Modeling of essential oil extraction process: application for orange, pomelo, and lemongrass - Nguyen Dang Binh Thanh

Vietnam Journal of Science and Technology 56 (4A) (2018) 182-189 MODELING OF ESSENTIAL OIL EXTRACTION PROCESS: APPLICATION FOR ORANGE, POMELO, AND LEMONGRASS Nguyen Dang Binh Thanh, Nguyen Trung Dung, Ta Hong Duc* Hanoi University of Science and Technology, No. 1 Dai Co Viet road, Ha Noi *Email: duc.tahong@hust.edu.vn Received: 17 July 2018; Accepted for publication: 6 October 2018 ABSTRACT In this study, the kinetic models of steam distillation of orange (Citrus sinensis (L.) Osbeck), pomelo (Citrus grandis L.), and lemongrass (Cymbopogoncitratus) for the recovery of essential oils were developed. The model parameters wereestimated based on experimental data and comprehensive kinetic mechanismsof the solid-liquid extraction process. Numerical results showed that, the extraction mechanism of the three materials were best fit to the Patricelli two- stage model in which the diffusion of the oil was followed by the washing step. Moreover, the model parameters obtained from the measured data reflected clearly the nature of the two-stage extraction at which the kinetic rate of the washing step (surface extraction) was higher than that of in-tissue diffusion step. Thus, the kinetics of the extraction processes obtained from the present work could be usedfor the scale-up of the extraction process operating at a large scale and for the purpose of process control as well. Keywords: essential oil, steam distillation, modeling, optimization, kinetic. 1. INTRODUCTION Essential oilsextracted from sweet orange (Citrus sinensis (L.) Osbeck), pomelo (Citrus grandis L.), and lemongrass (Cymbopogoncitratus) have useful components in the production of food, pharmacy and perfume industries [1, 2, 3]. According to literature, essential oils can be extractedby different methods, from traditional techniques to novel techniques such as solvent extraction, steam distillation, hydrodistillation, microwave extraction, ultrasound extraction and supper critical CO2 extraction. Each method hasits own advantages and disadvantages [4, 5]. However, the steam distillation is commonly used for essential oil extraction due to safety, simplicity and environmental-friendly operations. In order to carry out a production line at a large scale, mathematical modeling is often considered as an inventive step. Mathematical modelscan help the design, optimization and control of a processby lowering the cost of trials and experiments [6]. Thus, mathematical modeling plays an important role in the selection of process conditions. Several theoretical, empirical, and semi-empirical models were reported for the solid-liquid extraction of bioactive substances from plant materials. Most studies were developed based onthe type ofone-stage Steam distillation modeling for essential oil extraction process 183 model. However this model type is not sufficient to captureall mechanisms of the extraction process due to the complicated nature of the plant oil deposition. Therefore, the objective of this work is to examine several two-stage models (washing and diffusive stage) and propose the best one that reflects well experimental data.The content of this study includes experimental conduction and kinetic modeling of essential oil steam distillation applied to sweet orange, pomelo, and lemongrass. 2.MATERIALS AND METHODS 2.1.Experimental lab-scale system An experimental lab-scale system for steam distillation is shown in Figure 1. The designed capacity of the still (stripper) was10 –15 kg per batch depending on type of plants (Figure 1b). In this work, sweet orange and pomelo peels were taken from Tra Vinh province while lemongrass was brought from Quang Nam. For each run, 75 liter of purified water was initially loaded to the stripping vessel and the raw materials were chopped up to the average size of 5 – 10 mm prior to the loading step. During the extraction, the mixture of thetreated plant materials and water was heated using a 6 kW electrical resistant heater and the system was operated at atmospheric pressure. The equipment was fitted with a tight lid to prevent oil and vapor from leaking out. The system is operated in a manner that the steam rising from the still strips the oil away from the plant materials and the vapor comprised of oil and steam is passed to a condenser where the vapor phase is condensed and separated. In the decanter (the oil–water separator), the essential oil is separated from water at the top of the separator since the density of the oil is lighter than that of the remaining liquid. Figure 1.