DOI:
10.1039/C6RA23188H
(Paper)
RSC Adv., 2016,
6, 111182-111189
Application of response surface methodology for investigation of membrane fouling behaviours in microalgal membrane bioreactor: the effect of aeration rate and biomass concentration
Received
17th September 2016
, Accepted 17th November 2016
First published on 17th November 2016
Abstract
This study was performed to investigate membrane fouling phenomena and to optimize fouling parameters in a submerged membrane bioreactor. An ideal experimental design was carried out based on central composite design (CCD) with response surface methodology (RSM). This RSM was used to calculate the effects of microalgae concentration, aeration rate and their interaction with membrane fouling and permeate flux. Based on the analysis of variance (ANOVA), the permeate flux model proved to be of very good fit with a very low probability value (<0.0001). High permeate flux occurred at low microalgae concentrations and optimum aeration rate (2.5 L min−1). Minimum total resistance was 2.65 × 10−11 m−1 at optimum aeration rate and microalgae concentration of 0.65 g L−1. Furthermore, a high microalgae concentration released larger amounts of extra-cellular polysaccharides (EPS), resulting in severe fouling.
1. Introduction
Microalgae are single cell organisms which are a promising source for producing a variety of products and biofuels. As an example, microalgae can be used in the pharmaceutical industry, beauty products and cosmetics, dietary supplements, removal of heavy metals, biological wastewater treatment, eliminating biological contaminants and the production of biofuels.1–4 Selection of suitable harvesting method of microalgae considering their low concentration and density (almost equal to density of water) in the culture medium and their small cellular size are two important reasons why nowadays, little attention is paid to microalgae despite their competence in production of various products.5,6 For proper selection of the harvesting technique, it is required to gain high biomass recovery at a modest cost.7 There are various harvesting methods developed for algae biomass collection including centrifugation, flocculation and filtration.8 Centrifugation is usually done in a high gravity force, resulting in high energy consumption, and high tension which can damage microalgae cells.9 Flocculation is very sensitive to pH. Moreover, plenty of coagulator is required, where this large amount can be toxic for algae, and also create a large amount of deposits.8 All these techniques are generally costly for large-scale algae cultivation and harvesting.5
Membrane technology has advantages over conventional cultivation and biomass separation methods, since it does not need any additional chemicals and it will not cause significant damage to cell structure. Increasingly, membrane filtration processes can achieve more complete recovery of algae biomass. The most important benefits of this method, when compared to conventional dewatering methods, is that membrane filtration is a relatively lower-energy consuming method for algae harvesting, but it faces issues of membrane fouling.7,10–12 Membrane filtration is done by two distinct methods in a membrane bioreactor (MBR): cross flow filtration or submerged membrane filtration. Most studies on microalgae harvesting emphasize ultra- or microfiltration with cross flow membrane structure. Cross flow MBRs are highly energy consuming because of high cross-flow velocity (CFV) and high shear stress, applied to the membrane surfaces. In cross flow MBRs, CFV variation contributes to changing the shear stress and fouling.13 Increased CFV has two opposite effects on fouling: microalgae is prevented to deposit on membrane surfaces because of high CFV, thus reducing fouling (positive impact), but the dramatic CFV elevation results in breaking microalgae cells and releasing extra-cellular polysaccharides (EPS), thereby intensifying fouling (negative impact).14,15 In submerged MBRs, the lack of any cross flow stream contributes to lower strain. Hence, in this structure, the required energy for filtration is less than cross flow MBRs. So this method is more efficient for filtration.16,17 In more recent research on the hydrodynamics of membrane fouling in submerged MBRs, the focus is on the aeration rate and membrane efficiency.18 In submerged MBRS, stress is usually applied to the system by air bubbles which provides oxygen required for the biomass and can disperse solid particles, with less fouling occurring on the membrane surface.19 Furthermore, aeration can be used to create shear stress on the membrane surface without the need for recirculation pump.18 Aeration is an effective parameter on biomass flow particle size, membrane filtration and fouling control. High aeration removes adhesion of biomass to the surfaces, but on the other hand, aeration causes failure in biomass flocs which can release high amounts of EPS. EPS can accumulate in membrane surfaces and increase membrane fouling. In this condition, colloid particles and EPS can be considered the most important factors in membrane fouling.13,20 Biomass concentration is the other parameter that has a significant effect on EPS and membrane fouling, where fouling is developed by increasing the biomass concentration.16,21,22
Most previous experiments on fouling phenomenon have adopted the common experimental method in which, among all the consider parameters, one parameter is altered while other parameters are remained constant. This method leads to conductance of numerous experiments; further, the concurrent effect of several parameters on the fouling is neglected, thereby lowering the operation efficiency, whereas the experimental design one can observe the concurrent effect of several parameters on the membrane fouling. For this purpose, it is possible to use response surface methodology (RSM). The RSM can be used to design statistical experiments, develop a mathematical model to predict out-put response and check its adequacy and eventually to optimize the operational conditions using a valid model.23,24
This study was conducted to investigate the parameters influencing membrane fouling to reduce fouling. Aeration rate and microalgae biomass concentration, two factors influencing fouling, were investigated to achieve minimum membrane fouling. RSM was used across 3 levels to develop a model and to optimize the effective parameters on fouling and simultaneously observe the interaction between operational factors. In addition, different kinds of membrane resistance, cake layer resistance and amounts of EPS were also determined and discussed in detail.
