Abstract:Fine chemicals produced via batch crystallization with properties dependent on the crystal size distribution require precise control of supersaturation, which drives the evolution of crystal size over time. Model predictive control (MPC) of supersaturation using a mechanistic model to represent the behavior of a crystallization process requires less experimental time and resources compared with fully empirical model-based control methods. Experimental characterization of the hexamine-ethanol crystallization system was performed in order to collect the parameters required to build a one-dimensional (1D) population balance model (PBM) in gPROMS FormulatedProducts software (Siemens-PSE Ltd.). Analysis of the metastable zone width (MSZW) and a series of seeded batch cooling crystallizations informed the suitable process conditions selected for supersaturation control experiments. The gPROMS model was integrated with the control software PharmaMV (Perceptive Engineering Ltd.) to create a digital twin of the crystallizer. Simulated batch crystallizations were used to train two statistical MPC blocks, allowing for in silico supersaturation control simulations to develop an effective control strategy. In the supersaturation set-point range of 0.012-0.036, the digital twin displayed excellent performance that would require minimal controller tuning to steady out any instabilities. The MPC strategy was implemented on a physical 500 mL crystallizer, with the simulated solution concentration replaced by in situ measurements from calibrated attenuated total reflection-Fourier transform infrared (ATR-FTIR) spectroscopy. Physical supersaturation control performance was slightly more unstable than the in silico tests, which is consistent with expected disturbances to the heat transfer, which were not specifically modeled in simulations. Overall, the level of supersaturation control in a real crystallizer was found to be accurate and precise enough to consider future adaptations to the MPC strategy for more advanced control objectives, such as the crystal size.

Physics-Informed Machine Learning for MPC: Application to a Batch Crystallization Process

Physics-informed Neural Networks to Solve Lumped Kinetic Model for Chromatography Process.

Developing Physics-Informed Neural Networks for Model Predictive Control of Periodic Counter-Current Chromatography

Spatial Batch Optimal Design Based on Self-Learning Gaussian Process Models for LPCVD Processes

Latent-Variable Nonlinear Model Predictive Control Strategy For A Ph Neutralization Process

Adaptive nonlinear model predictive control for a class of multivariable chemical processes

Iterative Learning Model Predictive Control for a Class of Continuous/Batch Processes

Physics-informed recurrent neural network modeling for predictive control of nonlinear processes

Long Short-Term Memory Neural Networks for Modeling Dynamical Processes and Predictive Control: A Hybrid Physics-Informed Approach

A method for the rapid creation of AI driven crystallization process controllers

Recurrent Neural Network-Based Model Predictive Control for Continuous Pharmaceutical Manufacturing

Online Learning‐Based Predictive Control of Crystallization Processes under Batch‐to‐Batch Parametric Drift

Highly Efficient Crystallization Studies through Machine Learning and Automation

Explicit machine learning-based model predictive control of nonlinear processes via multi-parametric programming

Adaptive Predictive Control Algorithm for Batch Processes: Application to a Rotational Molding Process

Iterative Learning Model Predictive Control for Multi-Phase Batch Processes

Combined Iterative Learning and Model Predictive Control Scheme for Nonlinear Systems

Development of a Digital Twin for the Prediction and Control of Supersaturation during Batch Cooling Crystallization

Model Predictive Control of Melt Pool Size for the Laser Powder Bed Fusion Process Under Process Uncertainty

Iterative Learning Model Predictive Control Based on Iterative Data-Driven Modeling

Model predictive control of nonlinear processes using transfer learning-based recurrent neural networks