Format And Type Fonts CHEMICAL ENGINEERING TRANSACTIONS VOL. 45, 2015 A publication of The Italian Association of Chemical Engineering www.aidic.it/cet Guest Editors: Petar Sabev Varbanov, Jiří Jaromír Klemeš, Sharifah Rafidah Wan Alwi, Jun Yow Yong, Xia Liu Copyright © 2015, AIDIC Servizi S.r.l., ISBN 978-88-95608-36-5; ISSN 2283-9216 DOI:10.3303/CET1545274 Please cite this article as: Doraiselvan K., Yusup S., Wai C.K., Muda N.S., 2015, Optimization studies on catalytic pyrolysis of empty fruit bunch (efb) using l9 taguchi orthogonal array, Chemical Engineering Transactions, 45, 1639-1644 DOI:10.3303/CET1545274 1639 Optimization Studies on Catalytic Pyrolysis of Empty Fruit Bunch (EFB) Using L9 Taguchi Orthogonal Array Kirenraj Doraiselvan a , Suzana Yusup* a , Cheah K. Wai a , Nur S. Muda a a Biomass Processing Lab, Centre of Biofuel and Biochemical, Green Technology Mission Oriented Research, Chemical Engineering Department, University Technology PETRONAS Bandar Seri Iskandar, 31750 Tronoh, Perak, Malaysia drsuzana_yusuf@petronas.com.my In this investigation, catalytic pyrolysis of empty fruit bunch (EFB) was studied and optimized in terms of bio-oil and char yields by using Taguchi L9 Orthogonal Array method. The effects of pyrolysis temperatures, catalyst loadings and particle sizes on the product yields were investigated and discussed in this paper. The catalytic pyrolysis is performed in a semi-batch reactor which is externally heated by an electrical vertical split tube furnace. Under the nitrogen flow rates of 60 ml/min, 80 mL/min and 100 mL/min, 15 g of EFB biomass with the particle sizes of 100 µm, 250 µm, 500 µm were thermally decomposed at three pyrolysis temperatures of 573 K, 673 K and 773 K, along with Zeolite HZSM-5 catalyst loadings of 1 wt%, 5 wt% and 12 wt%. From the product analysis, a maximum liquid bio-oil yield of 64.4 wt% were obtained at catalyst loading of 1.5 wt.%, reaction temperature of 773 K, nitrogen flow rate of 100 mL/min, and particle size of 250 µm. Meanwhile, at catalyst loading of 3.25 wt.%, reaction temperature of 773 K, nitrogen flow rate of 60 mL/min and particle size of 250 µm, the bio char yield formed was as low as 18.6 wt.% which could be mainly attributed to secondary cracking of char residue. 1. Introduction With the escalating global tensions associated with fossil fuel shortage and environmental consequences of an ever increasing consumption of non-renewable resources, many research programs are devoted in developing new renewable energy resources with the hope in replacing the depleting crude fuel energy. Biomass has been discovered as the most attractive renewable energy source in comparison to wind and solar because biomass can be converted into liquid, solid and gaseous fuels and a range of speciality chemicals (Choi et al., 2014). With the proper catalyst and biomass feedstock, liquid fuel with high calorific value and good hydrocarbon distribution can be thermally produced from catalytic cracking of carbon- carbon bonds and de-oxygenation reaction in pyrolysis process. Pyrolysis has been widely recognized as one of the most promising technologies for liquid bio-oil production along with the co-formation of solid char and gaseous products. Bio oil is a complex liquid mixture of water, ketones, sugars, furans, phenols, aldehydes, and guaiacols. Typically, it has to be further processed into liquid transportation fuels and value-added products, such as food flavorings, resins, fertilizer and fine chemicals, through cracking process. (Bridgwater and Peacocke, 1999). In recent years, much pyrolysis works have been focused on using various types of biomass resources such as corn stalk, food waste, pine wood biomass and others (Zhang et al., 2015). In the present study, Empty Fruit Brunch (EFB) is