Development of Predictive Control System Using Just-In-Time Modeling and Enthalpy-Aware Control in Air Conditioners for Large-Scale Data Center

In this research, we propose an energy-efficient predictive control system with outside-air-introduced-type air conditioner in large-scale data center. The system consists of two parts; a predictive part and a control part. We propose to use a dynamic model construction technique called just-in-time (JIT) modeling in the predictive part. It can estimate the air temperature and humidity in the DC room and temperature at the inlet of servers at a high accuracy while the update of data center model is constructed naturally. We also propose to use a control algorithm that utilizes air enthalpy at outside and DC room in addition to the air temperature and humidity in the control part. Owing to the enthalpy, we achieve an effective energy-saving with both wet and dry air. We introduced this predictive control system to a large-scale data center and had verified the improvement in the energy-saving performance for few month. The air temperature and humidity after one hour could be successfully estimated with an accuracy of correlation coefficient of 0.97 or more using the proposed JIT modeling. Furthermore, we demonstrated that the power consumption of the DC room was significantly reduced by 28.9% using the proposed control algorithm compared to that of the conventional control algorithm.