Comparison of Numerical Modeling to Experimental Data in a Small Data Center Test Cell

Information Technology (IT) equipment compaction has become a significant air cooling challenge at the data center level. Computational Fluid Dynamics and Heat Transfer (CFD/HT) models have been employed as the dominant technique for the design and optimization of both new and existing data centers. Understanding the limitations of a CFD/HT models’ ability to predict the actual data center temperature field and flow characteristics becomes a critical component in optimizing the actual data center rather than optimizing the model of the data center. This is most important near the IT equipment where temperature and flow specifications from the IT equipment manufacturers must be maintained for reliable operation. This study is a continuation of earlier comparisons of CFD/HT models to experimentally measured temperature and flow fields in a small data center test cell. This study compares the experimentally collected data for three different layouts of perforated tiles to a CFD/HT model with seven turbulence models not previously evaluated. Insight into the location of the deviation between the different turbulence models and experimental data are discussed, along with the computational effort involved in running the CFD/HT models. It was found that the zero equation (or mixing length model) and the Spalart-Allamaras turbulence models produced the smallest deviations from experimental data, but the former required only a fifth of the computational effort of the latter. The laminar flow model required the least computational effort, running more than twice as fast as the zero equation turbulence model, and produced deviations similar to those of the six different k-ε turbulence models.