incorporated as the biomass feedstock due to its vast availability and quantity. In Malaysia, a total of 18 million fibrous fibres were produced as biomass waste throughout Malaysia in the year of 2010 and expected more for subsequence years (Abdullah and Gerhauser, 2008). Thus, an attempt on using EFB as the feedstock in thermal pyrolysis process is carried out with the hope of improving the yield of bio-oil produced from conventional biomass resources. To improve and optimize the bio-oil production, Taguchi’s Orthogonal Array method, a robust statistical and mathematical technique widely employed in many engineering applications, especially in manufacturing sector is incorporated into the present investigation. It is capable of studying multiple quality characteristic aspects and determining 1640 the most significant parameters of the processes. Unlike Response Surface Methodology (RSM), the orthogonal array method allows easier analysis with minimum number of experimental runs and the analysis result is valid over the entire region spanned by the control factors and its respective settings. 2. Materials and methods 2.1 Feedstock Preparation and Characterization Empty Fruit Bunch (EFB) was obtained from FELCRA Berhad Nasaruddin, Bota, Perak, Malaysia. The particle size of the EFB is reduced with a Cutting Mill and is sieved to a particle size of 100 µm, 250 µm, 500 µm. At 110 °C the samples are dried in an oven for more than 24 h. The proximate analysis of the bulk EFP is conducted using Thermogravimetry EXSTAR TG/DTA 6300 and ultimate analysis is carried out using LECO 932 CHNS Analyzer. Table 1: Characteristics of Empty Fruit Bunch (EFB) Proximate Analysis (wt%) Ultimate Analysis (wt%) Moisture 5.3 Carbon, C 45.01 Volatile Matter 71.00 Hydrogen, H 4.88 Fixed Carbon 16.80 Oxygen, O 49.02 Ash 6.90 Nitrogen, N 0.78 High Heating Value, HHV (MJ/Kg) 20.20 Sulphur, S 0.31 2.2 Catalyst Preparation and Characterization The zeolite HZSM-5 employed in this experimental work was obtained from Zeolyst International with the / mole ratio of 30 and surface area of 400 m 2 /g. Brunauer–Emmett–Teller (BET) specific surface area and mircropore volume of the catalyst samples were determined by using Surface Area Analyser Micromeritics ASAP 2021. For the preparation, the catalysts materials were calcined under nitrogen flow at a temperature of 773 K for 5 h in tube furnace reactor. 2.3 Experimental Setup Prior the experiment starts, the initial weight of empty borosilicate tube, glass wool, catalyst, condenser and rubber tube are weighed and recorded. 15 g of EFP biomass is introduced into the borosilicate tube before the glass wool and catalysts are filled in. The zeolite catalyst is weighed accordingly as indicated in the Table 2. With nitrogen gas flow of 500 mL/min, the tube is tighten and purged for 5 min in order to drive out all of the oxygen or residue air contents present in the tube. After 5 min of degassing, the nitrogen flow and the temperature are set at the desired flow rate and reaction temperature respectively with the flow controllers. For the heating rate, the furnace is kept heating up at 20 °C/min. After heating up the pyrolyser unit, the liquefied product is collected in the condenser unit at which the mixture of vapour and nitrogen carrier gas is passed through the ice bath before condensing. After completion of each run, the reactor is left to cool down to room temperature before analysis of product is conducted. After the experiment, the condenser, rubber tube, borosillicate tube with glass wool and the catalyst are weighed individually. Table 2: Experimental runs and corresponding parameters determined from L9 Taguchi Orthogonal Array Run Catalysis Loading (wt%) Temperature ( ) Nitrogen Flow Rate (mL/min) Particle Size (µm) 1 3.25 573 