(372m) Modeling and Control of Batch Processes with Recycle

Batch processes are prevalent across various industries such as chemical, mechanical, biochemical, agriculture, and pharmaceutical, especially for producing high-value products. Given the limited production volume in batch processes, ensuring consistent high-quality products is crucial. Model Predictive Control (MPC) is a widely adopted control technology in these industries due to its capability to forecast the process's future behavior and generate an optimal input sequence while adhering to specified constraints [1,2,3]. Rotational molding, also known as roto-molding, is a prime example of a batch process extensively used in industries for manufacturing hollow plastic products. The profitability of producing such items can be greatly enhanced by implementing optimal control strategies to minimize waste costs. Achieving this requires a comprehensive focus on modeling, particularly for processes like rotational molding where the quality variables that need to be controlled are not directly measurable or available during the process. Additionally, in these industries, there are instances where products fail to meet specified standards, often due to disturbances during a batch or impurities in raw materials. In such cases, it becomes desirable to disassemble the product if feasible and reuse its components along with the raw material to attempt the production of a higher-quality product.

The present manuscript addresses the problem of economically achieving a user-specified set of product qualities in an industrial complex batch process and additionally accounting for the recycled material, illustrated through a lab-scale uni-axial rotational molding (also known as roto-molding) setup. To this end, a data-driven Economic MPC (EMPC) formulation is developed and implemented to achieve product specification via constraints on the predicted quality variables. First, a state-space dynamic model of the roto-molding process is built using previous batch data generated in the lab using an uni-axial roto-molding setup. The dynamic model captures the internal mold temperature trajectory for an input sequence (combination of two heaters and compressed air). This model is then supplemented by a partial-least-squares quality model, which relates key quality variables (sinkhole area and impact energy) with the terminal (states) prediction. The complete model is then placed within the EMPC scheme which minimizes the cost associated with inputs and allows the user to specify required product quality via constraints on the quality variables. Results achieved from experimental studies illustrate the capability of the proposed EMPC scheme in lowering the process cost (energy requirements) for two manifestations of the economic objective.

References:

1. Bonvin, D. Optimal operation of batch reactors—a personal view. Journal of Process Control 1998, 8, 355–368.

2. Flores-Cerrillo, J.; MacGregor, J. F. Control of particle size distributions in emulsion semibatch polymerization using mid-course correction policies. Industrial and Engineering Chemistry Research 2002, 41, 1805–1814

3. Mhaskar, P.; Garg, A.; Corbett, B. Modeling and Control of Batch Processes; Advances in Industrial Control; Springer International Publishing: Cham, 2019.