Optimization of Data Center Cooling Efficiency Using Reduced Order Flow Modeling Within a Flow Network Modeling Approach

Much of the energy use by data centers is attributed to the energy needed to cool the data centers. Thus, improving the cooling efficiency and thermal management of data centers can translate to direct and significant economic benefits. However, data centers are complex systems containing a significant number of components or sub-systems (e.g., servers, fans, pumps, and heat exchangers) that must be considered in any synergistic data center thermal efficiency optimization effort. The Villanova Thermodynamic Analysis of Systems (VTAS) is a flow network tool for performance prediction and design optimization of data centers. VTAS models the thermodynamics, fluid mechanics, and heat transfer inherent to an entire data center system, including contributions by individual servers, the data center airspace, and the HVAC components. VTAS can be employed to identify the optimal cooling strategy among various alternatives by computing the exergy destruction of the overall data center system and the various components in the system for each alternative. Exergy or “available energy” has been used to identify components and wasteful practices that contribute significantly in cooling inefficiency of data centers including room air recirculation — premature mixing of hot and cold air streams in a data center. Flow network models are inadequate in accurately predicting the magnitude of airflow exergy destruction due to simplifying assumptions and the three-dimensional nature of the flow pattern in the room. On the other hand, CFD simulations are time consuming, making them impractical for iterative-based design optimization approaches. In this paper we demonstrate a hybrid strategy, in which a proper orthogonal decomposition (POD) based airflow modeling approach developed from CFD simulation data is implemented in VTAS for predicting the room airflow exergy destruction. The reduced order POD tool in VTAS provides higher accuracy than 1-D flow network models and is computationally more efficient than 3-D CFD simulations. The present VTAS – POD tool has been applied to a data center cell to illustrate the use of exergy destruction minimization as an objective function for data center thermal efficiency optimization.