2. Materials and methods
2.1. Growth of algal cultures
Green pure Chlorella vulgaris microalgae were prepared from Shahid Beheshti University in Tehran and were grown in standard growth medium culture (BG 11), which contains NaNO3 1.5 g, K2HPO4·3H2O 0.04 g, KH2PO4·3H2O 0.2 g, EDTA 0.0005 g, Fe ammonium citrate 0.005 g, citric acid 0.005 g, Na2CO3 0.02 g and 1 mL of trace metal solution per litre, pH 7.3. The trace metal solution contains H3BO3 2.85 g, MnCl2·4H2O 1.8 g, ZnSO4·7H2O 0.02 g, CuSO4·5H2O 0.08 g, CoCl2·6H2O 0.08 g and Na2MoO4·2H2O 0.05 g per litre.25 The experimental conditions were remained constant at pH 7.3 and temperature 25 °C in all runs. All experimental means were sterilized individually in autoclave at 125 °C.
2.2. Experimental setup
The filtration performance tests were performed in a lab-scale vacuum filtration set-up, illustrated in Fig. 1. A vacuum filtration unit applying for the experiments was joined with a vacuum gauge and a needle valve adjusting the pressure. The microfiltration membrane used in this study was mixed-cellulose ester (MCE) with 0.22 μm nominal pore size and 0.0018 m2 effective membrane area. The submerged membrane photobioreactor was made up of a Plexiglass cylinder, with an effective volume of 2 litre. The Plexiglass module was fixed and attached by two glass bases to the wall of chamber. The light intensity was 2000 lx. In each experiment, microalgae medium was transferred from elementary cultivated sources into the feed tank in 3 distinct concentrations (0.3, 0.065 and 1 g L−1). Air was blown across varied aeration rates (0, 2.5 and 5 L min−1) into the chamber using spargers which were located on the bottom of container. The distance between the holes of sparger was equal and the developed air bubbles were controlled uniformly and homogenously. The air bubbles were identical and controlled to be uniform during the runs. In addition, to avoid the sedimentation of biomass in any run, especially the runs without aeration rate, magnetic stirrers was used at 300 rpm.
 |
| Fig. 1 A schematic diagram of experimental set-up. | |
2.3. Measurement of permeate flux
In this study, transmembrane pressure (TMP) was set at 0.5 bar and the permeate volume was measured every 30 minutes in constant TMP. Permeate flux was calculated by eqn (1) |
 | (1) |
where V is a permeate volume (L); t is a time (h); A is an effective filtration area (m2).16
2.4. Measurement of resistance
Membrane resistance can be measured quantitatively using the resistance – in series model based on Darcy's law as follows: |
 | (2) |
where Rt is the total resistance, ΔP is the TMP (N m−2), η is the permeate viscosity (Pa s), J is the permeate flux (L m−2 h−1). The total filtration resistance is the sum of the membrane resistance (Rm), the cake resistance from the cake layer formed on the membrane surface (Rc) and the irreversible resistance caused by pore blockage (Rir).26
In the filtration experiment, every new membrane was first filtered by deionized (DI) water in operational conditions to find membrane resistance (Rm).16 Then the microalgal suspension was filtered with permeate flux and total resistance being calculated according to eqn (1) and (2) respectively. After recording permeate flux, membrane surfaces were cleaned twice by DI water and a sponge to remove cake layer deposited on membrane surfaces.27,28 In the next step, algal suspension was replaced by DI water in MBR under the same operating conditions to get irreversible resistance which is not removed easily and required chemical cleaning.29 All the resistances can be calculated from experimental data using following equations:30,31
|
 | (3) |
|
 | (4) |
where
Jw0 is the initial water flux and
Jw1 is the final water flux after removing the cake layer. All resistances were measured in m
−1.
