Data Center Airflow Prediction With an Enhanced Potential Flow Model

Abstract

Potential flow models (PFM) have been implemented for a variety of applications, including data center airflow and temperature estimation. As an approximate solution to the data center room physics, potential flow models have great value in their simplicity and the limited computational effort required providing estimates. However, potential flow models lack the ability to capture the effects of buoyancy, which can affect airflow patterns within data centers. We show how this effect can be simulated within PFM; resulting in a model we call Enhanced PFM (EPFM). This model is only marginally more complex to implement than PFM and retains much of the properties of the original PFM, specifically its simplicity and stability. Solution time, about double that of PFM, is still only a small fraction of that of CFD, while empirical tests show a marked improvement in the prediction of key data center temperatures.