Expeditious Data Center Sustainability, Flow, and Temperature Modeling: Life-Cycle Exergy Consumption Combined with a Potential Flow Based, Rankine Vortex Superposed, Predictive Method
Traditionally data centers were designed with reliability, functionality, and up front cost as the primary design drivers, with secondary consideration given to cost of operation and sustainability. In the past decade, data center design has shifted giving more weight to operational energy use and sustainability. This thesis outlines two work streams aimed at allowing data center designers and operators to design more efficient, cost effective, sustainable data centers. The first track deals with evaluating data center sustainability quickly and easily. The metric chosen to quantify data center sustainability is life-cycle exergy consumption. Other sustainability indicators are investigated, and reasons for choosing this metric for data centers are provided. Two case studies are provided to illustrate the sustainability ramifications of different design and operation decisions. A Life-Cycle Assessment (LCA), perhaps the most widely accepted industry practice of assessing the environmental ramifications of a product or process, is also completed for comparison purposes. A MATLAB based graphical user interface (GUI) developed to perform a life-cycle exergy consumption analysis (LCEA) on data centers was designed and its usefulness is demonstrated.
The second track of this thesis deals with expeditious velocity and temperature field prediction in data centers, developed to be an ultrafast alternative to conventional computational fluid dynamics (CFD) code. This expeditious predictive scheme is based on a potential flow velocity prediction method with a Rankine vortex superposition correction scheme to correct for physical processes neglected by the relatively simpler but much more computationally efficient potential flow equations. Energy considerations are accomplished through convective energy transport equations. It is shown that without the Rankine vortex superposition correction, potential flow is inadequate for accurately predicting velocity and temperature fields in data centers. It is also shown that with the Rankine vortex superposition, in areas of interest within data centers, temperature predictions are as accurate or are more accurate than a commercially available, conventional CFD software package that uses the full Navier-Stokes and energy equations to solve for velocity and temperature fields.
Three case studies are carried out based on experiments that were conducted in a data center
at HP Labs in Palo Alto, California. Measurements taken in that data center are used for comparison with CFD predictions from ANSYS FLUENT and a MATLAB based, potential flow with Rankine vortex code named COMPACT (Compact Model of Potential Flow and Convective Transport) . FLUENT's temperature field predictions are also compared with COMPACT's predictions for both computational speed and accuracy.
The ultimate goal of these two streams, when combined, is to give a data center designer or operator an ultrafast, comprehensive tool to make data center sustainability decisions while constrained by operational temperature limit considerations. It is shown in this work that such a goal is attainable.