Digital Design of an Agrochemical Crystallization Process via Two-Dimensional Population Balance Modeling

Solution crystallization is commonly used in the agrochemical industry to isolate and purify the active ingredient (AI). The generated particles' characteristics, typically needle-like shape in the agrochemical industry, can affect the performance of subsequent unit operations (filtration, washing, and drying). Previous experimental findings highlighted the impact of poor crystallization control on subsequent unit operations. By following the model-free Quality-by-Control (QbC) framework, the improved crystallization process highlighted the effectiveness of temperature cycling to improve particle properties and filtration time. However, the model-free QbC approach generally leads to suboptimal process performance. To further optimize the process while reducing experimental time, cost, and efforts, in this paper, a model-based QbC framework was applied to a commercial agrochemical crystallization process. A digital model of the system was developed using a two-dimensional population balance model (2D-PBM), which was calibrated with carefully designed parameter estimation experiments to estimate the 2D kinetics. Sensitivity analysis via different methods highlighted the estimability of certain parameters, assessed the quality of the estimated parameters, and tested the reliability of the model. Lastly, the digital model was applied to an optimization framework to perform in silico experiments and digital design of an optimal operating procedure. By validating the optimization results experimentally, the nominal temperature profile improved the content uniformity of the crystal product by reducing the size distribution span and minimizing the generation of fine particles.