Developing a Predictive Control System for Optimal Chiller Operation

In a world where decarbonisation is an urgent priority, the building sector plays a pivotal role in mitigating carbon emissions. To meet this challenge, it is necessary to adapt HVAC systems and develop advanced controllers, of which Model Predictive Control (MPC) stands out. This work focuses on the development of an MPC designed to optimise the operation of a refrigeration plant. To achieve an accurate controller, detailed models of the main components of the system, such as the chiller and the thermal energy storage tank, have been integrated into a simplified, control-oriented plant model. Various options have been analysed to eliminate the need for binary variables and conditional structures in the model. The developed MPC belongs to the field of nonlinear programming and aims to minimise consumption or cost and to guarantee demand coverage. The controller has been tested on a digital twin of the plant, and calculations have been performed to determine the energy and cost savings achieved.