(a) A typical diagram of a steam distillation system, (b) A photo of a lab-scale steam distillation system: (1) Water, (2) Steam Water, (3) Plant material, (4) Steam and essential oil, (5) Cold water, (6) Hot water, (7) Water and Essential oil, (8) Separator, (9) Essential oil, (10) Water. The solid – liquid extraction was carried out for atotal of 160 minutes in which the oil recovery was measured at proper extraction time for kinetic study. Accumulated oil yield obtained from the experiments was recorded for the analysis of oil recovery. Composition of the oil obtained from the extraction of each raw material was analyzed by Gas Chromatography- Mass Spectrometry (GC-MS) on a capillary column (30 m, 0.32 mm i.d., 0.25 µm film thickness). Temperature of the column was initially set to 40 oC for 2 min, and then gradually increased to 225 oC at the rate of 4 oC/min. The extracted oil was diluted by acetone 99.99 % at Nguyen Dang Binh Thanh, Nguyen Trung Dung, Ta Hong Duc 184 the volumetric ratio of 3:100. Temperature of the injector and detector was set at 290 and 175 oC, respectively. The carrier gas (Helium gas) flow was maintained at the rate of 2.2 mL/min and the split ratio was 1:100. 2.2. Mathematical models Kinetics modeling of solid – liquid extraction required an understanding of extraction mechanism. In the Table 1, several two-stage modeling studies have been conducted to describe extraction of different substances from various materials[7]. The extraction yield is obtained using the following equation:
              (1) Table 1. Two-stage models for the extraction of plant materials [7]. No. Model Equation Parameter 1 Parabolic Diffusion Model
    √(T1) K1 – washing kinetic coefficient K2 – diffusive kinetic coefficient 2 Elovich Model
     ! (T2) K1 – washing kinetic coefficient K2 – diffusive kinetic coefficient 3 Patricelli Model
  "#1 % &'(%)  *#1 % &'(%)(T3) K1, K2 – kinetic coefficient for the washing and the diffusion stage A, B – final yield for washing and diffusion stage 4 So and Macdonald Model
  "#1 % &'(%)  *#1 % &'(%)  +#1 % &'(%,)(T4) K1, K2, K3 – kinetic coefficient for washing, first diffusion and second diffusion stage A, B, C – final yield for washing, first diffusion, and second diffusion stage 2.3. Statistical analyses Mathematical modeling of the solid liquid extraction required the statistical methods of regression and correlation analysis for the model verification. The validation of models could be judgedon the basis of different statistical methods. The most widely used method in literature was root mean square error (RMSE) analysis, which was determined as follows. -./0  1∑ 3453678  (2) The concordance between the experimental data and calculatedvalues were also examined by the coefficient of determination (R2), -  1 % ∑ 345367398∑ 34:67398 (3) Steam distillation modeling for essential oil extraction process 185 where Yi – experimental value of the yield;
5 – predicted value of the yield using the regression model;
: - arithmetic average value of the experimental yield; n – number of experimental points. 3. RESULTS AND DISCUSSION 3.1. Experimental data Experiments of the essential oil extraction from each raw material type (sweet orange, pomelo, and lemongrass) were carried out on the lab-scale system mentioned above. For each batch experiment, the extracted oil volume with respect to extraction time was recorded so that the oil yield could be estimated as a function of processing time. Total extraction of each raw material type (sweet orange, pomelo, and lemongrass) was conducted in 150 min starting from the first liquid drop obtained at the decanter. Details of the experimental data were given in Table 2 and the description of the experimentation can be found elsewhere [5]. Table 2. Experimental data of oil recovery from the extraction of sweet orange, pomelo, and lemongrass. Extraction Time (min) Sweet Orange Pomelo Lemongrass Extracted Oil (mL) Oil Yield (-) Extracted Oil (mL) Oil Yield (-) Extracted Oil (mL) Oil Yield (-) 0 0 0 0 0 0 0 10 6.5 0.333 4.5 0.529 3.0 0.4 20 12 0.615 6 0.706 4.5 0.6 30 15 0.769 6.5 0.765 5.0 0.667 60 - - 8.0 0.941 6.5 0.867 90 18.5 0.949 8.5 1.0 7.0 0.933 120 