- Email: [email protected]
Optimized conditions for phytoremediation of diesel by Scirpus grossus in horizontal subsurface flow constructed wetlands (HSFCWs) using response surface methodology
Journal of Environmental Management 140 (2014) 152e159
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Optimized conditions for phytoremediation of diesel by Scirpus grossus in horizontal subsurface flow constructed wetlands (HSFCWs) using response surface methodology Israa Abdul Wahab Al-Baldawi a, b, d, *, Siti Rozaimah Sheikh Abdullah a, Hassimi Abu Hasan a, Fatihah Suja b, Nurina Anuar a, Idris Mushrifah c a
Department of Chemical and Process Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia Department of Civil and Structural Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia Tasik Chini Research Centre, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, Selangor, Malaysia d Department of Biochemical Engineering, Al-khwarizmi College of Engineering, University of Baghdad, Baghdad, Iraq b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 October 2013 Received in revised form 12 March 2014 Accepted 17 March 2014 Available online 21 April 2014
This study investigated the optimum conditions for total petroleum hydrocarbon (TPH) removal from diesel-contaminated water using phytoremediation treatment with Scirpus grossus. In addition, TPH removal from sand was adopted as a second response. The optimum conditions for maximum TPH removal were determined through a Box-Behnken Design. Three operational variables, i.e. diesel concentration (0.1, 0.175, 0.25% Vdiesel/Vwater), aeration rate (0, 1 and 2 L/min) and retention time (14, 43 and 72 days), were investigated by setting TPH removal and diesel concentration as the maximum, retention time within the given range, and aeration rate as the minimum. The optimum conditions were found to be a diesel concentration of 0.25% (Vdiesel/Vwater), a retention time of 63 days and no aeration with an estimated maximum TPH removal from water and sand of 76.3 and 56.5%, respectively. From a validation test of the optimum conditions, it was found that the maximum TPH removal from contaminated water and sand was 72.5 and 59%, respectively, which was a 5 and 4.4% deviation from the values given by the Box-Behnken Design, providing evidence that S. grossus is a Malaysian native plant that can be used to remediate wastewater containing hydrocarbons. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Pilot scale constructed wetlands TPH removal Aeration Box-Behnken Design
1. Introduction One of the major threats to water quality is chemical pollution, especially from the petrochemical, pesticide and pharmacy industries, among others. Diesel is a source of organic pollution in the environment that can pose challenges to remediation design because it comprises hundreds of compounds. Regarding the fact that hydrocarbon compounds are found in industrial effluents, they can impose a great threat to the environment and humans. Current methods applied to remove organic components from wastewater include adsorption on activated carbon (Petrova et al., 2010), chemical oxidation (Liang et al., 2011), electrochemical (Souza and Ruotolo, 2013) and others. Even so, these methods present certain
* Corresponding author. Department of Chemical and Process Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia. Tel.: þ60 3 89216407; fax: þ60 3 89216148. E-mail addresses: [email protected], [email protected] (I.A.W. AlBaldawi), [email protected] (S.R. Sheikh Abdullah). http://dx.doi.org/10.1016/j.jenvman.2014.03.007 0301-4797/Ó 2014 Elsevier Ltd. All rights reserved.
disadvantages, such as low efficiency and high cost (Konnerup et al., 2009). Therefore, new green technologies are being pursued as alternative wastewater treatment methods for the removal of toxic organic pollutants, such as phytoremediation (Huang et al., 2004). This is an environmentally friendly engineering technology that has been successful in cleaning up the environment in a cost effective way without destroying the site (Nivala et al., 2013). Diesel oil is a complex fuel mixture with hydrocarbon molecules containing from 8 to 40 atoms of carbon, and are generally heavier than those found in gasoline (Vieira et al., 2009). Chapelle (1999) mentioned that under both aerobic and anaerobic conditions, microorganisms are generally able to degrade hydrocarbons in contaminated soil. Constructed wetlands (CWs) have been commonly used in water treatment systems as a substitute for conventional methods due to their low energy requirements as well as easy operation and maintenance (Garcia et al., 2010; Chen et al., 2012). However, the data regarding the use of CWs for treating water contaminated with hydrocarbons have been sparse until now. Phytoremediation in CWs has been successfully used to
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
remove organic contaminants from industrial wastewater as a natural in situ water treatment method (Al-Baldawi et al., 2013a; Afferden et al., 2011; Fountoulakis et al., 2009). Wetlands can be used for the biological treatment of wastewater by plants which enhance microbial growth and thus promote contaminant degradation by supplying oxygen and root exudates in the rhizosphere zone (Stottmeister et al., 2003; Seeger et al., 2011). Wetland systems are not expensive and require less maintenance than the traditional remediation technologies because they utilize naturally occurring physical, chemical and biological processes to remove contaminants (Zhang et al., 2010). Response Surface Method (RSM) is an efficient statistical tool that can be used for modelling and optimization of more than one process variable (El-Ghenymy et al., 2012). Using 3D response surface plots, one can better understand the relationship between process variables and responses of experiments. In order to reduce laboratory experiments and to save time and cost, the application of RSM is a recommended method for the soft modelling of pilot scale studies, especially for the phytoremediation process. Demima et al. (2014) conducted a central composite design analysis of a phytoremediation process to optimize the factors that influenced the removal of heavy metals from an aqueous solution by L. gibba and to investigate their impact on the process. Recently, RSM has been applied to the optimization of several water treatment processes, such as the electrochemical process (Wu et al., 2012; El-Ghenymy et al., 2012), ozonation (Zeng et al., 2012), ultrafiltration (Yuliwati et al., 2012), nano-porous membranes (Salahi et al., 2013) and biofilm reactors (Hasan et al., 2011; Muhamad et al., 2013). Since the optimization of phytoremediation pilot scale studies is rare, in this study we have adopted RSM using the Box-Behnken Design (BBD) to optimize the performance of the phytoremediation of wastewater contaminated with diesel. In the present study, we used horizontal subsurface flow constructed wetlands (HSFCWs) using Scirpus grossus to remediate water contaminated with different diesel concentrations with various aeration rate options for a period of 72 days in a field approach. We aimed to maximise the TPH removal efficiency from water and sand using RSM through a Box-Behnken experimental design by optimising the diesel concentration, retention time and aeration rate.
153
Fig. 1. Pilot scale constructed wetland design.
2.2. Analyses of total petroleum hydrocarbons (TPH) Three water samples from each constructed wetland (100 mL each) were collected periodically in clean schott duran bottle on each sampling day to extract TPH. To analyse the sand, three replicates of 10 g sand were collected at 5 cm depth from the top surface of sand layer periodically in clean schott duran bottle from each constructed wetland on the same sampling days to extract TPH. The TPH concentration measurements were carried out according to the methodology described in our previous study (AlBaldawi et al., 2013c).
2. Materials and methods 2.3. Optimization conditions with the Box-Behnken design 2.1. Pilot scale setup and operation The pilot scale constructed wetland system was designed as a tank made of fiberglass with a thickness of 0.5 cm. The dimension of each pilot scale constructed wetland was L180 W 90 H 90 cm. It was filled with coarse gravel (V 20e50 mm) with a height of 15 cm for the lowest layer, then 15 cm of fine gravel (V 10e20 mm) for the second layer and 20 cm of fine sand (V 0.5e1 mm) for the upper layer. A schematic diagram of the pilot scale constructed wetland and a complete substrate constructed wetland setup are depicted in Fig. 1. The pilot scale constructed wetland was operated in a horizontal subsurface flow system (HSFCWs) configuration. The pilot scale was run with three diesel concentrations and aeration rates for 72 days. Synthetic wastewater was prepared by mixing diesel with water in different concentrations (0.0, 0.1, 0.175, 0.25% Vdiesel/ Vwater). In the aerated constructed wetlands, air was supplied through air diffuser pipes located 20 cm below the bed surface. More details on the operations of the constructed wetlands can be referred in our previous study (Al-Baldawi et al., 2013b). A perennial native Malaysian plant, the bulrush S. grossus, was used in this study. Fifty plants were planted in each constructed wetland at a depth of 2e10 cm in the sand medium.
