CASTOR OIL TRANSESTERIFICATION-EXPERIMENTAL AND MODELING STUDIES

Abstract

Energy demand is growing at a fast pace probably in every country of the word. This is because of increase in population in some parts of the globe and also due to enhanced impetus on the development of infrastructure of all kinds and related human related services. The demand of energy driven sectors, are mostly met by fossil fuels namely Petroleum, Natural Gas and Coal whose reserves are constantly depleting. Use of fossil fuels also results into adverse impact on environment and climatic conditions (Wang et al., 2012). Therefore, there has been a growing interest in searching alternative renewable and eco-friendly sources of energy (Caballero and Guirardello, 2013). It is expected that these alternative sources can replace fossil fuels to a large extent or can at least supplement them. Due to this reason, Biodiesel has become more attractive alternative and sustainable fuel source (Balat and Balat, 2010). Biodiesel is made from renewable biological sources namely vegetable oils and animal fats (Ma and Hanna, 1999). Biodiesel is a nontoxic, biodegradable and environmental friendly fuel. It also provides lower hydrocarbon and carbon monoxide emissions, higher cetane number, less smoke and particulate matter (Atadashi et al., 2012).Generally, vegetable oils or animal fats can be easily converted to biofuel by transesterification reaction with an alcohol in the presence of a catalyst. The main product is mixture of fatty acid alkyl ester (FAAE), called as biodiesel (Badday et al., 2014), and glycerol is produced as a byproduct (Fayyazi et al., 2013). Leung et al. (2010) in an excellent review has described various vegetable oil feedstocks, which are used for producing biodiesel. These may be classified as edible vegetable oils, non-edible vegetable oils, and others. Edible oils include Soybean, Rapeseed, Sunflower, Palm, Peanut, Corn, Camelina, canola, cotton, pumpkin, while non-edible oils are Jatropha, Pongamia, Sea mango, Palanga, Tallow, Nile tilapia, poultry, and used cooking oils fall in the category of others. Biodiesel produced from these edible oils is a suitable substitute for diesel fuel. In India as well as in other countries, edible oils are not recommended to be used as raw material for biodiesel. Its use is likely to increase the cost of edible oils, which are used by human beings. Therefore, it is desirable and preferred to use non-edible oils for this purpose. ii Leung et al. (2010) have pointed out in their review that significant amount of research work has been done on non-edible oils namely, Jatropha curcas, Pongamia pinnata, Sea mango, Palanga, Tallow, Nile tilapia etc and used cooking oil. Several excellent reviews have been published recently on Biodiesel from vegetable oils. These are due to Issariyakul and Dalai (2014), Atabani et al. (2013), and Lin et al. (2011). After critical assessment of published work, it has been observed that relatively less amount of research work has been done on the transesterification of castor oil- a non-edible vegetable oil. About castor oil, Berman et al. (2011) states "Castor oil is one of the most promising non-edible oil crops, due to its high annual seed production and yield, and since it can be grown on marginal land and in semiarid climate". Scholz and Silva (2008) have reviewed the "Prospects and risks of castor oil as a fuel", and recommended that a better option possibly is its transesterification and addition of the product biodiesel to fossil diesel fuel. Similar view point has been expressed by Shrirame et al. (2011). Berman et al. (2011) have evaluated the fuel related properties of castor oil Biodiesel and recommended its use for blending with diesel with a maximum limit of 10 %. Therefore, it was thought proper to conduct focused review of literature on castor oil transesterification related to process development, property estimation, cost estimation, reaction kinetics and optimization. Main observations are given below. (i) Most of the research papers are from the year 2010 and onwards. (ii) Ethanol and methanol both have been used for transesterification. With ethanol several catalysts namely, KOH, NaOH, KOCH3, NaOCH3, H2SO4, HCl, K2CO3 and CaCO3, (iii) In-situ transesterification of castor oil seeds has also been done. have been used. However sodium methoxide, tin (IV) complex, solid potassium and cesium salts of 12-Tunstophosphoric acid have been used as catalyst with methanol. (iv) Transesterification under supercritical conditions has also been investigated. (v) Transesterification with methanol using basic catalyst (KOH, NaOH, KOCH3, NaOCH3, NaOC2H5 (vi) Commonly used molar ratio of methanol to oil varies between 4:1 to 12.5:1. Higher molar ratios 50:1, 225:1 and 250:1 have also been used. ) has been studied. iii (vii) Operating temperature varies from 25° - 65 °C, but in one study, experiments have also been conducted at 80 °C. (viii) Experimental results have been utilized to obtain kinetic rate constants of kinetic models of first order irreversible, and second order irreversible and reversible reactions. Most of the researchers have assumed pseudo first order kinetics. (ix) Design of experiment and Response Surface Method (RSM) has been used to obtain optimized conditions of transesterification reaction. It may be concluded from the above summary that the transesterification of castor oil with methanol in the presence of acid catalyst have not been investigated from the point of view of determining reaction kinetics, and optimized operation conditions, parameters of kinetic models. For using the alkaline catalyst it is desired that the FFA content in the oil should be less than 1% (Tiwari et al., 2007). In case FFA content exceeds this limit, the soap formation