19.5 1.0 8.5 1.0 7.5 1.0 135 19.5 1.0 - - - - 150 19.5 1.0 8.5 1.0 7.5 1.0 Measurement results showed that, at the first period of the extraction (about 20 min) the oil yield increased significantly with time. Then, at the second step, the extraction rate tended to decrease. According to these phenomena, it can be explained that, at the initial step, oil deposited on the surface of the raw material was washed and entrained by the steam. This step often occurred in a short time which accounted for the instant washing step. The rest part of the oil deposited in the plant’s tissues was extracted at a lower rate due to the nature of the desorption mechanism. Thus, the essential oils recovered from plants can be captured well by the two-stage soli-liquid extraction. The composition of each essential oil calculated from GC-MS analysis were given in Table 3. It can be seen that D-Limonene was the major component of the essential oils extracted from sweet orange (95.59 %) and pomelo (82.54 %) since these two materials come from the same family. Farhat et al. [3] also reported that the composition of Limonene in the oil extracted from Nguyen Dang Binh Thanh, Nguyen Trung Dung, Ta Hong Duc 186 orange peel by steam distillation and microwave steam distillation was around 95% and in the work of Chen et al. [8], Limonene concentration obtained from microwave extraction of pomelo peel was in the range of 78 – 87 %. In the case of lemongrass oil, the total content of Citral (including citronellal, citronellol, and geraniol) is around 69 % (see Table 3). Cassel et al. [9] reported in one of their work that, citral concentration of the lemongrass oil extracted by steam distillation is 63.5 %, while the total content of this component was in the range of 73 – 85 % in a study of Desai et al. [10]. Table 3. Composition of the essential oils extracted from sweet orange, pomelo, and lemongrass. Sweet Orange Pomelo Lemongrass Compound Content (%) Time (min) Compound Content (%) Time (min) Compound Content (%) Time (min) α-pinene, (-)- 0.51 5.27 α-Pinene 1.27 5.29 Limonene 3.295 9.685 2-β-pinene 0.10 6.38 β-Myrcene 1.50 6.69 Citronellal 31.043 13.84 β-Myrcene 1.65 6.64 α-Phellandrene 1.48 7.15 Citronellol 10.003 15.92 D-Limonene 95.59 7.771 D-Limonene 82.54 7.79 Geraniol 27.864 17.06 Octanal 0.20 30.61 γ-Terpinene 8.41 8.59 Dibutyl phthalate 0.70 31.44 Nootkatone 1.09 28.41 3.2. Kinetic model parameters The four previously described models (see Table 1) were tested for the extraction ofsweet orange, pomelo, and lemongrass. Tables 4, 5, and 6 showed the corresponding results of nonlinear regression and statistical analyses for the development of the kinetic models. Numerical calculations showed that Patricelli model was the best fit for all materials (pomelo, sweet orange and lemongrass) selected in this study. It can be observed that the Patricelli model has high coefficient of determination R2 = 0.993 and low value of RMSE (RMSE = 0.023) for pomelo; R2 = 0.997 and RMSE = 0.017 for sweet orange; and R2 = 0.999, RMSE = 0.011 for lemongrass. Table 4. Coefficients and statistical parameters of extraction modeling for sweet orange. Model Coefficients RMSE R2 K1 K2 K3 A B C Parabolic Diffusion 0.287 0.064 - - - - 0.072 0.946 Elovich 0 0.204 - - - - 0.051 0.973 Patricelli 0.048 0.0002 - 0.964 1.089 - 0.017 0.997 So and Macdonald 0.0001 0.048 0 0.871 0.964 1.968 0.018 0.997 Steam distillation modeling for essential oil extraction process 187 Table 5. Coefficients and statistical parameters of extraction modeling for pomelo. Model Coefficients RMSE R2 K1 K2 K3 A B C Parabolic Diffusion 0.525 0.043 - - - - 0.056 0.962 Elovich 0.187 0.170 - - - - 0.031 0.988 Patricelli 0.131 0.014 - 0.648 0.424 - 0.023 0.993 So and Macdonald 0.107 0.0005 0 0.870 1.974 4.984 0.048 0.972 Table 6. Coefficients and statistical parameters of extraction modeling for lemongrass. Model Coefficients RMSE R2 K1 K2 K3 A B C Parabolic Diffusion 0.286 0.064 - - - - 0.048 0.972 Elovich 0 0.203 - - - - 0.028 0.990 Patricelli 0.140 0.023 - 0.359 0.671 - 0.011 0.999 So and Macdonald 0.995 0.039 0.001 0.173 0.705 0.714 0.014 0.998 Figure 2. Extraction kinetics of pomelo, sweet orange, and lemongrass. 