2.5. Analytical methods
2.5.1. Determination of cell density of C. vulgaris. In order to have considered concentration for any new experiment, sample of microalgal medium was collected to measure the concentration of Chlorella vulgaris in algal medium using spectrophotometer (UNIC 2100) at 580 nm wavelength. The absorbance at 580 nm was calibrated by the dry cell weight (DCW) method.32 Then, the weight of C. vulgaris was obtained from the calibration curve. |
Microalgal cell dry weight (g L−1) = 0.653 × OD580 R2 = 0.99
| (6) |
2.5.2. EPS extraction and analysis. EPS classified to soluble microbial products (SMP) and EPS which extracted from cell floc (eEPS).21 Several methods have been applied to extract EPS from culture include ammonium hydroxide extraction, sodium hydroxide extraction, ethylene diamine tetra acetic acid (EDTA) extraction, ultrasonication, high speed centrifugation, extracting by boiling or autoclaving and formaldehyde and sodium hydroxide extraction.26 In the present study, EPS extraction method was heating which was performed by Morgan and et al.33 Microalgae suspension was collected to analyzed protein extracellular polymeric substance and polysaccharide extracellular polymeric substance based on the method of Lowry et al. and Dubois et al., respectively using UV/VIS spectrophotometer (WTW, SpectroFlex 6600, Germany).34,35 Several steps were done to calculate eEPS and SMP as described on the following. A microalgae suspension was first mixed using magnet stirrer at ambient temperature for 5 minutes then dewatered by centrifuging in a 10 mL test tube at 5000 rpm for 5 min (Hettich, Universal 320, Germany). After, microalgae suspension was filtered throughout the 1.2 μm filter paper to determine SMP. Removed microalgae on the filter were suspended in DI water and were shaken for 1 min. Then the solution was heated at 80 °C for 10 min. After heating, the solution was centrifuged at 7000 rpm for 10 min. Finally the eEPS sample was provided after filtration through 1.2 μm filter paper.
2.6. Experimental design and statistical analysis
Experimental design of the process for maximize permeate flux was employed using RSM, which is a useful mathematical-statistical method to design experiments and to optimize independent variables in chemical reaction and industrial process.24 In this study, RSM was used to assess the relationship between response (permeate flux) and variables, as well as, to optimize variable conditions to predict the best value of response. For this aim, central composite design (CCD), which is ideal for sequential experimentation, was carried out. CCD confirms plenty amount of information to check the lack of fit when sufficient number of experimental value exists.14 RSM and CCD were generated with the help of Design Expert software (version 0.0.7). Microalgae concentration and aeration rate were the significant independent variables in this study which were varied over tree levels between −1 to 1 at the determined ranges. Table 1 provides the independent variable and their variation levels in actual and coded values. The total numbers of experiments for the two factors with four replications to enhance error were obtained as 13. According to the CCD, a full quadratic approximation, which can be used for developing a second-order response surface models, is written in general forms as follows: |
 | (7) |
where y is the response; xi and xj are the variable; β0 is constant coefficient; βj, βjj and βij are the interaction coefficients of linear, quadratic and second-order terms, respectively; k is the number of studied factors; and ei is the error. Analysis of variance (ANOVA) with the 95% confidence level was used to evaluate the interaction between the effective variable and the response. The modality of the fit of the polynomial model was represented by the coefficient of determination (R2) and the adjusted R2 (Radj2).36
Table 1 Area of experimentation in actual and coded values of factors
Factor |
Actual factor name |
Units |
Level in coded and actual values |
Low actual |
High actual |
Low coded |
High coded |
+α |
A |
Aeration rate |
L min−1 |
0 |
2.5 |
−1 |
1 |
|
B |
Microalgae biomass concentration |
mg L−1 |
