Optimization and Artificial Neural Network (ANN) modeling of oil expression from enzyme treated Jatropha curcas L. (Ratanjot) on a hydraulic press
Loading...
Date
2007-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand)
Abstract
Rapid urbanization, agricultural mechanization and increase in vehicular population enhance demand for fuel. So meeting the energy requirements in a sustainable manner is a major challenge. Among the many tree species which can yield oil as a source of energy in the form of bio-diesel, Jatropha curcas L. (Ratanjot) has been found most suitable due to its various favorable attributes like hardy nature, short gestation period, high oil recovery and quality of oil
Study was conducted to optimize and to develop Artificial Neural Network (ANN) models for oil expression from Jatropha curcas L. (Ratanjot) on a hydraulic press. Experiments were planned using full factorial design in two phases. In the first phase of experimental design, three levels of husk percentage, five levels of pressure and five levels of holding time were taken as independent variables. In a similar way five levels of enzyme concentration, five levels of pressure and five levels of holding time were taken as independent parameters in the second phase of experiments. Line curves, surface plots and iso-oleum curves were developed to show the effect of independent parameters on oil expression. Empirical mathematical models representing oil expression in terms of single and multiple responses of process parameters were developed using SPSS software. Optimization of variables was performed by partial differentiation of multiple regression equation with respect to each variable and then solving the coefficient matrix on MATLAB software. In order to have a better prediction of unseen input conditions within the experimental range Artificial Neural Network (ANN) modeling of oil expression process was carried out using back propagation algorithm and MATLAB software.
Enzymatic treatment substantially enhances oil expression from 87% with hydraulic pressing alone to 91% for hydraulic pressing with enzymatic treatment. The optimum conditions of husk percentage, pressure and holding time for maximum oil recovery were obtained as 87.40%, 45.63 MPa and 27.09 min respectively. Optimum conditions of enzyme concentration, pressure and holding time for maximum oil recovery were obtained as 110.73mg/100 g dry matter, 43.83 MPa and 17.42 min respectively. Optimum architecture of ANN for training at different husk percentages was found to be two hidden layers with 8 and 11 nodes in first and second hidden layer while that for samples at different enzyme concentrations was found to be two hidden layers with 9 and 11 nodes in first and second hidden layer respectively. Both in case of training and testing results of output predicted by ANN architecture shows good agreement with experimental values.
Description
Keywords
null