Chillers are one of the most sophisticated and most important equipment in cooling plants. Due to its complexity and importance, a cost effective and accurate control technique is a necessity to ensure system reliability and longevity. As the chiller contains multiple inputs and outputs, two advanced multivariable control techniques were selected to control the chillers cooling capacity, exit chilled liquid temperature of the evaporator as well as to reject disturbances. The advanced techniques covered in this framework are the Linear Quadratic Integral (LQI) and Model Predictive Control (MPC). Both were successfully applied to the chiller’s model for multiple testing cases and simulations. Then the results were compared to the industry standard control technique, the PID. Successful use of Genetic Algorithm (a machine learning method) is also presented in this thesis as a method for tunning the controller weights. It was deduced from the simulation test results that the LQI and MPC outperformed the traditional PID controllers in terms of energy efficiency and transient response. An energy savings of around 10% to 20% is seen on the compressor’s electric power, lower overshoot/undershoot for most outputs and faster settling time. Also, the MPC controller had the ability to incorporate input constraints into the problem formulation and use quadratic programming to find a solution to the constraint optimization problem.
Library of Congress Subject Headings
Refrigeration and refrigerating machinery--Automatic control; Compressors--Automatic control; Genetic algorithms
Electrical Engineering (MS)
Awad, Suliman, "Advanced Control of Vapor Compression Liquid Chillers" (2023). Thesis. Rochester Institute of Technology. Accessed from