abstract
- AbstractThe modeling of emulsion polymerization (EP) commonly involves the coupling of kinetic and thermodynamic models. High computational resources are commonly used to compute the monomer partition within the aqueous phase, monomer droplets and polymer particles using complex or simplified approaches such as the simple partition model, the Morton-Vanzo equations and semi-empirical expressions. It would be desirable to develop a simpler or surrogate model, based on thermodynamic principles, which can be readily incorporated to existing or new EP models avoiding the implementation of additional time-consuming algorithms and convergence issues. Herein, a surrogate model using artificial neural networks (an artificial intelligence (AI) tool) trained through the solution of the Morton-Vanzo equations is developed. This model is coupled to EP models to analyze data for the ab initio homopolymerization of styrene (STY) and methyl methacrylate (MMA) with potassium persulfate (KPS) as initiator and sodium dodecyl sulfate (SDS) as surfactant at 60 °C. The proposed model offers a simpler implementation of the monomer partition calculation and is readily coupled to population balance equations such as the pseudo-bulk and 0¿1-2 models. The AI-based approach matches well the outputs of the thermodynamic model with similar simulation times.