0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 140 160 Ex tr a ct io n yi el d (-) Time (min) Pomelo: Exp. Sweet Orange: Exp. Lemongrass: Exp Pomelo: Patricelli Model Sweet Orange: Patricelli Model Lemongrass: Patricelli Model Nguyen Dang Binh Thanh, Nguyen Trung Dung, Ta Hong Duc 188 Experimental data of the oil yield obtained from experiments were depicted in comparison with predicted model in Figure 2. It can be seen that, the Patricelli model captured well the two- stage extraction mechanism of theoil deposition in the plants. For the extraction of pomelo and sweet orange, the washing stage was more important than diffusion stage. This may be explained that the amount of essential oil located on the surface of the skin was higher than the oil deposited inside the plant tissues. For instance, the oil deposited on the skin surface of sweet orange peels occupied more than 96 % of the total oil yield (see Table 2). However, due to the fiber structure of lemongrass, the amount of in-cell oil (67.1 %) was higher than the oil deposited on the cell surface (35.9 %).In addition, numerical results given in Table 2 also showed that the extraction rate of the surface oil was higher than that of in-tissue oil since the values of K1 were always higher than that of K2 for all selected materials (sweet orange, pomelo, and lemongrass). Details of kinetic models for the extraction of selected materials in this work were described in Equations (4), (5), and (6) as follows. For sweet orange:
  0.964#1 % &'(%0.048)  1.089#1 % &'(%0.0002) (4) For pomelo:
  0.648#1 % &'(%0.131)  0.424#1 % &'(%0.014) (5) For lemongrass:
  0.359#1 % &'(%0.139)  0.671#1 % &'(%0.023) (6) 4. CONCLUSIONS Kinetics of theessential oil extraction from different plants (sweet orange, pomelo, and lemongrass) using steam distillation were developed on the basis of semi-theoretical models. The results showed that all models selected are in a good agreement with experimental data. Howerver, the Patriicelli model, in which both washing and desorption steps were accounted for, can capture wellthe extraction kinetics of all materials (sweet orange, pomelo, and lemongrass) considered in the present work. The proposed mathematical models can be useful for the process design of large scale systems and for the purpose of process control. Acknowledgements. This work is funded by the Hanoi University of Science and Technology (HUST) under project T2017-PC-020. REFERENCES 1. Balti M. A., Hadrich B., Kriaa K., and Kechaou N. - Lab-scale extraction of essential oils from Tunisian lemongrass (Cymbopogon flexuosus), Chemical Engineering and Processing: Process Intensification 124 (2018) 164-173. 2. Desai M. A., and Parikh J. - Extraction of essential oil from leaves of lemongrass using microwave radiation: optimization, comparative, kinetic, and biological studies, ACS Sustainable Chemistry & Engineering 3 (2015) 421-431. 3. Farhat A., Fabiano-Tixier A.S., Maataoui M. E., Maingonnat J.F., Romdhane M., and Chemat F. - Microwave steam diffusion for extraction of essential oil from orange peel: Kinetic data, extract’s global yield and mechanism, Food Chemistry 125 (2011) 255-261. Steam distillation modeling for essential oil extraction process 189 4. Carlson L. H. C., Machado R. A. F., Spricigo C. B., Pereira L. K., and Bolzan A. - Extraction of lemongrass essential oil with dense carbon dioxide, The Journal of Supercritical Fluids 21 (2001) 33-39. 5. Thanh N. D. B., Duc T. H., and Dung N. T. - Kinetics and modeling of oil extraction from vietnam lemongrass by steam distillation, Vietnam Journal of Science and Technology 55 (2017) 58-65. 6. Cassel E., Vargas R. M. F., Martinez N., Lorenzo D., and Dellacassa E. - Steam distillation modeling for essential oil extraction process, Industrial Crops and Products 29 (2009) 171-176. 7. HadrichB., DimitrovK., and Kriaa K. - Modelling investigation and parameters study of polyphenols extraction from Carob (Ceratonia siliqua L.) using experimental factorial design, Journal of Food Processing and Preservation 41 (2017) 1-10. 8. Chen Q., Hu Z., Yao F. Y. D., and Liang H. - Study of two-stage microwave extraction of essential oil and pectin from pomelo peels, LWT - Food Science and Technology 66 (2016) 538-545. 9. Cassel E.and VargasR. M. F. - Experiments and modeling of the Cymbopogon winterianus essential oil extraction by steam distillation,Journal of Mexican Chemical Society 50 (2006) 126-129. 10. Desai E., Parikh J., and DeA K. - Modelling and optimization studies on extraction of lemongrass oil from Cymbopogon flexuosus (Steud.) Wats, Chemical Engineering Research and Design 92 (2014) 793-803.