Alkaline fusion approach to Zeolite Y synthesis from Ahoko Metakaolin: computer-aided simulation, response surface and ANFIS modelling with uncertainty quantification

Abstract

This study aimed to simulate zeolite Y synthesis through the alkaline fusion method, optimization of process variables, and Adaptive Neuro-Fuzzy Inference System (ANFIS) modeling of utility and waste generated during the synthesis of zeolite Y. The steady-state simulation was performed using ASPEN Batch Process Developer (ABPD), and the regression model and ANFIS input data were achieved using Box-Behnken Design (BBD). ANFIS models were developed for utility, the ratio of Waste Water Generated (WWG), and, the ratio of Sodium Metasilicate (SM) in the waste stream. Monte Carlo Simulation (MCS) was used to study the sensitivity of variables and predict the uncertainty of the developed regression models. The ABPD produced 0.00104 kg of zeolite Y per batch at an estimated batch time of 73.17 h and a flowrate of 0.0000193 kg/h. The optimum variables that minimized the utility (8.17 kJ), and the ratios of WWG (0.398) and SM (0.601) are 1 g of NaOH, 6.43 g of Na2SiO3, and 207 ml of water respectively. Comparison of RSM and ANFIS models for the prediction of the utility, and the ratios of WWG and SM show that both model types can predict the responses with R2 values ranging between 0.998 and 1. The uncertainty analysis values obtained from the MCS for the three responses showed that the obtained uncertainty values are negligible to developed models. Therefore, proof of concept validation of alkaline fusion synthesis of zeolite Y was modeled and, information on energy required for synthesis and volume of WWG from the synthesis of zeolite Y was presented.