Tuning the Model Predictive Control of a Crude Distillation Unit.

Area of Science:

• Chemical Engineering
• Process Control
• Optimization

Background:

• Industrial Crude Distillation Units (CDUs) require precise control for optimal performance.
• Model Predictive Control (MPC) is a key technology for managing complex industrial processes.
• Tuning MPC parameters is crucial for effective process management, especially with real-time optimization targets.

Purpose of the Study:

• To develop and evaluate a novel multi-objective optimization approach for tuning MPC parameters in a CDU.
• To compare the proposed tuning method against an existing multi-objective optimization technique.
• To assess the computational load and tuning effort associated with the new approach.

Main Methods:

• A realistic CDU model and MPC model were used, assuming a nominal case where both models are identical.
• Process outputs were controlled within zones rather than at fixed set points.
• A multi-objective optimization strategy was employed to determine weights for output error, input deviation from targets, and input moves.
• Simulations were conducted to compare the proposed method with an existing approach.

Main Results:

• The proposed multi-objective optimization approach achieved performance comparable to an existing method.
• The new tuning method demonstrated a significantly lower computational load compared to the existing approach.
• The tuning effort required by the proposed method was considerably less than conventional ad-hoc procedures.

Conclusions:

• The developed multi-objective optimization approach provides an efficient and effective way to tune MPC for CDUs.
• This method offers a practical alternative to existing techniques, reducing computational burden and tuning complexity.
• The findings suggest potential for broader application in industrial process control optimization.