RSM/ANN based modeling of methyl esters yield from Anacardium occidentale kernel oil by transesterification, for possible application as transformer fluid

Abstract

This study is on the modeling of methyl esters production process; obtained by the transesterification of Anacardium occidentale kernel (AOK) oil (AOKO), using artificial neural network (ANN) and response surface methodology (RSM). AOKO was obtained from the kernels/seeds of Anacardium occidentale tree. The oils were extracted from the kernels using solvent extraction method. The physicochemical properties of AOKO and Anacardium occidentale kernel oil methyl esters (MAOKOt) were determined using standard methods. Fatty acids composition was determined using gas chromatography (GC). At modeling conditions of temperature (65 ​°C), mole ratio (7:1), catalyst concentration (2.5 ​wt %), stirring speed (600 ​rpm) and time (150 ​min), the RSM predicted and validated methyl ester yields were 94.82%, and 94.70%, respectively; while ANN predicted and validated yields were 93.21% and 93.33%, respectively. The physicochemical characterization results of AOKO and MAOKOt samples, show that their respective viscosity, dielectric strength (DS), pour and flash points were (20.01 and 10.97 mm2s-1), (25.34 and 38.60 ​kV), (11 and 5 ​°C), and (270 and 288 ​°C). These results indicated the MAOKOt sample’s potential use as transformer fluid. The GC result indicated that MAOKOt was unsaturated. Finally, on the basis of the gotten model results, ANN was adjudged as a better predictive model, when compared to RSM.