Multiobjective nonlinear model predictive control of pharmaceutical batch crystallizers.

Area of Science:

• Chemical Engineering
• Process Control
• Crystallization Technology

Background:

• Batch crystallization is crucial for pharmaceutical and chemical industries.
• Existing model predictive control (MPC) methods often require complex tuning with weighting functions.
• Developing advanced control strategies for batch crystallization is essential for product quality and process efficiency.

Purpose of the Study:

• To implement a rigorous multiobjective nonlinear model predictive control (MPC) procedure for batch crystallization.
• To demonstrate the effectiveness of MPC without using weighting functions or additional restrictive constraints.
• To apply the developed MPC technique to diverse crystallization scenarios involving different substances.

Main Methods:

• A multiobjective nonlinear model predictive control (MPC) approach was developed and applied.
• The optimization language Pyomo, integrated with the GAMS interface, was utilized for problem-solving.
• Three distinct batch crystallization cases were modeled and simulated: unseeded paracetamol, seeded potassium nitrate, and temperature-controlled citric acid anhydrate.

Main Results:

• The implemented MPC procedure successfully controlled batch crystallization processes.
• The absence of weighting functions and restrictive constraints simplified the control strategy.
• Effective control was demonstrated across unseeded, seeded, and temperature-controlled crystallization systems.

Conclusions:

• Multiobjective nonlinear MPC offers a robust and effective control solution for batch crystallization.
• This approach provides a simplified yet powerful alternative to traditional MPC methods.
• The successful application to paracetamol, potassium nitrate, and citric acid anhydrate highlights the versatility of the technique.