80 500 2 3.25 673 100 100 3 5 573 100 250 4 1.5 573 60 100 5 5 673 60 500 6 1.5 673 80 250 7 5 773 80 100 8 1.5 773 100 500 9 3.25 773 60 250 2.4 Bio-oil and Bio-char Yields Analysis For the product analysis, the weight of bio-oil formed is computed by the weight difference of the weight of the condenser before and after each complete run. To calculate the weight of bio-char produced, the total weight of the glass tube after reaction is deducted with the total weight of the glass tube, catalyst and 1641 glasswool before the reaction. With reference to Rahman et al (2014), the percentage of char and liquid product yields were defined as Eq(1). (1) In the context of experimental optimization, Design Expert 8.0.6.1 software, a statistical software which incorporates Taguchi’s L9 Orthogonal Array method is used in order to determine the optimized operating conditions. A total of 9 significant experimental runs is determined from the statistical software instead of 81 runs as the present study involves 4 reaction parameters (Catalyst Loadings, Temperature, Nitrogen Flow and Particles Size) and 3 levels each. These 9 experiment runs helped in the systematic approach for data analysis and identification of parameters that affect the oil production. Using the Design Expert software, the relationship between the variables is established and analysis is conducted for each parameter. The experiment is repeated at different operating conditions based on Taguchi L9 Orthogonal Array as shown in Table 2. To determine the effects of each variable on the product yields, a variance index, Signal-to-Noise (S/N) ratio is calculated by using the equations as shown as Eq (2) and (3). Among the three quality characteristics, “the bigger the better” is used as the determining factor to define and verify the liquid product yields, whereas “the smaller the better” is used to verify the solid product yields since the higher liquid yield and lower solid yield is desirable from the pyrolysis process. For maximizing the liquid product yield, the following definition of S/N ratio can be calculated as below:            r i i yrN S 1 2 11 log10 (2) For minimizing the solid yield, the following definition of S/N ratio can be calculated as below:            r i i r y N S 1 2 log10 (3) Where r is the number of tests in trial, yi is the experimental response at i th repetition. In order to evaluate the significance of each individual factor on product yields, the average S/N ratio for each factor f at level j is computed for each factor level as shown in Eq (4): 3 at level j factor i values for Sum of S/N N S  (4) 3. Results and discussion 3.1 Optimum Conditions by Taguchi L9 Orthogonal Array From Table 5, the maximum bio-oil yield was found at catalyst loading of 1.5 wt.%, temperature of 773 K, nitrogen flow rate of 100 mL/min and particle size of 250 µm with the yield as high as 64.43 wt.%. On another hand, it was also found that the lowest char residue was 18.6 wt% which occurred at catalyst loading of 1.5 wt.%, temperature of 773 K nitrogen flow rate of 60 mL/min and particle size of 250 µm. As can be seen from Table 3, the range of S/N ratio is the highest referring to particle size, followed by pyrolysis temperature, then nitrogen flow rate and catalyst loadings. Such ranking indicates that the particle size has a more pronounced effect on the liquid product yield and a small change in the particle size causes a larger influence on the liquid yield, resulting in larger S/N ratio range. Whereas, from Table 4, the factors affecting the solid yield have the opposite ranking as to that affecting the liquid product yields. This is in good agreement with the fact that the higher the liquid yields, the lower the solid yields. 