 1091Â cmâˆ’1 exhibits the character of carbohydrates or carbohydrates-like substances (Croue et al., 2003). The very broad and small peaks around 400 to 800 cm-1 could hardly be characterized. From the FT-IR spectra, it was clear that the membrane foulant not only consisted of EPS (proteins, polysaccharides, etc.) but also of other organic substances. The identical spectra are clearly shown when comparing the between the different membranes and to the activated sludge. It means that the major constituent of foulant was from the deposited activated sludge into the surface of membrane. Identical spectra were also reported by Z. Wang et al., (2009) who analyzed the membrane foulant and EPS and Zhou et al., (2007) who analyzed the activated sludge.
Fig. 9. FT-IR spectra of activated sludge and membrane foulant for six membranes.
18.104.22.168 SEM Analysis
SEM photographs of freshly prepared, fouled and physically cleaned membranes are given in Fig. 10. The active layer in contact with the mixed liquor was observed a "gel"-like layer. The physical structure of the fouling layer appeared to be compact and sticking to the membrane surface. Also, it was very difficult to remove the compacted gel-layer by routine aeration. Consequently, it leads to the increase of TMP and caused severe membrane fouling.
Fig. 10. Typical SEM photographs of (A) new membrane surface, (B) fouled membrane surface, (C) physical cleaned membrane surface, (D) cross section of new membrane, (E) cross section of cleaned membrane
This study describes the treatment of molasses wastewater in a lab-scale MBR using lab-made phase inversed membranes and diluted molasses as feed. The following conclusions can be drawn from this study:
The rapid biomass growth and fast adaptation was shown during the fed-batch operation period. However, due to accumulation of non degraded melanoidins the mixed liquor solution became more brown in time.
The continuous MBR operation occurred under a stable temperature, ph and DO value and more than 80.9%, 90.1%, and 30% of COD, TN and color were removed.
Each PSF membrane concentration (8, 12 and 16%) was characterized by its own TMP profile. This could be explained by the different membrane properties (critical flux) and the local flux concept, which is supported by the uneven fouling distribution.
A higher membrane-polymer concentration showed an improved melanoidins rejection. This indicates that the membrane interception for high molecular substances in mixed liquor. Taking into account these results, we suggest the use of smaller pore size membranes to retain high molecular weight substances and provide an improved and prolonged contact time with the activated sludge and facilitates their efficient removal.
Fouling autopsy showed no remarkable differences between the organic constituents among the membranes cake layer, and between the cake layer and activated sludge. According to our opinion, the overall part of the cake layer originated from compacted deposited activated sludge that eventually initiates the fouling.