Artificial Neural Network and Response Surface Methodology for Predicting and Maximizing Biodiesel Production from Waste Oil with KI/CaO/Al2O3 Catalyst in a Fixed Bed Reactor

Abstract

AbstractBiodiesel from waste oil is produced using heterogeneous catalyzed transesterification in a fixed bed reactor (FBR). Potassium iodide/calcium oxide/alumina (KI/CaO/Al2O3) catalyst was prepared through the processes of calcination and impregnation. The novel catalyst was analyzed with X‐ray diffraction (XRD), scanning electron microscopy (SEM), and energy dispersive X‐ray spectrometer (EDX). The design of experiment (DoE) method resulted in a total of 20 experimental runs. The significance of 3 reaction parameters, namely catalyst bed height, methanol to waste oil molar ratio, and residence time, and their combined impact on biodiesel yield is investigated. Both the artificial neural network (ANN) based on artificial intelligence (AI) and the Box‐Behnken design (BBD) based on response surface methodology (RSM) were utilized in order to optimize the process conditions and maximize the biodiesel production. A quadratic regression model was developed to predict biodiesel yield, with a correlation coefficient (R) value of 0.9994 for ANN model and a coefficient of determination (R2) value of 0.9986 for BBD model. The maximum amount of biodiesel that can be produced is 98.88 % when catalyst bed height is 7.87 cm, molar ratio of methanol to waste oil is 17.47 : 1, and residence time is 3.12 h. The results of this study indicate that ANN and BBD models can effectively be used to optimize and synthesize the highest %yield of biodiesel in a FBR.