Optimum production of biodiesel by ultrasonication from novel candidature Kigelia africana seeds: A comparison of RSM and ANN expert systems

Abstract

ABSTRACT Biodiesel production by ultrasonication was carried out from non- edible Kigelia africana oil using a novel heterogeneous base catalyst by the calcination of Pyrgostylus striatulus shells. A comparative study of expert optimization systems namely Response surface methodology (RSM) and Artificial neural intelligence (ANN) was executed on optimization of the process parameters for the transesterification process. The process parameters considered were alcohol to oil ratio (9:1 to 15:1 mol/mol), catalyst weight (3–5 wt.%) and ultrasonication time (30–60 min) on methanolysis process. Statistical error analysis was performed. Quadratic regression model with R2 value of 0.9854 and 0.9788 was observed for ANN and RSM, respectively. The results of the current work indicated ANN model was a cut above RSM. The physio-chemical properties of biodiesel produced were assessed and found to be within the standard ASTM limits indicating the bio-fuel to be suitable.