A modified time increment Monte Carlo based optimization framework for PSD dynamics in poly(methyl) methacrylate suspension polymerization

Stochastic evolution of the particle size distribution (PSD) in poly(methyl methacrylate) suspension polymerization is influenced by the transient equilibrium between turbulent breakage and coalescence. A high-fidelity Constant Number Event Driven Monte Carlo framework resolves these dynamics by explicitly coupling conversion-dependent thermophysical properties specifically viscosity amplification and dynamic interfacial tension to hydrodynamics inspired kernels. Bayesian optimization via a Tree structured Parzen Estimator minimizes a composite loss function of Mean Squared Error and Earth Mover’s Distance ensuring high fidelity topological fits across the 10–150 µm domain. Experimental data at 700–900 rpm and 0.1–0.3 wt% PVA reveal a Sauter mean diameter (d 32 ) range of 30.9–67.0 µm with peak refinement (30.9 ± 15.1 µm) achieved at 900 rpm and 0.2 wt% PVA. The framework reproduces mass weighted PSDs with Root Mean Square Errors (RMSE) consistently below 0.073 and a minimum of 0.036. Kinetic parameters optimized at a single intermediate speed (800 rpm) successfully predict PSD evolution at uncalibrated hydrodynamic boundaries (700 and 900 rpm) with less than 4% RMSE deviation. Sensitivity analysis of agitation speed perturbations (±5% to ±20%) demonstrates that while RMSE increases from a baseline of 0.052 to a plateau of 0.078, the morphological shape of the PSD remains invariant, confirming the structural robustness of the mechanistic framework. These results provide a statistically validated platform for the rational design and hydrodynamic scale-up of poly(methyl methacrylate) suspension polymerization.