In the optimization study of the pilot scale system, the TPH concentration in water and sand was set using a BBD. The interaction between the main factors of diesel concentration, retention time and aeration rate, and the response of TPH removal efficiency in contaminated water and sand were investigated. The results were then analysed to develop an appropriate model for these factors. The response data of TPH removal from contaminated water was derived from the pilot scale study. The variability factors included in the design were diesel concentration (0.1, 0.175 and 0.25% Vdiesel/Vwater), retention time (14, 43 and 72 days) and aeration rate (0, 1 and 2 L/min) (Table 1). 2.4. Box-Behnken Design (BBD) The effects of diesel concentration (X1), retention time (X2) and aeration rate (X3) on TPH removal from contaminated water and sand was investigated using BBD. The design simulated 17 total experiment including five replicates to assess the error magnitude that occurs randomly. The dependent variable was TPH removal in water and sand. BBD was used for the statistical design of experiments and data analysis. In the optimization, the responses were
154
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
coupled to selected variables by linear or quadratic models. The mathematical relations between the responses and these variables can be represented by quadratic models Eq. (1) (Kayan and Gözmen, 2012):
Y ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ ε
(1)
where Y is the response (TPH removal efficiency); x1, x2, and x3 represent the effect of the independent variables; x21, x22, x23 are the square effects; and x1x2, x1x3, and x2x3 are the interaction effects; b0 is constant; b1, b2 and b3 are the linear coefficients; b12, b13 and b23 are the interaction coefficients; and b11, b22, and b33 are quadratic coefficients; ε is the random error. The main aim of the design experiment was to optimize the response (Y) based on the factors investigated using the BBD model (Gan and Latiff, 2011; Kousha et al., 2012). Design ExpertÒ software (version 6.0.10, Stat-Ease, USA) was used to simulate the experimental run and optimize TPH removal in water and sand. The statistical significance of the first order model equation was determined by performing analysis of variance (ANOVA). Especially, a valid model must be significant based on the F-value and Pvalue in contrast to a lack of fit (insignificant). Furthermore, the proportion of variance shown by the multiple coefficients of determination, R2, should be close to 1 as possible to show a good correlation between the experimental and predicted values (Muhamad et al., 2013; Zeng et al., 2012). For optimization, the diesel concentration was set as the maximum, the retention time was set in the given range and the aeration rate was set as the minimum in order to maximize the removal of TPH from water and sand. The selected conditions were set since 1) the bulrush Scirpus grossus has the ability to tolerate toxicity due to high diesel concentrations; 2) the system can potentially be operated within the given time period of 72 days (Al-Baldawi et al., 2013b) and 3) we aimed to minimize the aeration rate for cost-saving reactor operation.
Table 1 Variables and their levels in the experimental design. Variables
Diesel concentration Retention time Aeration rate
Units
(Vdiesel/Vwater) % Day L/min
Symbols
X1 X2 X3
Code levels 1
0
þ1
0.1 14 0
0.175 43 1
0.25 72 2
The model was used to determine the TPH removal from water and sand in the pilot scale operation with optimal conditions. A normal plot of the residuals was analysed to check the normality assumption. The diagnostic of normal residuals demonstrated in Fig. 2 indicates that the residual behaviour followed a normal distribution and was quadratic, which is the more important assumption for checking statistical modelling. Therefore, the quadratic model built was adequate. The actual and predicted values of responses are listed in Table 2 with errors from 14.1e 11.5% between the actual and predicted values. Fig. 3 shows the predicted output values versus actual experimental values for TPH removal from water. From this figure, it can be noted that the values calculated using the predictive quadratic model were in good agreement with the experimental values with a satisfactory correlation between these values. Therefore, the developed model is suitable for predicting the efficiency of TPH removal from water under the investigated conditions. The results show that TPH removal from water and sand ranged from 33.3 to 88.3% and 18.9e67.3%, respectively. The maximum TPH
3. Results and discussion 3.1. Evaluation of simulated run by BBD BBD was selected to determine the relationship between the factors (X1, X2, X3) and the response function (Y). A quadratic model was built to fit the results, of which the coefficients were calculated by multiple regression analysis. The quadratic model obtained is shown in Eqs. (2) and (3).
Y1 ¼ þ75:08 þ 1:55*X1 þ 20:66*X2 þ 2:85*X3 þ 1:32*X12 11:43*X22 5:33*X32 0:91*X1 *X2 þ 1:03*X1 *X3 þ 0:94*X2 *X3
(2)
Y2 ¼ þ48:31 þ 1:66*X1 þ 15:90*X2 0:20*X3 þ 2:92*X12 10:46*X22 þ 5:86*X32 þ 2:12*X1 *X2 þ 5:93*X1 *X3 þ 5:70*X2 *X3
(3)
with, Y1 ¼ % TPH removal from water Y2 ¼ % TPH removal from sand X1 ¼ diesel concentration (Vdiesel/Vwater) X2 ¼ retention time (days) X3 ¼ aeration rate (L/min)
Fig. 2. Normal probability plot of studentized residuals of the quadratic model for TPH removal efficiency from: (a) water and (b) sand.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
155
Table 2 Box-behnken design matrix for three test factors in uncoded units along with the observed and predicted responses for TPH removal from water and sand. Run
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Variables
TPH from water
TPH from sand
X1: Diesel concentration (%v/v)
X2: Retention time (days)
X3: Aeration rate (L/min)
Actual value (%)
Predicted value (%)
Actual value (%)
Predicted value (%)
0.1 0.25 0.1 0.25 0.1 0.25 0.1 0.25 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175
14 14 72 72 43 43 43 43 14 72 14 72 43 43 43 43 43
1 1 1 1 0 0 2 2 0 0 2 2 1 1 1 1 1
39.8 47.6 84.1 88.3 72.2 70.3 69.7 72.0 33.3 71.5 43.2 85.2 78.4 85.4 61.0 83.6 67.0
45.6 48.7 86.9 90.0 63.4 66.5 69.1 72.2 44.3 85.6 50.0 91.3 67.8 67.8 67.8 67.8 67.8
28.1 24.6 52.7 57.7 58.7 52.7 49.6 67.3 33.8 57.2 18.9 65.0 59.8 35.1 40.0 49.6 57.1
30.0 33.3 61.8 65.1 46.1 49.4 45.7 49.0 31.8 63.6 31.4 63.2 47.5 47.5 47.5 47.5 47.5
removal efficiency from water (88.3%) was found in Run 4 under the experimental conditions of diesel concentration (X1 ¼ 0.25%), retention time (X2 ¼ 72 days) and aeration rate (X3 ¼ 1 L/min). Meanwhile, the maximum TPH removal from sand was observed in Run 8 with diesel concentration (X1 ¼ 0.25%), retention time (X2 ¼ 43 days) and aeration rate (X3 ¼ 2 L/min). These situations varied depending on the response required. Therefore, optimum process situation should be considered in order to obtain maximum TPH removal efficiency from water and sand.
