The Woodruff School

SUMMARY

Data centers are computing infrastructure facilities that house arrays of electronic racks containing high power dissipation data processing and storage equipment whose temperature must be maintained within allowable limits. The heat generated by the electronic equipment and the costs of powering the cooling systems in data centers are increasing continually. In this research, the sustainable and reliable operations of the electronic equipment in data centers are shown to be possible through the Open Engineering Systems paradigm shift in the next generation data centers. A design approach is developed to bring adaptability and robustness, two main features of open systems, in the multi-scale convective systems such as data centers. The presented approach is centered on the integration of three constructs: a) a Proper Orthogonal Decomposition (POD) based multi-scale modeling approach, b) compromise Decision Support Problem (cDSP), and c) robust design, in order to overcome the challenges in thermal-fluid modeling, having multiple objectives, and inherent variability management, respectively. Two new POD based reduced order thermal modeling methods are presented to simulate multi-parameter dependent temperature field in complex multi-scale convective systems such as data centers. The methods results in average error norm of ~ 6%, while it can be up to ~250 times faster than CFD/HT simulation in an iterative optimization technique. Also, a simpler POD based reduced order modeling approach centered is validated against experimental measurements in an operational data center facility. The open design method application for an energy efficient air cooled data center cell with an annual increase in the power consumption shows a 12-46% reduction in the energy consumption of the center in addition to being adjustable to the newer IT equipment and higher heat loads, compared with a traditional design. Also, the method can reduce the variability in the thermal response by 74%. Also, the POD based modeling and power profiling of IT equipment is used to bring adaptability for coordinated minimization of cooling and IT power consumption in future open data centers. This coordination results in 12-70% saving in the total energy consumption of a data center cell in different scenarios.

SUBJECT:	Ph.D. Dissertation Defense

BY:	Emad Samadiani

TIME:	Tuesday, July 28, 2009, 11:00 a.m.

PLACE:	MARC Building, 114

TITLE:	Energy Efficient Thermal Management of Data Centers Via Open Multi-Scale Design

COMMITTEE:	Dr. Yogendra Joshi, Co-Chair (ME) Dr. Farrokh Mistree, Co-Chair (ME) Dr. Janet K. Allen (ME) Dr. Karsten Schwan (COC) Dr. Thorsten Stoesser (CEE) Dr. Hendrik F. Hamann (IBM) Dr. Madhusudan K. Iyengar (IBM)

SUMMARY Data centers are computing infrastructure facilities that house arrays of electronic racks containing high power dissipation data processing and storage equipment whose temperature must be maintained within allowable limits. The heat generated by the electronic equipment and the costs of powering the cooling systems in data centers are increasing continually. In this research, the sustainable and reliable operations of the electronic equipment in data centers are shown to be possible through the Open Engineering Systems paradigm shift in the next generation data centers. A design approach is developed to bring adaptability and robustness, two main features of open systems, in the multi-scale convective systems such as data centers. The presented approach is centered on the integration of three constructs: a) a Proper Orthogonal Decomposition (POD) based multi-scale modeling approach, b) compromise Decision Support Problem (cDSP), and c) robust design, in order to overcome the challenges in thermal-fluid modeling, having multiple objectives, and inherent variability management, respectively. Two new POD based reduced order thermal modeling methods are presented to simulate multi-parameter dependent temperature field in complex multi-scale convective systems such as data centers. The methods results in average error norm of ~ 6%, while it can be up to ~250 times faster than CFD/HT simulation in an iterative optimization technique. Also, a simpler POD based reduced order modeling approach centered is validated against experimental measurements in an operational data center facility. The open design method application for an energy efficient air cooled data center cell with an annual increase in the power consumption shows a 12-46% reduction in the energy consumption of the center in addition to being adjustable to the newer IT equipment and higher heat loads, compared with a traditional design. Also, the method can reduce the variability in the thermal response by 74%. Also, the POD based modeling and power profiling of IT equipment is used to bring adaptability for coordinated minimization of cooling and IT power consumption in future open data centers. This coordination results in 12-70% saving in the total energy consumption of a data center cell in different scenarios.