may occur which should inhibit the suppression of ester (Canakci and Gerpen, 2001). In the present case FFA content in castor oil is more than 1%, so acid catalyst have been chosen for castor oil transesterification. In view of the above discussion, experimental studies on castor oil transesterification have been planned to be conducted in this thesis in two types of Lab Reactors, Small Lab Reactor (500 cc) and the Large Lab Reactor (3 L), with sulphuric acid as a catalyst. Main emphasis is on kinetic modeling, use of RSM and Artificial Neural Network (ANN) for determining optimized conditions, and kinetic parameters. The results are summarized below: Small lab reactor: The operating conditions used for experimental studies were methanol/oil molar ratio = 6:1, catalyst =1 % concentrated H2SO4 (% v/v of castor oil), temperature = 35° to 65 °C at an interval of 5 °C, and 600 RPM. Experimental results have been analyzed with respect to three kinetic schemes namely, first order pseudo irreversible reaction, second order irreversible reaction, and the reversible reaction. iv (a) By using first order pseudo irreversible reaction kinetics, and experimental data, rate constant k (mm-1 Activation Energy, E = 38.283 kJ/mol ; Arrhenius constant, A= 1461.0345 min) have been computed at various temperatures. This provides the values of activation energy, and Arrhenius constant as given below. (b) Analysis of experimental data by second order irreversible reaction kinetics provides the following values : -1 Activation Energy, E = 38.611 kJ/mol ; Arrhenius constant = 343836.48 ml/ (mol.min) Fitting of irreversible second order kinetic model is somewhat better than that of irreversible first order reaction with respect to correlation coefficients. (c) Kinetic model for reversible reaction with forward reaction as pseudo first order, and backward reaction as second order has also been used to analyze the experimental results and the obtained model is as given below: Where Xc = fractional formation of FAME; k1(min-1) = 216.264 exp ; Gas constant, R = ; Temperature T = K Equilibrium constant, K = ; and K35 = 0.008895 This kinetic model is applicable in the temperature range 35° to 60 °C. (d) This transesterification reaction is endothermic in nature, and its heat of reaction, computed in this work is 23.560 kcal/g mol. (e) In order to evaluate the usability of FAME (biodiesel) produced, its several properties have been experimentally determined. From these properties, it is concluded that the biodiesel product formed should be used for blending with diesel oil in appropriate quantity to bring its properties within acceptable limits as viscosity of FAME product and its water content both are on higher side. These conclusions are in accordance with pervious similar findings reported in the literature [Canoira et al. (2010)]. v Large Lab Reactor: The experimental data taken in this reactor have been analyzed by response surface Methodology (RSM), and Artificial Neural Networks (ANN). Main Conclusions are: RSM Modeling: The central composite design (CCD) of the RSM was used to decide the number of experiments, to be conducted. Range of operating conditions was: Methanol to oil ratio = 6:1 to 25:1, catalyst amount (vol. %) = 1 to 3, temperature (°C) = 40 to 60. An experiment was conducted up to 4 hours duration. Samples for analysis were taken in between at regular intervals. A RSM model has been developed by using experimental data of all sets at 4 hour time only. The model in terms of coded variables for predicting % FAME yield for given values of methanol to oil molar ratio, catalyst amount, and the temperature. ANOVA has been used to evaluate the adequacy of the RSM model. The model has also been validated with two additional sets of experimental data. The model predictions are within ± 5 % deviation with respect to experimental results. RSM model has been used to optimize the experimental conditions. These are methanol to oil ratio = 25:1, catalyst amount = 3 vol %, temperature = 60 °C. RSM model yields % FAME yield as 75.67 % while that obtained experimentally is 76.95 %. Using the model, the effects of variation in operating conditions on % FAME yield have also been studied. ANN Model: The development of ANN model has been done by using fractional formation of FAME (Xc ANN model has been validated with two additional sets of experimental data, obtained by conducting separate experiments, and the ANN model predictions are within ± 4 % deviation. ), versus time (t) experimental data. The developed model is a Feed forward Neural Network (FFNN). There are four input neurons corresponding to four input variables namely methanol to oil molar ratio, catalyst amount, temperature, and time, and one output neuron corresponding to the fractional formation of FAME. There is one hidden layer in optimized ANN model consisting of 12 neurons. In this model tangent sigmoid function has been used as the activation function in the hidden layer and linear function has been used as the activation function in the output layer. vi A kinetic model has been developed by using Xc various time t data namely, 9:1, 12:1, 15:1, 18:1, and the experimental data at 25:1. The model is applicable at the optimum temperature = 60 °C and catalyst amount = 3 % v/v. The kinetic model provides the effect of change in the methanol to oil molar ratio on Xc at optimum operating conditions. The model is reproduced below. With at . Where t = time (min) , k1 = 0.000295 (l/gmol.min), k2 = 0.026775 (l/gmol.min) The developed kinetic model predicts the fractional formation of FAME within ± 10 % deviation It is our view that the results of this study may be useful for designing a batch or continuous flow reactor for castor oil transesterification at optimum operating conditions.