ANOVA was performed to test the significance and adequacy of the model. The ANOVA results of the suggested model for TPH removal from water and sand are presented in Table 3 and indicate that the equation effectively represents the relationship between the response (TPH removal from water and sand) and the significant input variables. From the ANOVA (Table 3), the F-value for the model was more than 5.96 for TPH removal from water and 4.52 for TPH removal from sand. The high F-value indicates that most of the variation in the response can be explained by the regression equation, which implies that the terms in the model have a significant effect on the response. The associated P-value was used to estimate whether the F-value is large enough to indicate statistical significance. The P-values for TPH removal from water and sand were found to be lower than 0.05, indicating that the model is statistically significant at the 95% probability level. For the optimization of TPH removal, a “good fit” model is required to avoid poor or unclear results (Mayers et al., 2009). To ensure the adequacy of the proposed model, the R2, adjusted R2, F test, and a lack-of-fit test were evaluated. The ANOVA results showed significant (P < 0.05) response surface models with good R2 (0.8846) (Table 3) and adjusted R2 (0.73) values for the model Y1 and R2 (0.8853) and adjusted R2 (0.66) values for the model Y2, indicating the significance of the models (Khuri and Cornell, 1996). “Adeq Precision” measures the signal to noise ratio. A ratio greater than 4 is desirable. For the present study, the signal to noise ratio was found to be 7.44 (for TPH removal from water) and 6.62 (for TPH removal from sand), which indicates an adequate signal. Therefore, the quadratic model can be used to navigate the design space. Table 3 also presents the results of lack-of-fit tests for TPH removal from water and sand. The F-values of the lack-of-fit tests for the both models showed that variations in the data around the fitted model were not significant relative to the pure error, suggesting significant model correlation between the variable and process response. There were 84 and 92.6% chance that the lack-offit F-value could have occurred due to noise for TPH removal from water and sand. Therefore, the suggested model was statistically significant, and showed agreement between the observed and predicted responses. 3.2. Optimization of operational conditions
Fig. 3. Actual versus predicted values for TPH removal from: (a) water and (b) sand.
The target of optimizing the process was to find the optimum operation conditions leading to maximum TPH removal from water
156
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
Table 3 ANOVA analysis of the quadratic model to the TPH removal from medium. Source
Sum of squares
TPH removal from water Model 4207.451
Degrees freedom
Mean square
F-value
9
Prob > F
5.965
0.0140
Significant
0.245
0.6360
43.546
0.0003
0.826
0.3936
0.094
0.7684
7.015
0.0330
1.529
0.2562
0.042
0.8430
0.054
0.8223
0.045
0.8380
0.277
0.8403
Not significant
4.524 0.306 27.980 0.004 0.498 6.370 2.000 0.250 1.946 1.795
0.0296 0.5975 0.0011 0.9490 0.5032 0.0396 0.2002 0.6326 0.2057 0.2222
Significant
0.147
0.9262
Not significant
467.495 X1
19.182
1 19.182
X2
3413.144
1 3413.144
X3
64.755
1 64.755
X21
7.348
1 7.348
X22
549.856
1 549.856
X23
119.815
1 119.815
X1X2
3.311
1 3.311
X1X3
4.259
1 4.259
X2X3
3.531
1 3.531
Residual
548.656
7
Lack of fit
94.252
3
78.379 31.417 Pure error
454.404
4 113.601
Cor total 4756.107 R2 ¼ 0.8846 R2adj ¼ 0.7363 TPH removal from sand Model 2943.696 X1 22.108 X2 2023.009 X3 0.318 2 X1 36.008 X22 460.556 X23 144.579 X1X2 18.051 X1X3 140.695 X2X3 129.769 Residual 506.106 Lack of fit 50.353 Pure error 455.753 Cor total 3449.801 R2 ¼ 0.8853 R2adj ¼ 0.6647
16 Adeq precision ¼ 7.4387 9 1 1 1 1 1 1 1 1 1 7 3 4 16 Adeq precision ¼ 6.6236
327.077 22.108 2023.009 0.318 36.008 460.556 144.579 18.051 140.695 129.769 72.301 16.784 113.938
Values of ‘Prob > F’ less than 0.05 indicate model terms are significant.
and sand. The desirability function methodology was used for this optimization. When operating in the pilot scale, variables of diesel concentration and retention time were set within a certain range and the aeration rate was minimized. By using the function of
numerical optimization in the Design Expert software, we found a desirability of 0.883 for the maximum TPH removal efficiencies (Fig. 4). The maximum TPH removal from water and sand was 76.3% and 56.5%, respectively, at the optimized conditions of 0.25% diesel
Fig. 4. Desirability slope for numerical optimization conditions.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
157
3.3. The relationship between the response and variables under the optimum conditions
concentration (X1), 63 days retention time (X2) and 0 L/min aeration rate (X3) (Fig. 4). Validation experiments were carried under at the optimum conditions to confirm the predicted optimum response. The measured results under the optimum conditions were about 72.5% for TPH removal from water and 59% for TPH removal from sand. These results were very close to the predicted values, indicating the adequacy of the obtained model to optimize TPH removal from water and sand. The deviations between the measured and predicted values were within 5%. Therefore, it can be concluded that the regression models were appropriate in their reduced forms.
In this study, the aim of optimization was to find the conditions that provide maximum TPH removal from water in a phytoremediation process. The three dimensional (3D) response surface is a graphical representation of the regression function. The 3D response surface plots (Figs. 5 and 6) show how TPH removal from water (response variable Y1) and sand (response variable Y2) relate to the factors of diesel concentration (X1), retention time (X2) and aeration rate (X3) through quadratic model equations (Eqs. (2) and
Fig. 5. Response surface for TPH removal from water as a function of the varibles: (a) diesel concentration (X1) and retention time (X2), (b) diesel concentration (X1) and aeration rate (X3) and (c) retention time (X2) and aeration rate (X3).
Fig. 6. Response surface for TPH removal from sand as a function of the varibles: (a) diesel concentration (X1) and retention time (X2), (b) diesel concentration (X1) and aeration rate (X3) and (c) retention time (X2) and aeration rate (X3).
158
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
(3)). Figs. 5a and 6a shows the 3D surface plot of the effect of diesel concentration (X1) and retention time (X2) on TPH removal from water and sand. It can be seen that a diesel concentration within the range of 0.1e0.25% (Vdiesel/Vwater) led to steady TPH removal from water and sand. This might due to the range of diesel concentrations selected were not significantly different. The effect of retention time (X2) was an increase in the response within 14e72 days. It is generally believed that the long retention time of a phytoremediation operation leads to high TPH removal. Figs. 5b and 6b depict the effect of diesel concentration (X1) and aeration rate (X3) on TPH removal from water and sand. It was observed that TPH removal from water and sand increased with an increase in the aeration rate. It was clearly shown that there was a steady increase in the TPH removal from water when the aeration rate was increased from 0 to 1 L/min and later gradually decreased when the aeration was further increased to 2 L/min; the aeration rate was effectively positive until 2 L/min for TPH removal from sand. The actual TPH removal from water was 88.3% with an aeration rate of 1 L/min, which is enough on the pilot scale with a subsurface flow system to enhance oxygen availability for biodegradation. However, the maximum actual TPH removal from sand (67.3%) was achieved with an aeration rate of 2 L/min because the contaminant degradation was enhanced in sand medium by supplying aeration to accelerate the aerobic biodegradation process (Seeger et al., 2011). Figs. 5c and 6c show the effect of retention time (X2) and aeration rate (X3) on TPH removal from water and sand. It was observed that TPH removal increased with an increase in the aeration rate, but the effect of aeration rate was not significantly different (P > 0.05) when applying a high level of 2 L/min. This demonstrates that longer retention time reduces more TPH concentration in water, but the diesel concentration and aeration rate do not have much effect on TPH removal efficiency. 4. Conclusions The optimization of a pilot constructed wetland model associated with the efficiency of TPH removal from water and sand was performed using the Box-Behnken Design. The results show that retention time was the most significant factor for the process, while no or little effect of diesel concentration and aeration rate was seen on TPH removal efficiency from water and sand. The coefficients of determination (R2) for the model equations (Y1 and Y2) were greater than 0.85 with the probability values (P < 0.05) demonstrating significance for the regression model to predict the responses. The experimental values were close to the predicted theoretical values, indicating that the models could be validated for the optimization of TPH removal from water and sand. The optimum conditions of the process were found to be at a diesel concentration of 0.25%, retention time of 63 days and no aeration with 76.3% and 56.5% maximum TPH removal from water and sand, respectively. These experimental results were found to be in good agreement with the predicted values obtained by the models. Therefore, the TPH phytoremediation process in constructed wetlands is a potential option technology for the treatment of wastewater contaminated with diesel. Since this study was performed batchwise, a future work will look on continuous operation of horizontal subsurface flow system constructed wetland. Acknowledgements The authors would like to thank Universiti Kebangsaan Malaysia (UKM-KK-03-FRGS0119-2010) and the Tasik Chini Research Centre for supporting this research project. They also acknowledge with thanks to the Iraqi Ministry of Higher Education and Scientific Research for providing a doctoral scholarship for the first author.
