Effect of Thermal Characteristics of Electronic Enclosures on Dynamic Data Center Performance

Abstract

Data centers are the facilities that house large number of computer servers that dissipate high power. Considering the dynamics of the data centers, their efficient thermal management is a big challenge that needs to be addressed. Computational analysis using a CFD code is very useful technique that helps the engineer to understand and solve the data center cooling problem. Several ongoing numerical modeling research efforts assume the computer room air conditioning (CRAC) units as fixed flow devices with constant temperature boundary condition. In reality, CRAC supply temperature is governed by the thermal characteristic curve, as specified by vendor. In this paper, study is presented by incorporating the CRAC thermal characteristic curve in the numerical model. Case studies are presented to show how the segregated high and low powered clusters in a data center may affect the supply temperatures from the CRAC in their vicinity. Another concern that is crucial in analyzing data centers performance precisely is the effect of buoyancy and thermal mass on the facility environment. In some cases, the effect of thermal mass and buoyancy may cause unexpected behaviors such as temperature overshoot or rapid variations in temperature. Non-dimensional parameters are used to demonstrate the effects of thermal mass and buoyancy.