Application of MPC to a MIMO process

Abstract

Maintaining precise levels of liquid in storage tanks is crucial across various industries, including chemical manufacturing, water treatment, food and beverage production, and climate control for buildings. This dissertation explores the application of the Model Predictive Control approach to a Coupled Tank System provided by Kentridge Instruments Pte Ltd. This system was chosen for its ability to present a complex control problem due to the interaction between the tanks and the constraints on water levels and flow rates, making it an ideal candidate for Model Predictive Control application. The system modeling is based on the First Principles approach, using the mass balance equation to describe water volume in each tank. This method derives dynamic differential equations and accounts for water flow, identifying essential system parameters for an approximate state-space model of the Coupled Tank System. Model Predictive Control (MPC) predicts system behavior to compute optimal control actions over a time horizon. MATLAB and Simulink simulated the system, focusing on flow rate, water level, and other constraints parameters. The results showed MPC enhances control in multiple input/output systems, improving operational efficiency and performance reliability in liquid management applications.