0.3 |
0.65 |
−1 |
1 |
|
3. Result and discussion
3.1. The effects of hydrodynamic condition and biomass concentration
The permeate flux was measured as a function of time until it remained constant for 30 min at constant TMP of 0.5 bar. Fig. 2 depicts permeate flux vs. time for all 13 runs which were done in different conditions of aeration rate and microalgae concentration. As can be seen, the permeate flux followed the same trend across all runs; it plunged dramatically within 200 min due to particle and collide deposition on membrane surfaces or inside membrane pores, resulting in membrane fouling and reducing membrane performance.5,13,14,18 Afterwards permeate flux had a gradual decrease because of cake layer formation on the membrane surfaces. Cake layer can act as a pre-filter substrate which prevented the deposition of particles into the pores and severe fouling.13,37 Having passed the time around 1400 min, permeate flux remained constant for 30 minutes before stopping the experiment. Such a process for changes in the membrane flux has also been reported in many studies; in this study, reduction of permeate flux and achievement of constant flux occurred in a longer time in comparison with previous studies owing to using lower pressure or difference in the membrane pore size.14,15,37
 |
| Fig. 2 Permeate flux decline versus time of 13 runs. | |
3.2. Predicted model using RSM
Permeate flux is usually evaluated as the desired response in modelling of membrane filtration used in calculation of other parameters such as different kinds of resistance and fouling. In this study, permeate flux, which was stabilized for 30 minutes in each run, was labelled as a response flux. Table 2 shows the all 13 runs of the CCD experimental design and the experimental value of the response flux. Student t-test was used to calculate the significance of regression coefficients. Insignificant coefficients were omitted to achieve a more suitable regression model using back-ward elimination procedure. Eqn (8) indicates the final reduced regression model in terms of coded factors for the permeate flux: |
J = 19.15 + 2.69A − 0.79B + 1.23AB − 3.14A2
| (8) |
Table 2 Design of experiments and output responses based on actual values
Run |
Aeration rate (L min−1) |
Biomass concentration (mg L−1) |
Flux (L m−2 h−1) |
1 |
2.5 |
0.65 |
18 |
2 |
0 |
1 |
11.16 |
3 |
2.5 |
0.65 |
19.487 |
4 |
5 |
0.3 |
18 |
5 |
0 |
0.65 |
13.8 |
6 |
2.5 |
0.3 |
20 |
7 |
0 |
0.3 |
15 |
8 |
2.5 |
0.65 |
18.02 |
9 |
5 |
0.65 |
19 |
10 |
2.5 |
0.65 |
20.5 |
11 |
5 |
1 |
19.1 |
12 |
2.5 |
1 |
18 |
13 |
2.5 |
0.65 |
20.04 |
The ANOVA regression is shown in Table 3. Montgomery studies were used for calculation of the statistical estimators.38 As can be seen, the response surface model for permeate flux and all terms in ANOVA analysis was significant on account of low probability values in reduced regression model. F-Value of 28.44 and p-value less than 0.0001 stated that the reduced quadratic model was valid. In addition, there was only a 0.01% chance that a model F-value could have occurred due to the noise. It was accepted that if adequate precision rate was more than 4, the predicted model was correct. In this modelling, this ratio was equal to 16.115 indicating an adequate signal for the models to be used to navigate design space. Moreover, it showed that the lack of fit of dates was not significant, revealing that the indicated model terms were significant for the considered response.39 Joglekar and May found that R2 should be a minimum of 0.8 to fit a proper model.40 In this study, R2 was +0.93; meaning that the model was fit and it explained more than 99% of the data deviation. In addition, adjusted and predicted determination coefficients (Radj2 = 0.90 and Rpre2 = 0.87) were in a good match with R2. According to p-value less than 0.0001 for aeration rate and microalgae biomass concentration, it can be concluded that both factors were significant parameters on response flux variation. The parameter influence on response flux could be arranged as A2 < A < B < B2. Hence, aeration rate had the most considerable effect on the process. Eqn (9) is rewritten below in terms of the actual factors.
