SUMMARY

Data centers are facilities that house arrays of electronic racks containing high power dissipation data processing and storage equipment whose temperature must be maintained within allowable limits. This research introduces the concept of “openness” in data centers and explores required design framework for new energy efficient thermal solutions to continue the sustainable and reliable operation of the electronic equipment in data centers. Application of this framework for current air-cooled data centers and future advanced high heat load racks is shown to result in a great save in the used energy and operation cost of data centers for both short term and long term investment. While time-consuming and costly CFD/HT models are currently used to obtain the thermal field inside air-cooled data centers, this research results in an efficient and experimentally validated multi-scale method for Proper Orthogonal Decomposition (POD) based reduced order modeling of air temperature field in typical raised floor data centers, which is utilized with design methodologies to design an open and energy efficient air-cooling system. This method brings some insight to determine the appropriate number of observations required to generate the basis functions of the POD and the optimal number of modes to retain when computing the linear combination of the basis functions, two issues for which currently there is no rigorous theory. Also, regarding the thermal solution of advanced electronic racks of future data centers, a multi-scale water-cooled system is designed and built to cope with a mocked electronic rack dissipating a heat load at least two times larger than the currently available high power dissipating racks in the industry. This open water-cooled solution has potential of indefinite growth and sustainable operation for many years beyond a typical air-cooling solution and may result in a paradigm shift in the manner electronic racks are now being cooled in datacenters.