References Afferden, M., Rahman, K.Z., Mosig, P., De Biase, C., Thullner, M., Oswald, S.E., Müller, R.A., 2011. Remediation of groundwater contaminated with MTBE and benzene: the potential of vertical-flow soil filter systems. Water Res. 45, 5063e 5074. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013a. A phytotoxicity test of bulrush (Scirpus grossus) grown with diesel contamination in a free-flow reed bed system. J. Hazard. Mater. 252e253, 64e69. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013b. Effect of aeration on hydrocarbon phytoremediation capability in pilot sub-surface flow constructed wetland operation. Ecol. Eng. 61, 496e500. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013c. Phytotoxicity test of Scirpus grossus on diesel-contaminated water using a subsurface flow system. Ecol. Eng. 54, 49e56. Chapelle, F.H., 1999. Bioremediation of petroleum hydrocarbon-contaminated ground water: the perspectives of history and hydrology. Ground Water 37, 122e132. Chen, Z., Wu, S., Braeckevelt, M., Paschke, H., Kästner, M., Köser, H., Kuschk, P., 2012. Effect of vegetation in pilot-scale horizontal subsurface flow constructed wetlands contaminated with a low and high chlorinated hydrocarbon. Chemosphere 89, 724e731. Demima, S., Drouiche, N., Aouabed, A., Benayad, T., Couderchet, M., Semsari, S., 2014. Study of heavy metal removal from heavy metal mixture using the CCD method. J. Ind. Eng. Chem. 20, 512e520. El-Ghenymy, A., Garcia-Segura, S., Rodríguez, R.M., Brillas, E., El Begrani, M.S., Abdelouahid, B.A., 2012. Optimization of the electro-Fenton and solar photoelectro-Fenton treatments of sulfanilic acid solutions using a pre-pilot flow plant by response surface methodology. J. Hazard. Mater. 221e222, 288e297. Fountoulakis, M.S., Terzakis, S., Kalogerakis, N., Manios, T., 2009. Removal of polycyclic aromatic hydrocarbons and linear alkylbenzene sulphonates from domestic wastewater in pilot constructed wetlands and a gravel filter. Ecol. Eng. 35, 1702e1709. Gan, C., Latiff, A.A., 2011. Extraction of antioxidant pectic-polysaccharide from mangosteen (Garcinia mangostana) rind: optimization using response surface methodology. Carbohydr. Polym. 83, 600e607. Garcia, J., Rousseau, D.P.L., Morato, J., Lesage, E.L.S., Matamoros, V., Bayona, J.M., 2010. Contaminant removal processes in subsurface-flow constructed wetlands: a review. Environ. Sci. Technol. 40, 561e661. Hasan, H.A., Abdullah, S.R.S., Kamarudin, S.K., Kofli, N.T., 2011. Response surface 2þ methodology for optimization of simultaneous COD, NHþ removal 4 -N and Mn from drinking water by biological aerated filter. Desalination 275, 50e61. Huang, X., El-Alawi, Y., Penrose, D.M., Glick, B.R., Greenberg, B.M., 2004. A multiprocess phytoremediation system for removal of polycyclic aromatic hydrocarbons from contaminated soils. Environ. Pollut. 130, 465e476. Kayan, Berkant, Gözmen, Belgin, 2012. Degradation of Acid Red 274 using H2O2 in subcritical water: application of response surface methodology. J. Hazard. Mater. 201e202, 100e106. Khuri, A.I., Cornell, J.A., 1996. Response Surfaces: Designs and Analyses, second ed. New York, Marcel Dekker. Konnerup, D., Koottatep, T., Brix, H., 2009. Treatment of domestic wastewater in tropical, subsurface flow constructed wetlands planted with Canna and Heliconia. Ecol. Eng. 35, 248e257. Kousha, M., Daneshvar, E., Dopeikar, H., Taghavi, D., Bhatnagar, A., 2012. Boxe Behnken design optimization of Acid Black 1 dye biosorption by different brown macroalgae. Chem. Eng. J. 179, 158e168. Liang, S.H., Kao, C.M., Kuo, Y.C., Chen, K.F., Yang, B.M., 2011. In situ oxidation of petroleum-hydrocarbon contaminated groundwater using passive ISCO system. Water Res. 45, 2496e2506. Mayers, R.H., Montgomery, D.C., Anderson-Cook, C.M., 2009. Response Surface Methodology: Process and Product Optimization Using Designed Experiments, third ed. John Wily & Sons, New Jersey. Muhamad, M.H., Sheikh Abdullah, S.R., Mohamad, A., Abdul Rahman, R., Kadhum, A.H., 2013. Application of response surface methodology (RSM) for optimisation of COD, NH3eN and 2,4-DCP removal from recycled paper wastewater in a pilot-scale granular activated carbon sequencing batch biofilm reactor (GAC-SBBR). J. Environ. Manage. 121, 179e190. Nivala, J., Headley, T., Wallace, S., Bernhard, K., Brix, H., Afferden, M., Müller, R.A., 2013. Comparative analysis of constructed wetlands: the design and construction of the ecotechnology research facility in Langenreichenbach, Germany. Ecol. Eng. 61, 527e543. Petrova, B., Budinova, T., Tsyntsarski, B., Kochkodan, V., Shkavro, Z., Petrov, N., 2010. Removal of aromatic hydrocarbons from water by activated carbon from apricot stones. Chem. Eng. J. 165, 258e264. Salahi, A., Noshadi, I., Badrnezhad, R., Kanjilal, B., Mohammadi, T., 2013. Nanoporous membrane process for oily wastewater treatment: optimization using response surface methodology. J. Environ. Chem. Eng. 1, 218e225. Seeger, E.V., Kuschk, P., Fazekas, H., Grathwohl, P., Kaestner, M., 2011. Bioremediation of benzene-, MTBE- and ammonia-contaminated groundwater with pilotscale constructed wetlands. Environ. Pollut. 159, 3769e3776. Souza, R.B.A., Ruotolo, L.A.M., 2013. Electrochemical treatment of oil refinery effluent using boron-doped diamond anodes. J. Environ. Chem. Eng. 1, 544e 551.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159 Stottmeister, U., Wießner, A., Kuschk, P., Kappelmeyer, U., Kästner, M., Bederski, O., Müller, R.A., Moormann, H., 2003. Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotechnol. Adv. 22, 93e117. Vieira, P.A., Vieira, R.B., Faria, S., Ribeiro, E.J., Cardoso, V.L., 2009. Biodegradation of diesel oil and gasoline contaminated effluent employing intermittent aeration. J. Hazard. Mater. 168, 1366e1372. Wu, J., Zhang, H., Oturan, N., Wang, Y., Chen, L., Oturan, M.A., 2012. Application of response surface methodology to the removal of the antibiotic tetracycline by electrochemical process using carbon-felt cathode and DSA (Ti/RuO2eIrO2) anode. Chemosphere 87, 614e620.
159
Yuliwati, E., Ismail, A.F., Lau, W.J., Ng, B.C., Mataram, A., Kassim, M.A., 2012. Effects of process conditions in submerged ultrafiltration for refinery wastewater treatment: optimization of operating process by response surface methodology. Desalination 287, 350e361. Zeng, Z., Zou, H., Li, X., Sun, B., Chen, J., Shao, L., 2012. Ozonation of acidic phenol wastewater with O3/Fe(II) in a rotating packed bed reactor: optimization by response surface methodology. Chem. Eng. Process 60, 1e8. Zhang, B., Zheng, J., Sharp, R., 2010. Phytoremediation in engineered wetlands: mechanisms and applications. Procedia Environ. Sci. 2, 1315e1325.