|
Flux = 17.081 + 2.670 (aeration rate) − 5.785 (biomass concentration) + 1.411 (biomass concentration) × (aeration rate) − 0.502 (aeration rate)2
| (9) |
Table 3 ANOVA analysis for the regression model
Source |
Sum of squares |
df |
Mean square |
F value |
p-Value Prob. > F |
|
Model |
85.11 |
4 |
21.28 |
28.44 |
<0.0001 |
Significant |
A-air |
43.42 |
1 |
43.42 |
58.03 |
<0.0001 |
|
B − C |
3.74 |
1 |
3.74 |
5 |
0.0557 |
|
AB |
6.1 |
1 |
6.1 |
8.15 |
0.0213 |
|
A2 |
31.85 |
1 |
31.85 |
42.56 |
0.0002 |
|
Residual |
5.99 |
8 |
0.75 |
|
|
|
Lack of fit |
0.68 |
4 |
0.17 |
0.13 |
0.9647 |
Not significant |
Pure error |
5.31 |
4 |
1.33 |
|
|
|
Cor total |
91.09 |
12 |
|
|
|
|
This actual model could be easily used for graphical representation and analysis. It is usually important that the actual data are fitted in the predicted model. The adequacy of the model was evaluated using diagnostic plot, such as normal probability plot of studentised residuals and a plot of predicted versus actual values. Fig. 3a–c illustrates normal probability plots of the studentised residual for response flux. These plots indicate that if residual points are located on a straight line, they will witness a normal distribution. As can be seen in the Fig. 3b, there are some scattered data in the normal distribution. While a little scattering in the data were expected, it can be presumed that the data are normally distributed. Hence, it strengthened the adequacy of the least-square fit. According to Fig. 3d, the predicted values obtained from the presented model match well with the real values. In addition, as can be seen in Fig. 3b and c, the trend of residual plots versus the run number and predicted response did not follow a specific pattern and the residual scattered randomly. It can be deduced that the developed model for prediction and optimization of permeate flux has been suitable for Chlorella vulgaris microalgae microfiltration.41
 |
| Fig. 3 Residual plots for the CCD design (a) normal plots of residuals (b) residuals vs. run (c) residuals vs. predicted and plot of predicted vs. actual values (d). | |
Fig. 4 demonstrates the three dimensional surface response as a function of the microalgal concentration and aeration rate. It also indicates the major effects and the interaction between the investigated parameters. As can be seen, maximum response flux was 20.05 L m−2 h−1 at a biomass concentration of 0.65 g L−1 and aeration rate of 2.5 L min−1. Minimum response flux was 11.16 L m−2 h−1 at 1 g L−1 microalgae biomass concentration and without aeration rate. Growth of aeration rate to its optimum value and reduction of biomass concentration result in rising permeate flux and decreasing flux reduction efficiency. Moreover, this figure shows that the effect of aeration rate on permeate flux at low microalgae concentrations was stronger than at high biomass concentrations. In other words, the effects of microalgae biomass concentration on fouling were much higher than aeration rate effects even in the optimum aeration rate. High aeration rate created more tension in the membrane surface; thus having microalgae cells adhere less to the membrane surfaces. Moreover, high aeration rate made microalgae suspend in the MBR. Hence, microalgae cells were unable to sediment on the membrane surface and no cake layer was formed.
 |
| Fig. 4 3D response surface plot indicate the effects of aeration rate and microalgae concentration on fouling. | |
3.3. Study of resistance across different microalgae biomass concentrations
Membrane resistance was calculated to determine the behaviour of membrane fouling using eqn (1)–(5). Fig. 5 illustrates Rt, Rm, Rir and Rc at different aeration rates and biomass concentrations. Total resistance was different for each run, but in all of them cake layer resistance accounted for more than 94% of total resistance, whereas irreversible resistance had a smaller share of total resistances (less than 5%). As can be seen, high total resistance value occurred at various microalgae concentrations when there was no aeration (3.93 × 1011 m−1 at microalgae concentration of 0.65 g L−1 and 4.86 × 1011 m−1 at microalgae concentration of 1 g L−1). In MBRs without aeration, the microalgae cells were diffuse into the pores of the membrane quickly and then form a cake layer on the membrane surface without any obstacle, which led to increased irreversible resistance and cake layer resistance.17 In these conditions, the values of irreversible resistance (9.75 × 1010 and 1.46 × 1011 m−1) were much higher in comparison with MBRs with aeration rate. Also, Cui et al. reported the same result about the effects of aeration rate on sludge submerged MBR as this study's and emphasized the superiority of MBRs with aeration rate resulting in diminished reversible and irreversible resistance.42
 |
| Fig. 5 Various filtration resistances after 1400 minutes of filtration at different aeration rate and microalgae concentration (1 to 13: number of runs). | |
