Data centers are facilities that house large numbers of computer servers that dissipate high power. With the rapid increase in the heat flux of such systems, their thermal management has become a challenge that needs to be addressed. Computational analyses using a CFD code is a very useful technique that helps the engineer to understand and solve the data center cooling problem. In this paper the state of the art of numerical modeling of data center is discussed. Representative systems are modeled using the two most prominent approaches. Variation in results with the addition of modeling details is presented. The effect of under floor parameters such as the conditioned chilled air supply flow rate, the under floor plenum depth, and the tile opening flow resistance, is discussed. Total flow rate delivered by the Computer Room Air Conditioning (CRAC) unit depends on blower and system characteristic curves, as specified by the vendor. Impact of plenum depth and tile resistance on total CRAC flow rate is discussed. Under floor blockages such as cables, pipes, and random materials, impede the flow of the cold air stream and yield unpredictable air flow patterns. Currently, models with idealized plenums are used for simulation of data centers. The effect of including blockages in CFD analyses is discussed. A novel approach that defines safe and critical paths under plenum for routing the blockages is presented.
Numerical Modeling of Data Center Clusters: Impact of Model Complexity
- Views Icon Views
- Share Icon Share
- Search Site
Bhopte, S, Iyengar, MK, Sammakia, B, Schmidt, R, & Agonafer, D. "Numerical Modeling of Data Center Clusters: Impact of Model Complexity." Proceedings of the ASME 2006 International Mechanical Engineering Congress and Exposition. Heat Transfer, Volume 3. Chicago, Illinois, USA. November 5–10, 2006. pp. 51-60. ASME. https://doi.org/10.1115/IMECE2006-13494
Download citation file: