Identification of Granule Growth Regimes in High Shear Wet Granulation Processes Using a Physics-Constrained Neural Network

The digitization of manufacturing processes has led to an increase in the availability process data, which enabled use data-driven models predict outcomes these processes. Data-driven are instantaneous simulate and can provide real-time predictions but lack any governing physics within their framework. When data deviates from original conditions, may not agree with physical boundaries. In such cases, first-principle-based have proven be effective computationally inefficient cannot solved real time. Thus, there remains a need develop efficient understanding about process. this work, we demonstrate addition physics-based boundary conditions constraints neural network improve its predictability for granule density size distribution (GSD) high shear granulation physics-constrained (PCNN) was better at predicting growth regimes when compared other networks no constraints. input that violated boundaries provided, PCNN identified points more accurately non-physics constrained networks, error <1%. A sensitivity analysis variables also performed understand individual effects on final outputs.