Rising aeration rate from 0 L min−1 to 2.5 L min−1 decreased fouling and thus increased permeate flux. Aeration rate elevated MBR turbulency and mass transfer coefficients. By increasing aeration rate, accumulated cells were distributed in the algal mixture. So microalgae accumulation and cake layer thickness on the membrane surfaces was reduced and the cake resistance was decreased even though irreversible resistance remained almost constant. It took a time to be formed a new layer of cake with the same thickness of cake layer without aeration on the membrane surface. But, during this time, the thin layer on the membrane surface formed to prevent the penetration of microalgae cells in membrane pores so total resistance decreased and permeate flux increased.14,42 Dramatic elevation of aeration rate (more than 2.5 L min−1) decreased permeate flux whereas increased total resistance and irreversible resistance (Fig. 5). In this condition, mechanism of fouling was totally different from non-aerated mechanism. Aeration rate more than 2.5 L min−1 increased shear stress and caused microalgae cells breakage. Hence, microalgae cells were converted to the colloids and small particles which were deposited on the membrane pores and increased irreversible resistance owing to small size of particles.12 Meng et al. studied sludge submerged MBRs across different aeration rate (150, 400 and 800 L min−1). They concluded that low aeration rates (150 L min−1) were not able to remove small particles from the membrane surfaces. On the other hand, high aeration rate (800 L min−1) caused high shear stress, cells breakdown and severe fouling.13 It was worth mentioned that EPS is one of the challengeable parameters affecting permeate flux in high aeration rate.14,15,17 Fig. 6 shows the amount of EPS per microalgae concentration in algal medium for all 13 runs. This ratio increased with growing aeration rate (run number of 11 and 12). Increasing aeration rate and thus shear stress resulted in more EPS extraction and also increasing in EPS transmission due to its deformability.43 In addition, EPS formed a sticky gel layer on the cell walls of deposited microalgae which may cause the membrane pore blocking and increasing membrane fouling.16 In other words, aeration rate has had an optimum of 2.5 L min−1, around which resistance and fouling were larger while permeate flux was lower.
 |
| Fig. 6 Ratio of EPS concentration per microalgae concentration in different microalgae concentration and aeration rate (1 to 13: run numbers). | |
3.4. Study of resistance across different microalgae biomass concentrations
According to ANOVA analysis, permeate flux and thus fouling and total resistance are strongly dependent on microalgae concentration. As can be found in Fig. 4, high aeration rate in low biomass concentration reduced fouling and increased permeate flux. However, in high biomass concentrations (1 g L−1), aeration rate did not have a significant effect on reduction of fouling. This was expected because of the fact that in high microalgae biomass concentrations (more than 0.6 g L−1), cake layer was formed faster and the concentration polarization (CP) became intense.15 Fig. 7 reveals the effect of microalgae biomass concentration on Rc, Rir and Rt at optimum aeration rate of 2.5 L min−1. At low biomass concentrations, Rc was low and the flux was high. Because the microalgae medium initially contained few cells; hence, their adhesion was low and compaction did not occur between them. By further development, Rc experienced a gradual surge from 2.59 × 1011 m−1 at biomass concentration of 0.3 g L−1 to 2.68 × 1011 m−1 at biomass concentration of 0.68 g L−1. In this condition, compaction and adhesion rose to the maximum value on the membrane surface; thus, cake layer was uniformly formed gradually.17 Afterwards, Rc rose linearly at a constant rate. Rir was remained constant from microalgae concentration of 0.3 g L−1 to microalgae concentration of 1 g L−1. Hence, biomass concentration did not have considerable effects on irreversible resistance. As can be found from figure, the trend of Rt is similar to trend of Rc. So it can be concluded that Rt was dependent on Rc, where elevation of Rt did not arise from Rir. Increasing cell numbers of microalgae also released more amount of EPS in algal mixture.44,45 As it was mentioned above, EPS increased cake layer resistance and fouling while decreased permeate flux.
 |
| Fig. 7 Relationship of algal deposition and various filtration resistances in different microalgae concentration (aeration rate: 2.5 L min−1). | |
3.5. Process optimization
Numerical optimization technique was applied to investigate the optimal conditions of the process. The goal was to maximize permeate flux. Predictably, maximum flux would be reached in low microalgae concentration at 0.3 g L−1 and close to the optimum aeration rate at 3.1 L min−1. The maximum predicted value of the flux was 20.11 L m−2 h−1 and the desirability was 0.958.
