Variable-Speed Compressor Enabled Energy Efficient Cooling of Blade Servers.
Abstract: Largely due
to the power consumption of their thermal management systems, data centers, consume over
2% of the electricity produced in the United States. Therefore a fundamental
understanding of data center energy management is both environmentally and economically
important. In previous work, models were developed to explore the energy consumption of
the thermal management. Both ... read moreconventional air-cooling and hybrid cooling, where
microprocessors are liquid cooled and other components are air cooled, were considered.
These models were based on chillers which are not optimal for demonstrating energy
savings as the compressor is not variable speed and thus consumes essentially the same
amount of energy irrespective of its set point. In addition, the analytical models
include several oversimplifications especially with respect to the chiller sub-models
(e.g., compressor work computation and heat transfer characteristics of the condenser,
etc.). The analytical model developed in this thesis was based on the physical
properties and operating characteristics of real world components, which were validated
and integrated based on thermo-fluid principles. As a result of exercising the model
between the air-cooled and hybrid cooled systems it was found that the refrigerant mass
flow rate can be reduced by 40-50% or an energy savings of up to 420 W. This
relationship is similar to the previous work conducted by Hwang, however the amount of
savings is different due to modeling real world components. Additionally the model
demonstrates the sensitivity to ambient air temperature, as temperature was increased
from 20 ºC to 40 ºC, through a 58% reduction in system performance or an
increase in compressor work from 224 W to 525 W. The model also revealed various
relationships between the system performance as related to the system pressure rise and
the refrigerant mass flow rate. The integrated analytical model shows there are
opportunities for energy reduction in the thermal management systems for data centers.
Additionally this thesis demonstrates the importance of component sizing, and the impact
ambient air temperature has on system performance. Finally this model found that the
system performance was a function of both the pressure rise over the compressor and the
refrigerant mass flow rate.
Thesis (M.S.)--Tufts University, 2011.
Submitted to the Dept. of Mechanical Engineering.
Advisors: Marc Hodes, and Vincent Manno.
Committee: Alan Lyons.
Keywords: Mechanical Engineering, and Energy.read less