Alkaline Hydrolysis Depolymerisation of Waste Polyethylene Terephthalate (PET): Optimisation, ANN, ANFIS, and Kinetic Study

Abstract Alkaline depolymerisation of waste polyethylene terephthalate (PET) is an effective hydrolysis method for producing high-quality terephthalic acid (TPA), which has various applications, including the synthesis of metal–organic frameworks (MOFs) for carbon capture. Despite existing research on alkaline PET depolymerisation, optimisation of reaction kinetics and the integration of machine learning for predictive modelling represent significant gaps. This study explores the optimisation and kinetics of PET chemical recycling via alkaline hydrolysis using sodium hydroxide (NaOH). Central composite design (CCD) experiments were conducted, yielding 21 trials with variables including temperature (160 to 240 °C), time (2 to 8 h), solid-to-liquid ratio (2 to 10 wt.%), and NaOH concentration (0.5 to 4 M). To enhance the dataset for machine learning, it was expanded from 21 experimental datasets to 100 through random interpolation. Both Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS) were applied for yield prediction, utilising MATLAB R2025a. Experiments in a laboratory-scale Teflon autoclave were conducted, and the optimal conditions were found to be: 203 °C, 5.225 h, a solid-to-liquid ratio of 5.800 wt.%, and NaOH concentration of 2.225 M, yielding 98.84% TPA. The reaction kinetics yielded an apparent activation energy of 85.58 kJ/mol. ANFIS demonstrated better predictive performance with an R 2 of 0.97 (RMSE = 4.502), compared to ANN’s 0.952 (RMSE = 2.79). This work innovatively combines response surface methodology, kinetic modelling, and machine-learning predictions in PET depolymerisation, contributing to sustainable chemical recycling and addressing the United Nations Sustainable Development Goals (SDG 9, SDG 12, and SDG 13), thereby advancing circular economy.