3.5.1. Validation of optimization model. A final experiment was done by optimization condition data to investigate the accuracy of optimization model. The actual value of permeate flux in optimization condition was 23.03.
4. Conclusion
In this study, RSM was used for modelling and optimization of the anti-fouling processes. Minimum MBR fouling was achieved in minimum microalgae concentration and optimum aeration rate. Based on the ANOVA analysis, the most significant parameter on fouling was aeration rate, which has had an optimum point where the point of permeate flux and fouling have had their maximum and minimum values, respectively. Moreover, in this point, total resistance variations resulted from cake layer resistance variation, while irreversible membrane resistance was as the same as low aeration rate resistance. According to ANOVA analysis, the interactions between the aeration rate and microalgae concentration were not negligible. Low aeration rate could not prevent the algal penetration on membrane pores thus membrane fouling happened while high aeration rate may cause cell breakage and release of more EPS which had significant effects on membrane fouling. EPS also grew by increasing microalgae concentration. More number of algal cells released more amounts of EPS which increased cake layer resistance and thus total resistance.
References
- P. Spolaore, C. Joannis-Cassan, E. Duran and A. Isambert, J. Biosci. Bioeng., 2006, 101, 87–96 CrossRef CAS PubMed.
- Y. Shen, RSC Adv., 2014, 4, 49672–49722 RSC.
- Y. Chen, Y. Wu, D. Hua, C. Li, M. P. Harold, J. Wang and M. Yang, RSC Adv., 2015, 5, 18673–18701 RSC.
- E. W. Wilde and J. R. Benemann, Biotechnol. Adv., 1993, 11, 781–812 CrossRef CAS PubMed.
- M. Bilad, H. A. Arafat and I. F. Vankelecom, Biotechnol. Adv., 2014, 32, 1283–1300 CrossRef CAS PubMed.
- H. Greenwell, L. Laurens, R. Shields, R. Lovitt and K. Flynn, J. R. Soc., Interface, 2009, rsif20090322 Search PubMed.
- F. Wicaksana, A. G. Fane, P. Pongpairoj and R. Field, J. Membr. Sci., 2012, 387, 83–92 CrossRef.
- F. Mohn, Algae biomass: production and use/[sponsored by the National Council for Research and Development, Israel and the Gesellschaft fur Strahlen-und Umweltforschung (GSF), ed. Gedaliah Shelef and Carl J. Soeder, Munich, Germany, 1980 Search PubMed.
- C.-Y. Chen, K.-L. Yeh, R. Aisyah, D.-J. Lee and J.-S. Chang, Bioresour. Technol., 2011, 102, 71–81 CrossRef CAS PubMed.
- N. Rossignol, L. Vandanjon, P. Jaouen and F. Quemeneur, Aquacult. Eng., 1999, 20, 191–208 CrossRef.
- N. Rossi, I. Petit, P. Jaouen, P. Legentilhomme and M. Derouiniot, Sep. Sci. Technol., 2005, 40, 3033–3050 CrossRef CAS.
- N. Rossi, P. Jaouen, P. Legentilhomme and I. Petit, Food Bioprod. Process., 2004, 82, 244–250 CrossRef.
- F. Meng, F. Yang, B. Shi and H. Zhang, Sep. Purif. Technol., 2008, 59, 91–100 CrossRef CAS.
- A. Ahmad, N. M. Yasin, C. Derek and J. Lim, Desalination, 2012, 302, 65–70 CrossRef CAS.
- N. Javadi, F. Z. Ashtiani, A. Fouladitajar and A. M. Zenooz, Bioresour. Technol., 2014, 162, 350–357 CrossRef CAS PubMed.
- M. R. Bilad, V. Discart, D. Vandamme, I. Foubert, K. Muylaert and I. F. Vankelecom, Bioresour. Technol., 2013, 138, 329–338 CrossRef CAS PubMed.
- S. Babel and S. Takizawa, Desalination, 2010, 261, 46–51 CrossRef CAS.
- F. Meng, S.-R. Chae, A. Drews, M. Kraume, H.-S. Shin and F. Yang, Water Res., 2009, 43, 1489–1512 CrossRef CAS PubMed.
