Modelling and optimizing the transesterification process of shea butter via CD-BaCl-IL catalyst using soft computing algorithms

Abstract

Shortage and environmental threats posed by fossil fuel have become a critical issue requiring searching for alternative energy sources. This study utilized clay-doped barium chloride and ionic liquid (CD-BaCl-IL) as a catalyst for optimizing biodiesel production from shea butter. 10% barium chloride and ionic liquid were blended with clay, dried, and then calcined for 4 h at 600 °C to develop the catalyst. The synthesized catalyst and the biodiesel were suitably characterized. Response surface methodology (RSM) implementing Central Composite Design (CCD) and Genetic Algorithm (GA) were employed to model and optimize the effect of the process parameters on the response. The model's capabilities were evaluated using coefficient of determination (R2) and mean square error (MSE). The second-order polynomial model is shown in the Analysis of Variance (ANOVA) with an (R2 -0.9945, Adjusted R2-0.9846, Predicted R2-0.8694) demonstrating the model's acceptance. Artificial Neural Network (ANN) and Adaptive Neurofuzzy Inference System (ANFIS) were used to assess the model's capability. The obtained statistical results - (R2 = 0.8694, MSE = 0.7035), ANN (R2 = 0.999, MSE = 0.0001026), ANFIS (R2 = 0.99, MSE = 0.000041), showed that ANFIS had the best prediction with the lowest MSE. ANFIS integrated with GA (GA-ANFIS) gave the best optimization with a biodiesel yield (96.72%) at a catalyst concentration of 4 wt%, a methanol/mol ratio of 10:1, a time of 2.5 h, a temperature of 70 °C, and an agitation speed of 400 rpm. The optimal developed biodiesel properties were successfully evaluated within the ASTM D 6751 standards.

Keywords

• Genetic algorithm
• Response surface methodology
• Shea butter biodiesel
• Transesterification
• Adaptive neuro-fuzzy inference system
• CD-BaCl-IL