Real-Time Optimisation of a Data Center Free-Cooling System via Dither-Free Least-Squares Extremum Seeking

This paper addresses the supervisory real-time optimisation (RTO) of a data center free-cooling (FC) system, with the aim of minimising auxiliary power consumption while meeting thermal requirements. The problem is handled in a measurement-based way by adapting the pump and fan speed references online through a dither-free Least-Squares Extremum Seeking Control (LS-ESC) strategy based on first-order least-squares gradient estimation. The approach is assessed in a Simscape-based simulation environment of a coupled thermofluid FC system under a selected representative operating condition. The results show that the proposed method can recover the desired inlet temperature at the data center side while reducing the auxiliary power demand and coordinating the control effort between pump and fan in a non-trivial way. Overall, the study suggests that dither-free LS-ESC can serve as a practical supervisory RTO tool for FC applications and, more generally, for coupled thermo-fluid energy systems.