The management of mixed solid waste presents a significant environmental challenge, with conventional disposal methods contributing to pollution and resource depletion. Catalytic co-pyrolysis offers a promising thermochemical pathway for converting heterogeneous waste streams into valuable hydrocarbon products, yet optimising feedstock composition and process conditions remains complex due to the multivariable nature of these systems.
This study presents an integrated experimental and modelling framework to develop optimised catalytic co-pyrolysis systems for mixed solid waste conversion. Our approach combines physics-based modelling with machine learning techniques to predict optimal feedstock compositions for hydrocarbon chain growth. Building upon established kinetic models developed earlier, we constructed a validated physics-based model that captures fundamental reaction mechanisms during co-pyrolysis.
Recognising the limitations of purely mechanistic approaches, we employed the physics-based model to generate synthetic datasets spanning a wide range of operating conditions. These data were subsequently used to train an artificial neural network capable of capturing the complex, nonlinear relationships between hydrogenation temperatures, reaction time, initial feedstock composition, and product distribution. This hybrid modelling strategy leverages the interpretability of physics-based models while harnessing the predictive power of data-driven approaches.
Preliminary results demonstrate that the integrated framework can effectively identify feedstock combinations and process parameters that maximise desired hydrocarbon yields. The model provides valuable insights into synergistic interactions between different waste components during catalytic conversion. This work contributes to the development of more efficient and flexible waste-to-energy systems, supporting circular economy objectives by transforming mixed solid waste into useful chemical feedstocks and fuels.