Steady State and Transient Experimentally Validated Analysis of Hybrid Data Centers

2015 
Data centers consume a considerable amount of energy which is estimated to be about 2% of the total electrical energy consumed in the U.S. in the year 2010, and this number continues to increase every year. Thermal management is becoming increasingly important in the effort to improve the energy efficiency and reliability of data centers. The goal is to keep the information technologies (IT) equipment temperature within the allowable range in high power density data centers while reducing the energy used for cooling. In this regard, liquid and hybrid air/water cooling systems are alternatives to traditional air cooling. In particular, these options offer advantages for localized cooling higher power racks which may not be manageable using the room level air cooling system without requiring significantly more energy. In this paper, a hybrid cooling system in data centers is investigated. In addition to traditional raised floor, cold aisle-hot aisle configuration, a liquid–air heat exchanger attached to the back of racks is considered. First of all, the paper presents a review of literature of the study of this heat exchanger strategy in the thermal management of a data center. The discussion focus on rear door heat exchanger (RDHx) performance, both the steady state and transient impact are analyzed. The studies show that under some circumstances, this hybrid approach could be a viable alternative to meet the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) recommended inlet air temperatures, while at the same time reducing the overall energy consumption in high density data centers. The hybrid design approach can also significantly improve the dynamic performance during rack power increases or computer room air conditioner (CRAC) unit failure. And then, additional parametric steady state and dynamic analyses, are presented in detail for the different scenarios.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    30
    References
    8
    Citations
    NaN
    KQI
    []