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
Optimized conditions for phytoremediation of diesel by Scirpus grossus in horizontal subsurface flow constructed wetlands (HSFCWs) using response surface methodology Israa Abdul Wahab Al-Baldawi a, b, d, *, Siti Rozaimah Sheikh Abdullah a, Hassimi Abu Hasan a, Fatihah Suja b, Nurina Anuar a, Idris Mushrifah c a
Department of Chemical and Process Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia Department of Civil and Structural Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia Tasik Chini Research Centre, Faculty of Science and Technology, Universiti Kebangsaan Malaysia, Selangor, Malaysia d Department of Biochemical Engineering, Al-khwarizmi College of Engineering, University of Baghdad, Baghdad, Iraq b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 28 October 2013 Received in revised form 12 March 2014 Accepted 17 March 2014 Available online 21 April 2014
This study investigated the optimum conditions for total petroleum hydrocarbon (TPH) removal from diesel-contaminated water using phytoremediation treatment with Scirpus grossus. In addition, TPH removal from sand was adopted as a second response. The optimum conditions for maximum TPH removal were determined through a Box-Behnken Design. Three operational variables, i.e. diesel concentration (0.1, 0.175, 0.25% Vdiesel/Vwater), aeration rate (0, 1 and 2 L/min) and retention time (14, 43 and 72 days), were investigated by setting TPH removal and diesel concentration as the maximum, retention time within the given range, and aeration rate as the minimum. The optimum conditions were found to be a diesel concentration of 0.25% (Vdiesel/Vwater), a retention time of 63 days and no aeration with an estimated maximum TPH removal from water and sand of 76.3 and 56.5%, respectively. From a validation test of the optimum conditions, it was found that the maximum TPH removal from contaminated water and sand was 72.5 and 59%, respectively, which was a 5 and 4.4% deviation from the values given by the Box-Behnken Design, providing evidence that S. grossus is a Malaysian native plant that can be used to remediate wastewater containing hydrocarbons. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Pilot scale constructed wetlands TPH removal Aeration Box-Behnken Design
1. Introduction One of the major threats to water quality is chemical pollution, especially from the petrochemical, pesticide and pharmacy industries, among others. Diesel is a source of organic pollution in the environment that can pose challenges to remediation design because it comprises hundreds of compounds. Regarding the fact that hydrocarbon compounds are found in industrial effluents, they can impose a great threat to the environment and humans. Current methods applied to remove organic components from wastewater include adsorption on activated carbon (Petrova et al., 2010), chemical oxidation (Liang et al., 2011), electrochemical (Souza and Ruotolo, 2013) and others. Even so, these methods present certain
* Corresponding author. Department of Chemical and Process Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Selangor, Malaysia. Tel.: þ60 3 89216407; fax: þ60 3 89216148. E-mail addresses: [email protected], [email protected] (I.A.W. AlBaldawi), [email protected] (S.R. Sheikh Abdullah). http://dx.doi.org/10.1016/j.jenvman.2014.03.007 0301-4797/Ó 2014 Elsevier Ltd. All rights reserved.
disadvantages, such as low efficiency and high cost (Konnerup et al., 2009). Therefore, new green technologies are being pursued as alternative wastewater treatment methods for the removal of toxic organic pollutants, such as phytoremediation (Huang et al., 2004). This is an environmentally friendly engineering technology that has been successful in cleaning up the environment in a cost effective way without destroying the site (Nivala et al., 2013). Diesel oil is a complex fuel mixture with hydrocarbon molecules containing from 8 to 40 atoms of carbon, and are generally heavier than those found in gasoline (Vieira et al., 2009). Chapelle (1999) mentioned that under both aerobic and anaerobic conditions, microorganisms are generally able to degrade hydrocarbons in contaminated soil. Constructed wetlands (CWs) have been commonly used in water treatment systems as a substitute for conventional methods due to their low energy requirements as well as easy operation and maintenance (Garcia et al., 2010; Chen et al., 2012). However, the data regarding the use of CWs for treating water contaminated with hydrocarbons have been sparse until now. Phytoremediation in CWs has been successfully used to
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
remove organic contaminants from industrial wastewater as a natural in situ water treatment method (Al-Baldawi et al., 2013a; Afferden et al., 2011; Fountoulakis et al., 2009). Wetlands can be used for the biological treatment of wastewater by plants which enhance microbial growth and thus promote contaminant degradation by supplying oxygen and root exudates in the rhizosphere zone (Stottmeister et al., 2003; Seeger et al., 2011). Wetland systems are not expensive and require less maintenance than the traditional remediation technologies because they utilize naturally occurring physical, chemical and biological processes to remove contaminants (Zhang et al., 2010). Response Surface Method (RSM) is an efficient statistical tool that can be used for modelling and optimization of more than one process variable (El-Ghenymy et al., 2012). Using 3D response surface plots, one can better understand the relationship between process variables and responses of experiments. In order to reduce laboratory experiments and to save time and cost, the application of RSM is a recommended method for the soft modelling of pilot scale studies, especially for the phytoremediation process. Demima et al. (2014) conducted a central composite design analysis of a phytoremediation process to optimize the factors that influenced the removal of heavy metals from an aqueous solution by L. gibba and to investigate their impact on the process. Recently, RSM has been applied to the optimization of several water treatment processes, such as the electrochemical process (Wu et al., 2012; El-Ghenymy et al., 2012), ozonation (Zeng et al., 2012), ultrafiltration (Yuliwati et al., 2012), nano-porous membranes (Salahi et al., 2013) and biofilm reactors (Hasan et al., 2011; Muhamad et al., 2013). Since the optimization of phytoremediation pilot scale studies is rare, in this study we have adopted RSM using the Box-Behnken Design (BBD) to optimize the performance of the phytoremediation of wastewater contaminated with diesel. In the present study, we used horizontal subsurface flow constructed wetlands (HSFCWs) using Scirpus grossus to remediate water contaminated with different diesel concentrations with various aeration rate options for a period of 72 days in a field approach. We aimed to maximise the TPH removal efficiency from water and sand using RSM through a Box-Behnken experimental design by optimising the diesel concentration, retention time and aeration rate.
153
Fig. 1. Pilot scale constructed wetland design.
2.2. Analyses of total petroleum hydrocarbons (TPH) Three water samples from each constructed wetland (100 mL each) were collected periodically in clean schott duran bottle on each sampling day to extract TPH. To analyse the sand, three replicates of 10 g sand were collected at 5 cm depth from the top surface of sand layer periodically in clean schott duran bottle from each constructed wetland on the same sampling days to extract TPH. The TPH concentration measurements were carried out according to the methodology described in our previous study (AlBaldawi et al., 2013c).
2. Materials and methods 2.3. Optimization conditions with the Box-Behnken design 2.1. Pilot scale setup and operation The pilot scale constructed wetland system was designed as a tank made of fiberglass with a thickness of 0.5 cm. The dimension of each pilot scale constructed wetland was L180 W 90 H 90 cm. It was filled with coarse gravel (V 20e50 mm) with a height of 15 cm for the lowest layer, then 15 cm of fine gravel (V 10e20 mm) for the second layer and 20 cm of fine sand (V 0.5e1 mm) for the upper layer. A schematic diagram of the pilot scale constructed wetland and a complete substrate constructed wetland setup are depicted in Fig. 1. The pilot scale constructed wetland was operated in a horizontal subsurface flow system (HSFCWs) configuration. The pilot scale was run with three diesel concentrations and aeration rates for 72 days. Synthetic wastewater was prepared by mixing diesel with water in different concentrations (0.0, 0.1, 0.175, 0.25% Vdiesel/ Vwater). In the aerated constructed wetlands, air was supplied through air diffuser pipes located 20 cm below the bed surface. More details on the operations of the constructed wetlands can be referred in our previous study (Al-Baldawi et al., 2013b). A perennial native Malaysian plant, the bulrush S. grossus, was used in this study. Fifty plants were planted in each constructed wetland at a depth of 2e10 cm in the sand medium.