3.2 Effect of Catalyst Loading and Particle Size on Bio-Oil and Residue Char Yields Based on the analysis result, pyrolysis of EFB produces the highest yield of bio-oil at intermediate catalyst loading of 1.5 wt.% and lowest yield of bio-char at catalyst loading of 3.25 wt.%. Such results are consistent with the work reported by Ahoa, et al. (2008) as increasing catalyst loadings lead to more 1642 biomass decomposition toward gas products due to the presence of strong bronsted acid sites on the H ZSM-5 surface. In the context of solid yield, the char yields decreases proportionally with higher catalyst loadings as the maximum char yield occurs at catalyst loading of 1.5 wt.% and the minimum char yield occurs at 3.25 wt.%. When the amount of catalysts loaded increases, more volatile matters are being catalyzed and reacted which contributed to a higher liquid product yield and consequently, a lower char yield. For the pyrolysis of EFB with particle size of 250 µm, it yielded a maximum liquid product of 35 wt.% before it dropped back to less than 20 wt.% at particle size of 500 µm. Such trend could be elucidated as the small biomass particles is more favorable to overheating and experienced secondary polymerization reaction which in turns resulted in high gas yield (Kamaroddin et al., 2014). Large particle size of 500 µm experienced large temperature gradient within the particles, which possibly leads to partial biomass decomposition and gives rise to an increase in the solid yields. Similar trends are reported in the research works made by Azri et al. (2009). 3.3 Effect of Reaction Temperature and Nitrogen Flow Rate on Bio-oil and Residue Char Yields For reaction temperature, the highest bio-oil yield and lowest char yield were obtained at the highest reaction temperature of 773 K which is in good agreement to that reported in Sulaiman (2011)’s work. At high temperature, more volatile matters are released from the biomass structure and decomposed before converted into the liquid product, which in turns resulted in the declining trend of char weight. The decrease in char yield could be due to primary decomposition of EFB at high temperature or by secondary decomposition of the char residue. In term of nitrogen flow rate, the maximum liquid bio-oil and bio-char was obtained at the highest flow rate of 100 mL. Table 3: Average S/N ratio of bio-oil yields at each factor and level Factors Levels S/N Average Range Rank Catalysis Loading (wt%) 1.5 32.4 4.7 4 3.25 27.7 5 31.6 Temperature ( ) 300 26.9 6.2 2 400 31.8 500 33.1 Nitrogen Flow Rate (mL/min) 60 30.9 5.3 3 80 27.7 100 33.1 Particle Size (µm) 100 20.9 11.8 1 250 32.8 500 27.2 Table 4: Average S/N ratio of solid yields at each factor and level Factors Levels S/N Average Range Rank Catalysis Loading (wt%) 1.5 -32.11 2.9 1 3.25 -31.57 5 -31.50 Temperature (°C) 300 -29.10 2.1 3 400 -31.93 500 -25.39 Nitrogen Flow Rate (mL/min) 60 -30.76 2.5 2 80 -29.16 100 -29.51 Particle Size (µm) 100 -32.10 1.9 4 250 -31.57 500 -31.50 1643 Table 5: Pyrolysis product yields based on Taguchi’s L9 Orthogonal Array estimations Run Catalysis Loading (wt%) Temperature (K) Nitrogen Flow Rate (mL/min) Particle Size (µm) Oil Yield (%) Char Yield (%) 1 1.5 573 60 100 32.40 40.30 2 1.5 673 60 100 44.13 41.23 3 1.5 773 60 100 50.03 34.57 4 3.25 573 60 100 21.17 30.57 5 3.25 673 60 100 32.90 31.50 6 3.25 773 60 100 38.80 24.83 7 5 573 60 100 28.10 32.73 8 5 673 60 100 39.83 33.67 9 5 773 60 100 45.73 27.00 10 1.5 573 80 100 28.37 43.20 11 1.5 673 80 100 40.10 44.13 12 1.5 773 80 100 46.00 37.47 13 3.25 573 80 100 17.13 33.47 14 3.25 673 80 100 28.87 34.40 15 3.25 773 80 100 34.77 27.73 16 5 573 80 100 24.07 35.63 17 5 673 80 100 35.80 36.57 18 5 773 80 100 41.70 29.90 19 1.5 573 100 100 42.30 48.30 20 1.5 673 100 100 54.03 49.23 21 1.5 773 100 100 59.93 42.57 22 3.25 573 100 100 31.07 