Identifying and controlling temperatures by airflow in data center's room: a data driven methodology for the management of CRAC units

This thesis focuses on the problem of energy consumption of the data center's cooling system. Three steps have been done to improve the energy percformance: study of the system with flow mixing detection, linear models design of the data center room temperature, implementation of two different controllers. The first, a Model Predictive Control designed on the models already developed. The second one a model free PID controller which has been used to evaluate the MPC performance.