Modeling and optimization of Thevetia peruviana (yellow oleander) oil biodiesel synthesis via Musa paradisiacal (plantain) peels as heterogeneous base catalyst: A case of artificial neural network vs. response surface methodology

•Thevetia peruviana oil was converted to biodiesel by 2-step transesterification.•A model was developed for the esterification step and process was optimized.•A novel solid catalyst from plantain peel was developed and characterized.•ANN and RSM models were developed and optimized for the transesterification step.•Properties of the biodiesel produced were within standard specifications.

In this work, response surface methodology (RSM) was used to optimize the pretreatment step (esterification) while the transesterification step was optimized using both RSM and artificial neural network (ANN). The acid value of yellow oleander oil (YOO) with high FFA was reduced to 1.72 mgKOH g−1 by esterification with a statistically significant quadratic model at the optimal condition of methanol–oil ratio 0.35 (v/v), H2SO4 0.78% (v/v), reaction time 60 min and reaction temperature 55 °C. RSM predicted optimal condition for the transesterification was methanol–oil ratio 0.3 (v/v), reaction time 1.5 h and calcinated plantain peels (CPP) amount 3.0% (w/v) with yellow oleander oil biodiesel (YOOB) yield of 95.25% (w/w), which was validated as 94.87% (w/w) while ANN predicted optimal condition was methanol–oil ratio 0.3 (v/v), reaction time 1.25 h and CPP amount 2.8% (w/v) with YOOB yield of 94.97% (w/w), which was validated as 95.09% (w/w). The results obtained showed that ANN was a better and more efficient optimization tool than RSM due to its higher value of R2 and lower value of AAD. The quality of the YOOB obtained was within the ASTM D6751 and DIN EN 14214 biodiesel specifications.