- T. Ueda, K. Hata, Y. Kikuoka and O. Seino, Water Res., 1997, 31, 489–494 CrossRef CAS.
- F. Fan, H. Zhou and H. Husain, Water Res., 2006, 40, 205–212 CrossRef CAS PubMed.
- P. Le-Clech, V. Chen and T. A. Fane, J. Membr. Sci., 2006, 284, 17–53 CrossRef CAS.
- I.-S. Chang, P. Le Clech, B. Jefferson and S. Judd, J. Environ. Eng., 2002, 128, 1018–1029 CrossRef CAS.
- C. Chellamboli and M. Perumalsamy, RSC Adv., 2014, 4, 22129–22140 RSC.
- C. Xiurong, J. Binhong and Y. Dailong, Environ. Pollut. Control, 2004, 26, 113–115 Search PubMed.
- B. Cheirsilp and S. Torpee, Bioresour. Technol., 2012, 110, 510–516 CrossRef CAS PubMed.
- H. Azami, M. H. Sarrafzadeh and M. R. Mehrnia, Desalination, 2011, 278, 373–380 CrossRef CAS.
- M. Rickman, J. Pellegrino and R. Davis, J. Membr. Sci., 2012, 423, 33–42 CrossRef.
- R. Field, D. Wu, J. Howell and B. Gupta, J. Membr. Sci., 1995, 100, 259–272 CrossRef CAS.
- A. Lim and R. Bai, J. Membr. Sci., 2003, 216, 279–290 CrossRef CAS.
- H. Yuan and Z. He, Bioresour. Technol., 2015, 195, 202–209 CrossRef CAS PubMed.
- L. Xu, G.-q. Zhang, G.-e. Yuan, H.-y. Liu, J.-d. Liu and F.-l. Yang, RSC Adv., 2015, 5, 22533–22543 RSC.
- A. Babaei, M. R. Mehrnia, J. Shayegan and M.-H. Sarrafzadeh, Process Biochem., 2016, 51, 1568–1575 CrossRef CAS.
- J. Morgan, C. Forster and L. Evison, Water Res., 1990, 24, 743–750 CrossRef CAS.
- O. H. Lowry, N. J. Rosebrough, A. L. Farr and R. J. Randall, J. Biol. Chem., 1951, 193, 265–275 CAS.
- M. Dubois, K. A. Gilles, J. K. Hamilton, P. Rebers and F. Smith, Anal. Chem., 1956, 28, 350–356 CrossRef CAS.
- M. J. Bashir, H. A. Aziz, M. S. Yusoff and M. N. Adlan, Desalination, 2010, 254, 154–161 CrossRef CAS.
- X. Zhang, Q. Hu, M. Sommerfeld, E. Puruhito and Y. Chen, Bioresour. Technol., 2010, 101, 5297–5304 CrossRef CAS PubMed.
- D. Montgomery, Design and analysis of experiments, John Wiley & Sons, 2008. Search PubMed.
- M. H. Muhamad, S. R. S. Abdullah, A. B. Mohamad, R. A. Rahman and A. A. H. Kadhum, J. Environ. Manage., 2013, 121, 179–190 CrossRef CAS PubMed.
- A. M. Joglekar, A. T. May, E. Graf and I. Saguy, Food Product and Development: From Concept to the Marketplace, 1987, pp. 211–230 Search PubMed.
- B. K. Körbahti, N. Aktaş and A. Tanyolaç, J. Hazard. Mater., 2007, 148, 83–90 CrossRef PubMed.
- Z. Cui, S. Chang and A. Fane, J. Membr. Sci., 2003, 221, 1–35 CrossRef CAS.
- I. Gómez, A. Wulff, M. Y. Roleda, P. Huovinen, U. Karsten, M. L. Quartino, K. Dunton and C. Wiencke, Bot. Mar., 2009, 52, 593–608 Search PubMed.
- A. Drews, C.-H. Lee and M. Kraume, Desalination, 2006, 200, 186–188 CrossRef CAS.
- J.-B. Castaing, A. Massé, M. Pontié, V. Séchet, J. Haure and P. Jaouen, Desalination, 2010, 253, 71–77 CrossRef CAS.
|
This journal is © The Royal Society of Chemistry 2016 |
Click here to see how this site uses Cookies. View our privacy policy here.