38.57 23 3.25 673 100 100 42.80 39.50 24 3.25 773 100 100 48.70 32.83 25 5 573 100 100 38.00 40.73 26 5 673 100 100 49.73 41.67 27 5 773 100 100 55.63 35.00 28 1.5 573 60 250 36.90 34.07 29 1.5 673 60 250 48.60 35.00 30 1.5 773 60 250 54.50 28.33 31 3.25 573 60 250 25.67 24.33 32 3.25 673 60 250 37.40 25.27 33 3.25 773 60 250 43.30 18.60 34 5 573 60 250 32.60 26.50 35 5 673 60 250 44.30 27.43 36 5 773 60 250 50.23 20.77 37 1.5 573 60 500 23.07 35.33 38 1.5 673 60 500 34.80 36.27 39 1.5 773 60 500 40.70 29.60 40 3.25 573 60 500 11.83 25.60 41 3.25 673 60 500 23.57 26.53 42 3.25 773 60 500 29.47 19.87 43 5 573 60 500 18.77 27.77 44 5 673 60 500 30.50 28.70 45 5 773 60 500 36.40 22.03 46 1.5 573 80 250 32.87 36.97 47 1.5 673 80 250 44.60 37.90 48 1.5 773 80 250 50.50 31.23 49 3.25 573 80 250 21.63 27.23 50 3.25 673 80 250 33.37 28.17 51 3.25 773 80 250 39.27 21.50 52 5 573 80 250 28.57 29.40 53 5 673 80 250 40.30 30.30 54 5 773 80 250 46.20 23.67 55 1.5 573 80 500 19.03 38.23 1644 Table 6: Pyrolysis product yields based on Taguchi’s L9 Orthogonal Array estimations (Cont.) Run Catalysts Loading (wt%) Temperature (K) Nitrogen Flow Rate (ml/min) Particles Size (µm) Oil Yield (%) Char Yield (%) 56 1.5 673 80 500 30.77 39.17 57 1.5 773 80 500 36.67 32.50 58 3.25 573 80 500 7.80 28.50 59 3.25 673 80 500 19.53 29.43 60 3.25 773 80 500 25.43 22.77 61 5 573 80 500 14.73 30.67 62 5 673 80 500 26.47 31.60 63 5 773 80 500 32.37 24.93 64 1.5 573 100 500 32.97 43.33 65 1.5 673 100 500 44.70 44.27 66 1.5 773 100 500 50.60 37.60 67 3.25 573 100 500 21.73 33.60 68 3.25 673 100 500 33.47 34.53 69 3.25 773 100 500 39.37 27.87 70 5 573 100 500 28.67 35.77 71 5 673 100 500 40.40 36.70 72 5 773 100 500 46.30 30.03 73 1.5 573 100 250 46.80 42.07 74 1.5 673 100 250 58.53 43.00 75 1.5 773 100 250 64.43 36.33 76 3.25 573 100 250 35.57 32.33 77 3.25 673 100 250 47.30 33.27 78 3.25 773 100 250 53.20 26.60 79 5 573 100 250 42.50 34.50 80 5 673 100 250 54.23 35.40 81 5 773 100 250 60.13 28.77 4. Conclusion With the incorporation of Taguchi’s Orthogonal Array Method, catalytic pyrolysis of palm oil waste biomass, Empty Fruit Bunch (EFB) was successfully optimized with the highest oil yield of 64.4 wt% and lowest char yield of 18.6 wt. The highest bio-oil yield was obtained at an optimum temperature of 773 K with the catalyst loading of 1.5 wt% and nitrogen flow rate of 100 mL/min as well as particle size of 250 µm. While, the lowest bio-oil char yield is achieved under temperature of 773 K, catalyst loadings of 1.5 wt%, nitrogen flow rate of 60ml/min and particle size of 250 µm. The most significant factor affecting the bio-oil yield is particle size, followed by pyrolysis temperature, then nitrogen flow rate and catalyst loadings. Acknowledgement The authors would like to thank Ministry of Higher Education for funding the research through LRGS and University Technology PETRONAS for the support. References Abdullah N., Gerhauser H., 2008, Bio-oil derived from empty fruit bunches, Fuel, 87, 2606- 2613. Choi J.H., Woo H.C., Suh D.J., 2014, Pyrolysis of Seaweeds for Bio-oil and Bio-char Production, Chemical Engineering Transactions, 37, 121-126. Bridgwater A.V., Peacocke G.V.C., 1999, Fast pyrolysis processes for biomass, Renewable and Sustainable Energy Reviews, 4, 1-73. Rahman A.A., Abdullah N., Sulaiman F., 2014, Temperature effect on the characterisation of pyrolysis products from oil palm fronds, Advance in Energy Engineering (AEE), 2, 14-21. Sulaiman F., Abdullah N., 2011, Optimum conditions for maximising pyrolysis liquids of oil palm empty fruit bunches, Energy, 36, 2352-2359. Zhang B., Zhong Z., Min M., Ding K., Xie Q., Ruan R., 2015, Catalytic fast co-pyrolysis of biomass and food waste to produce aromatics: Analytical Py–GC/MS study, Bioresource Technology, 189, 30-35.