In the optimization study of the pilot scale system, the TPH concentration in water and sand was set using a BBD. The interaction between the main factors of diesel concentration, retention time and aeration rate, and the response of TPH removal efficiency in contaminated water and sand were investigated. The results were then analysed to develop an appropriate model for these factors. The response data of TPH removal from contaminated water was derived from the pilot scale study. The variability factors included in the design were diesel concentration (0.1, 0.175 and 0.25% Vdiesel/Vwater), retention time (14, 43 and 72 days) and aeration rate (0, 1 and 2 L/min) (Table 1). 2.4. Box-Behnken Design (BBD) The effects of diesel concentration (X1), retention time (X2) and aeration rate (X3) on TPH removal from contaminated water and sand was investigated using BBD. The design simulated 17 total experiment including five replicates to assess the error magnitude that occurs randomly. The dependent variable was TPH removal in water and sand. BBD was used for the statistical design of experiments and data analysis. In the optimization, the responses were
154
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
coupled to selected variables by linear or quadratic models. The mathematical relations between the responses and these variables can be represented by quadratic models Eq. (1) (Kayan and Gözmen, 2012):
Y ¼ b0 þ b1 x1 þ b2 x2 þ b3 x3 þ b12 x1 x2 þ b13 x1 x3 þ b23 x2 x3 þ b11 x21 þ b22 x22 þ b33 x23 þ ε
(1)
where Y is the response (TPH removal efficiency); x1, x2, and x3 represent the effect of the independent variables; x21, x22, x23 are the square effects; and x1x2, x1x3, and x2x3 are the interaction effects; b0 is constant; b1, b2 and b3 are the linear coefficients; b12, b13 and b23 are the interaction coefficients; and b11, b22, and b33 are quadratic coefficients; ε is the random error. The main aim of the design experiment was to optimize the response (Y) based on the factors investigated using the BBD model (Gan and Latiff, 2011; Kousha et al., 2012). Design ExpertÒ software (version 6.0.10, Stat-Ease, USA) was used to simulate the experimental run and optimize TPH removal in water and sand. The statistical significance of the first order model equation was determined by performing analysis of variance (ANOVA). Especially, a valid model must be significant based on the F-value and Pvalue in contrast to a lack of fit (insignificant). Furthermore, the proportion of variance shown by the multiple coefficients of determination, R2, should be close to 1 as possible to show a good correlation between the experimental and predicted values (Muhamad et al., 2013; Zeng et al., 2012). For optimization, the diesel concentration was set as the maximum, the retention time was set in the given range and the aeration rate was set as the minimum in order to maximize the removal of TPH from water and sand. The selected conditions were set since 1) the bulrush Scirpus grossus has the ability to tolerate toxicity due to high diesel concentrations; 2) the system can potentially be operated within the given time period of 72 days (Al-Baldawi et al., 2013b) and 3) we aimed to minimize the aeration rate for cost-saving reactor operation.
Table 1 Variables and their levels in the experimental design. Variables
Diesel concentration Retention time Aeration rate
Units
(Vdiesel/Vwater) % Day L/min
Symbols
X1 X2 X3
Code levels 1
0
þ1
0.1 14 0
0.175 43 1
0.25 72 2
The model was used to determine the TPH removal from water and sand in the pilot scale operation with optimal conditions. A normal plot of the residuals was analysed to check the normality assumption. The diagnostic of normal residuals demonstrated in Fig. 2 indicates that the residual behaviour followed a normal distribution and was quadratic, which is the more important assumption for checking statistical modelling. Therefore, the quadratic model built was adequate. The actual and predicted values of responses are listed in Table 2 with errors from 14.1e 11.5% between the actual and predicted values. Fig. 3 shows the predicted output values versus actual experimental values for TPH removal from water. From this figure, it can be noted that the values calculated using the predictive quadratic model were in good agreement with the experimental values with a satisfactory correlation between these values. Therefore, the developed model is suitable for predicting the efficiency of TPH removal from water under the investigated conditions. The results show that TPH removal from water and sand ranged from 33.3 to 88.3% and 18.9e67.3%, respectively. The maximum TPH
3. Results and discussion 3.1. Evaluation of simulated run by BBD BBD was selected to determine the relationship between the factors (X1, X2, X3) and the response function (Y). A quadratic model was built to fit the results, of which the coefficients were calculated by multiple regression analysis. The quadratic model obtained is shown in Eqs. (2) and (3).
Y1 ¼ þ75:08 þ 1:55*X1 þ 20:66*X2 þ 2:85*X3 þ 1:32*X12 11:43*X22 5:33*X32 0:91*X1 *X2 þ 1:03*X1 *X3 þ 0:94*X2 *X3
(2)
Y2 ¼ þ48:31 þ 1:66*X1 þ 15:90*X2 0:20*X3 þ 2:92*X12 10:46*X22 þ 5:86*X32 þ 2:12*X1 *X2 þ 5:93*X1 *X3 þ 5:70*X2 *X3
(3)
with, Y1 ¼ % TPH removal from water Y2 ¼ % TPH removal from sand X1 ¼ diesel concentration (Vdiesel/Vwater) X2 ¼ retention time (days) X3 ¼ aeration rate (L/min)
Fig. 2. Normal probability plot of studentized residuals of the quadratic model for TPH removal efficiency from: (a) water and (b) sand.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
155
Table 2 Box-behnken design matrix for three test factors in uncoded units along with the observed and predicted responses for TPH removal from water and sand. Run
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Variables
TPH from water
TPH from sand
X1: Diesel concentration (%v/v)
X2: Retention time (days)
X3: Aeration rate (L/min)
Actual value (%)
Predicted value (%)
Actual value (%)
Predicted value (%)
0.1 0.25 0.1 0.25 0.1 0.25 0.1 0.25 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175 0.175
14 14 72 72 43 43 43 43 14 72 14 72 43 43 43 43 43
1 1 1 1 0 0 2 2 0 0 2 2 1 1 1 1 1
39.8 47.6 84.1 88.3 72.2 70.3 69.7 72.0 33.3 71.5 43.2 85.2 78.4 85.4 61.0 83.6 67.0
45.6 48.7 86.9 90.0 63.4 66.5 69.1 72.2 44.3 85.6 50.0 91.3 67.8 67.8 67.8 67.8 67.8
28.1 24.6 52.7 57.7 58.7 52.7 49.6 67.3 33.8 57.2 18.9 65.0 59.8 35.1 40.0 49.6 57.1
30.0 33.3 61.8 65.1 46.1 49.4 45.7 49.0 31.8 63.6 31.4 63.2 47.5 47.5 47.5 47.5 47.5
removal efficiency from water (88.3%) was found in Run 4 under the experimental conditions of diesel concentration (X1 ¼ 0.25%), retention time (X2 ¼ 72 days) and aeration rate (X3 ¼ 1 L/min). Meanwhile, the maximum TPH removal from sand was observed in Run 8 with diesel concentration (X1 ¼ 0.25%), retention time (X2 ¼ 43 days) and aeration rate (X3 ¼ 2 L/min). These situations varied depending on the response required. Therefore, optimum process situation should be considered in order to obtain maximum TPH removal efficiency from water and sand.
ANOVA was performed to test the significance and adequacy of the model. The ANOVA results of the suggested model for TPH removal from water and sand are presented in Table 3 and indicate that the equation effectively represents the relationship between the response (TPH removal from water and sand) and the significant input variables. From the ANOVA (Table 3), the F-value for the model was more than 5.96 for TPH removal from water and 4.52 for TPH removal from sand. The high F-value indicates that most of the variation in the response can be explained by the regression equation, which implies that the terms in the model have a significant effect on the response. The associated P-value was used to estimate whether the F-value is large enough to indicate statistical significance. The P-values for TPH removal from water and sand were found to be lower than 0.05, indicating that the model is statistically significant at the 95% probability level. For the optimization of TPH removal, a “good fit” model is required to avoid poor or unclear results (Mayers et al., 2009). To ensure the adequacy of the proposed model, the R2, adjusted R2, F test, and a lack-of-fit test were evaluated. The ANOVA results showed significant (P < 0.05) response surface models with good R2 (0.8846) (Table 3) and adjusted R2 (0.73) values for the model Y1 and R2 (0.8853) and adjusted R2 (0.66) values for the model Y2, indicating the significance of the models (Khuri and Cornell, 1996). “Adeq Precision” measures the signal to noise ratio. A ratio greater than 4 is desirable. For the present study, the signal to noise ratio was found to be 7.44 (for TPH removal from water) and 6.62 (for TPH removal from sand), which indicates an adequate signal. Therefore, the quadratic model can be used to navigate the design space. Table 3 also presents the results of lack-of-fit tests for TPH removal from water and sand. The F-values of the lack-of-fit tests for the both models showed that variations in the data around the fitted model were not significant relative to the pure error, suggesting significant model correlation between the variable and process response. There were 84 and 92.6% chance that the lack-offit F-value could have occurred due to noise for TPH removal from water and sand. Therefore, the suggested model was statistically significant, and showed agreement between the observed and predicted responses. 3.2. Optimization of operational conditions
Fig. 3. Actual versus predicted values for TPH removal from: (a) water and (b) sand.
