Process Optimization of Zinc Chloride Activated Carbon Production from Avocado Pear Seed Waste: An Assessment of Artificial Neural Network and Box-Behnken Design

Abstract

This study deals with the production of avocado pear seed activated carbons using acidic (sulphuric acid) reagent according to Response Surface Methodology (RSM) and Machine learning (ML) approaches. During the investigation, Artificial neural network (ANN) and Box Behnken Design (BBD) were used to assess the influence of the activation temperature (600 -900⁰C), activation time (60-120 mins), and impregnation ratio (0.5-1.5) on the achievable BET surface area. The optimization of sulphuric acid-activated avocado pear seed production was also comparatively examined using both BBD with the RSM approach and ANN neural network model to determine the optimum process conditions. The Analysis of Variance (ANOVA) unveiled that the significant factors were activation temperature, and impregnation ratio for the avocado pear seed acid activation process, as all their p-values were less than 0.1. The best process conditions discovered for producing optimal BET surface area of H2SO4 activated carbon were activation temperature (1045.73K), activation time (120mins), and impregnation ratio (1.21) respectively. The optimal BET surface area achieved for H₂SO₄-activated avocado pear seed (APS) was 517.8m².g^-1. The correlation coefficient (R) for the RSM and ANN BET models were found to be 0.88, and 0.9955 respectively. Based on these results, the ANN BET model was ascertained to be the most capable model for predicting and forecasting the achievable surface area of H₂SO₄-activated avocado pear seed (APS). The RSM and ANN neural network can be applied as effective analytical tools for optimizing the HAPS production process.