Data-Driven Modeling of a Pilot Plant Batch Reactor and Validation of a Nonlinear Model Predictive Controller for...

Area of Science:

• Chemical Engineering
• Process Control
• Nonlinear Systems

Background:

• Batch processing is vital in chemical industries, especially for high-quality, low-volume production.
• Batch reactors exhibit inherent nonlinearities that challenge conventional linear control strategies.
• Linear controllers often lead to aggressive manipulated variable actions and increased energy usage.

Purpose of the Study:

• To investigate and address the nonlinearities in a pilot plant batch reactor.
• To develop and implement a nonlinear model predictive controller (NMPC) for improved batch reactor performance.
• To compare the effectiveness of NMPC against conventional linear controllers.

Main Methods:

• Identification of nonlinear system dynamics using Autoregressive with Exogenous Inputs (ARX) and Nonlinear Autoregressive with Exogenous Inputs (NARX) models.
• Utilizing open-loop data from a pilot plant batch reactor for model identification.
• Design and implementation of a nonlinear model predictive controller (NMPC).

Main Results:

• The nonlinear model accurately captured the batch reactor's dynamics.
• The NMPC resulted in significantly smoother manipulated variable action compared to linear controllers.
• Reactor temperature response was stabilized, leading to more consistent operation.
• Both simulation and real-time experiments confirmed the effectiveness of the NMPC approach.

Conclusions:

• Nonlinear model predictive control is an effective strategy for managing nonlinearities in batch reactors.
• NMPC implementation leads to improved operational efficiency, including reduced energy consumption and smoother control.
• The developed nonlinear model provides a robust basis for advanced control of batch processes.