The target of optimizing the process was to find the optimum operation conditions leading to maximum TPH removal from water
156
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
Table 3 ANOVA analysis of the quadratic model to the TPH removal from medium. Source
Sum of squares
TPH removal from water Model 4207.451
Degrees freedom
Mean square
F-value
9
Prob > F
5.965
0.0140
Significant
0.245
0.6360
43.546
0.0003
0.826
0.3936
0.094
0.7684
7.015
0.0330
1.529
0.2562
0.042
0.8430
0.054
0.8223
0.045
0.8380
0.277
0.8403
Not significant
4.524 0.306 27.980 0.004 0.498 6.370 2.000 0.250 1.946 1.795
0.0296 0.5975 0.0011 0.9490 0.5032 0.0396 0.2002 0.6326 0.2057 0.2222
Significant
0.147
0.9262
Not significant
467.495 X1
19.182
1 19.182
X2
3413.144
1 3413.144
X3
64.755
1 64.755
X21
7.348
1 7.348
X22
549.856
1 549.856
X23
119.815
1 119.815
X1X2
3.311
1 3.311
X1X3
4.259
1 4.259
X2X3
3.531
1 3.531
Residual
548.656
7
Lack of fit
94.252
3
78.379 31.417 Pure error
454.404
4 113.601
Cor total 4756.107 R2 ¼ 0.8846 R2adj ¼ 0.7363 TPH removal from sand Model 2943.696 X1 22.108 X2 2023.009 X3 0.318 2 X1 36.008 X22 460.556 X23 144.579 X1X2 18.051 X1X3 140.695 X2X3 129.769 Residual 506.106 Lack of fit 50.353 Pure error 455.753 Cor total 3449.801 R2 ¼ 0.8853 R2adj ¼ 0.6647
16 Adeq precision ¼ 7.4387 9 1 1 1 1 1 1 1 1 1 7 3 4 16 Adeq precision ¼ 6.6236
327.077 22.108 2023.009 0.318 36.008 460.556 144.579 18.051 140.695 129.769 72.301 16.784 113.938
Values of ‘Prob > F’ less than 0.05 indicate model terms are significant.
and sand. The desirability function methodology was used for this optimization. When operating in the pilot scale, variables of diesel concentration and retention time were set within a certain range and the aeration rate was minimized. By using the function of
numerical optimization in the Design Expert software, we found a desirability of 0.883 for the maximum TPH removal efficiencies (Fig. 4). The maximum TPH removal from water and sand was 76.3% and 56.5%, respectively, at the optimized conditions of 0.25% diesel
Fig. 4. Desirability slope for numerical optimization conditions.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
157
3.3. The relationship between the response and variables under the optimum conditions
concentration (X1), 63 days retention time (X2) and 0 L/min aeration rate (X3) (Fig. 4). Validation experiments were carried under at the optimum conditions to confirm the predicted optimum response. The measured results under the optimum conditions were about 72.5% for TPH removal from water and 59% for TPH removal from sand. These results were very close to the predicted values, indicating the adequacy of the obtained model to optimize TPH removal from water and sand. The deviations between the measured and predicted values were within 5%. Therefore, it can be concluded that the regression models were appropriate in their reduced forms.
In this study, the aim of optimization was to find the conditions that provide maximum TPH removal from water in a phytoremediation process. The three dimensional (3D) response surface is a graphical representation of the regression function. The 3D response surface plots (Figs. 5 and 6) show how TPH removal from water (response variable Y1) and sand (response variable Y2) relate to the factors of diesel concentration (X1), retention time (X2) and aeration rate (X3) through quadratic model equations (Eqs. (2) and
Fig. 5. Response surface for TPH removal from water as a function of the varibles: (a) diesel concentration (X1) and retention time (X2), (b) diesel concentration (X1) and aeration rate (X3) and (c) retention time (X2) and aeration rate (X3).
Fig. 6. Response surface for TPH removal from sand as a function of the varibles: (a) diesel concentration (X1) and retention time (X2), (b) diesel concentration (X1) and aeration rate (X3) and (c) retention time (X2) and aeration rate (X3).
158
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159
(3)). Figs. 5a and 6a shows the 3D surface plot of the effect of diesel concentration (X1) and retention time (X2) on TPH removal from water and sand. It can be seen that a diesel concentration within the range of 0.1e0.25% (Vdiesel/Vwater) led to steady TPH removal from water and sand. This might due to the range of diesel concentrations selected were not significantly different. The effect of retention time (X2) was an increase in the response within 14e72 days. It is generally believed that the long retention time of a phytoremediation operation leads to high TPH removal. Figs. 5b and 6b depict the effect of diesel concentration (X1) and aeration rate (X3) on TPH removal from water and sand. It was observed that TPH removal from water and sand increased with an increase in the aeration rate. It was clearly shown that there was a steady increase in the TPH removal from water when the aeration rate was increased from 0 to 1 L/min and later gradually decreased when the aeration was further increased to 2 L/min; the aeration rate was effectively positive until 2 L/min for TPH removal from sand. The actual TPH removal from water was 88.3% with an aeration rate of 1 L/min, which is enough on the pilot scale with a subsurface flow system to enhance oxygen availability for biodegradation. However, the maximum actual TPH removal from sand (67.3%) was achieved with an aeration rate of 2 L/min because the contaminant degradation was enhanced in sand medium by supplying aeration to accelerate the aerobic biodegradation process (Seeger et al., 2011). Figs. 5c and 6c show the effect of retention time (X2) and aeration rate (X3) on TPH removal from water and sand. It was observed that TPH removal increased with an increase in the aeration rate, but the effect of aeration rate was not significantly different (P > 0.05) when applying a high level of 2 L/min. This demonstrates that longer retention time reduces more TPH concentration in water, but the diesel concentration and aeration rate do not have much effect on TPH removal efficiency. 4. Conclusions The optimization of a pilot constructed wetland model associated with the efficiency of TPH removal from water and sand was performed using the Box-Behnken Design. The results show that retention time was the most significant factor for the process, while no or little effect of diesel concentration and aeration rate was seen on TPH removal efficiency from water and sand. The coefficients of determination (R2) for the model equations (Y1 and Y2) were greater than 0.85 with the probability values (P < 0.05) demonstrating significance for the regression model to predict the responses. The experimental values were close to the predicted theoretical values, indicating that the models could be validated for the optimization of TPH removal from water and sand. The optimum conditions of the process were found to be at a diesel concentration of 0.25%, retention time of 63 days and no aeration with 76.3% and 56.5% maximum TPH removal from water and sand, respectively. These experimental results were found to be in good agreement with the predicted values obtained by the models. Therefore, the TPH phytoremediation process in constructed wetlands is a potential option technology for the treatment of wastewater contaminated with diesel. Since this study was performed batchwise, a future work will look on continuous operation of horizontal subsurface flow system constructed wetland. Acknowledgements The authors would like to thank Universiti Kebangsaan Malaysia (UKM-KK-03-FRGS0119-2010) and the Tasik Chini Research Centre for supporting this research project. They also acknowledge with thanks to the Iraqi Ministry of Higher Education and Scientific Research for providing a doctoral scholarship for the first author.
