Data center optimization using PID regulation in CFD simulations

Data center electric energy consumption becomes a major issue in HVAC (Heating Ventilating and Air Conditioning) design. In the 2011 guidelines, the ASHRAE (American Society of Heating Refrigerating and Air conditioning engineer) introduces several allowable working envelopes based on temperature and hygrometry. These envelopes encourage the designers to raise cooling air temperature, increasing the COP (Coefficient Of Performance) of the chiller and decreasing its energy consumption. But server's thermal behavior of recent blade server evolves. Their heat rejection is now mainly driven by leakage phenomenon at the CPU level and a FSCA (Fan Speed Control Algorithm) modulate their air flow rate in function of their inside temperature. In this paper we introduce a new way to model this kind of server in a Data center CFD (Computational Fluid Dynamic) code. It includes air speed control with a PID (Proportional Integral Derivative) algorithm, CPU leakage simulation and overall material electric consumption. We will then test this model with different cooling air temperature corresponding to the ASHRAE recommended range. We will try to select the appropriate cooling temperature set point to reach the best compromise between the chiller and the server energy consumption.