Thermodynamic analysis of coupled mechanical and power systems in data centers

Abstract

Efficient power system design in modern data centers is of great importance due to the rapid growth of their energy demand. This study determines the impact of modeling the inherent coupling between data center power distribution and cooling systems on their predicted data center energy consumption. A novel solution technique that performs this coupling was created in the flow network modeling tool Villanova Thermodynamic Analysis of Systems (VTAS). First, a standalone power distribution framework and its associated component models were implemented into the tool, which were used to compare the energy efficiency of AC and DC-based power distribution strategies. Then, the power and cooling systems were coupled in an analysis of a test data center, showing that the inclusion of the electrical system power losses in the calculations results in a fourfold increase in predicted cooling system partial power usage effectiveness (pPUE). The results demonstrate the significance of including the aforementioned coupling in predictions of data center energy efficiency.