References Afferden, M., Rahman, K.Z., Mosig, P., De Biase, C., Thullner, M., Oswald, S.E., Müller, R.A., 2011. Remediation of groundwater contaminated with MTBE and benzene: the potential of vertical-flow soil filter systems. Water Res. 45, 5063e 5074. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013a. A phytotoxicity test of bulrush (Scirpus grossus) grown with diesel contamination in a free-flow reed bed system. J. Hazard. Mater. 252e253, 64e69. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013b. Effect of aeration on hydrocarbon phytoremediation capability in pilot sub-surface flow constructed wetland operation. Ecol. Eng. 61, 496e500. Al-Baldawi, I.A., Sheikh Abdullah, S.R., Suja’, F., Anuar, N., Idris, M., 2013c. Phytotoxicity test of Scirpus grossus on diesel-contaminated water using a subsurface flow system. Ecol. Eng. 54, 49e56. Chapelle, F.H., 1999. Bioremediation of petroleum hydrocarbon-contaminated ground water: the perspectives of history and hydrology. Ground Water 37, 122e132. Chen, Z., Wu, S., Braeckevelt, M., Paschke, H., Kästner, M., Köser, H., Kuschk, P., 2012. Effect of vegetation in pilot-scale horizontal subsurface flow constructed wetlands contaminated with a low and high chlorinated hydrocarbon. Chemosphere 89, 724e731. Demima, S., Drouiche, N., Aouabed, A., Benayad, T., Couderchet, M., Semsari, S., 2014. Study of heavy metal removal from heavy metal mixture using the CCD method. J. Ind. Eng. Chem. 20, 512e520. El-Ghenymy, A., Garcia-Segura, S., Rodríguez, R.M., Brillas, E., El Begrani, M.S., Abdelouahid, B.A., 2012. Optimization of the electro-Fenton and solar photoelectro-Fenton treatments of sulfanilic acid solutions using a pre-pilot flow plant by response surface methodology. J. Hazard. Mater. 221e222, 288e297. Fountoulakis, M.S., Terzakis, S., Kalogerakis, N., Manios, T., 2009. Removal of polycyclic aromatic hydrocarbons and linear alkylbenzene sulphonates from domestic wastewater in pilot constructed wetlands and a gravel filter. Ecol. Eng. 35, 1702e1709. Gan, C., Latiff, A.A., 2011. Extraction of antioxidant pectic-polysaccharide from mangosteen (Garcinia mangostana) rind: optimization using response surface methodology. Carbohydr. Polym. 83, 600e607. Garcia, J., Rousseau, D.P.L., Morato, J., Lesage, E.L.S., Matamoros, V., Bayona, J.M., 2010. Contaminant removal processes in subsurface-flow constructed wetlands: a review. Environ. Sci. Technol. 40, 561e661. Hasan, H.A., Abdullah, S.R.S., Kamarudin, S.K., Kofli, N.T., 2011. Response surface 2þ methodology for optimization of simultaneous COD, NHþ removal 4 -N and Mn from drinking water by biological aerated filter. Desalination 275, 50e61. Huang, X., El-Alawi, Y., Penrose, D.M., Glick, B.R., Greenberg, B.M., 2004. A multiprocess phytoremediation system for removal of polycyclic aromatic hydrocarbons from contaminated soils. Environ. Pollut. 130, 465e476. Kayan, Berkant, Gözmen, Belgin, 2012. Degradation of Acid Red 274 using H2O2 in subcritical water: application of response surface methodology. J. Hazard. Mater. 201e202, 100e106. Khuri, A.I., Cornell, J.A., 1996. Response Surfaces: Designs and Analyses, second ed. New York, Marcel Dekker. Konnerup, D., Koottatep, T., Brix, H., 2009. Treatment of domestic wastewater in tropical, subsurface flow constructed wetlands planted with Canna and Heliconia. Ecol. Eng. 35, 248e257. Kousha, M., Daneshvar, E., Dopeikar, H., Taghavi, D., Bhatnagar, A., 2012. Boxe Behnken design optimization of Acid Black 1 dye biosorption by different brown macroalgae. Chem. Eng. J. 179, 158e168. Liang, S.H., Kao, C.M., Kuo, Y.C., Chen, K.F., Yang, B.M., 2011. In situ oxidation of petroleum-hydrocarbon contaminated groundwater using passive ISCO system. Water Res. 45, 2496e2506. Mayers, R.H., Montgomery, D.C., Anderson-Cook, C.M., 2009. Response Surface Methodology: Process and Product Optimization Using Designed Experiments, third ed. John Wily & Sons, New Jersey. Muhamad, M.H., Sheikh Abdullah, S.R., Mohamad, A., Abdul Rahman, R., Kadhum, A.H., 2013. Application of response surface methodology (RSM) for optimisation of COD, NH3eN and 2,4-DCP removal from recycled paper wastewater in a pilot-scale granular activated carbon sequencing batch biofilm reactor (GAC-SBBR). J. Environ. Manage. 121, 179e190. Nivala, J., Headley, T., Wallace, S., Bernhard, K., Brix, H., Afferden, M., Müller, R.A., 2013. Comparative analysis of constructed wetlands: the design and construction of the ecotechnology research facility in Langenreichenbach, Germany. Ecol. Eng. 61, 527e543. Petrova, B., Budinova, T., Tsyntsarski, B., Kochkodan, V., Shkavro, Z., Petrov, N., 2010. Removal of aromatic hydrocarbons from water by activated carbon from apricot stones. Chem. Eng. J. 165, 258e264. Salahi, A., Noshadi, I., Badrnezhad, R., Kanjilal, B., Mohammadi, T., 2013. Nanoporous membrane process for oily wastewater treatment: optimization using response surface methodology. J. Environ. Chem. Eng. 1, 218e225. Seeger, E.V., Kuschk, P., Fazekas, H., Grathwohl, P., Kaestner, M., 2011. Bioremediation of benzene-, MTBE- and ammonia-contaminated groundwater with pilotscale constructed wetlands. Environ. Pollut. 159, 3769e3776. Souza, R.B.A., Ruotolo, L.A.M., 2013. Electrochemical treatment of oil refinery effluent using boron-doped diamond anodes. J. Environ. Chem. Eng. 1, 544e 551.
I.A.W. Al-Baldawi et al. / Journal of Environmental Management 140 (2014) 152e159 Stottmeister, U., Wießner, A., Kuschk, P., Kappelmeyer, U., Kästner, M., Bederski, O., Müller, R.A., Moormann, H., 2003. Effects of plants and microorganisms in constructed wetlands for wastewater treatment. Biotechnol. Adv. 22, 93e117. Vieira, P.A., Vieira, R.B., Faria, S., Ribeiro, E.J., Cardoso, V.L., 2009. Biodegradation of diesel oil and gasoline contaminated effluent employing intermittent aeration. J. Hazard. Mater. 168, 1366e1372. Wu, J., Zhang, H., Oturan, N., Wang, Y., Chen, L., Oturan, M.A., 2012. Application of response surface methodology to the removal of the antibiotic tetracycline by electrochemical process using carbon-felt cathode and DSA (Ti/RuO2eIrO2) anode. Chemosphere 87, 614e620.
159
Yuliwati, E., Ismail, A.F., Lau, W.J., Ng, B.C., Mataram, A., Kassim, M.A., 2012. Effects of process conditions in submerged ultrafiltration for refinery wastewater treatment: optimization of operating process by response surface methodology. Desalination 287, 350e361. Zeng, Z., Zou, H., Li, X., Sun, B., Chen, J., Shao, L., 2012. Ozonation of acidic phenol wastewater with O3/Fe(II) in a rotating packed bed reactor: optimization by response surface methodology. Chem. Eng. Process 60, 1e8. Zhang, B., Zheng, J., Sharp, R., 2010. Phytoremediation in engineered wetlands: mechanisms and applications. Procedia Environ. Sci. 2, 1315e1325.