Critical Facilities: Power Audits
Part 1: Data Center Power Audits Can Improve Efficiency
Data Center Power Audits Can Improve Efficiency
By James McEnteggart - November 2012 - Data Centers
Power density for a data center can range from 25 W/sq. ft. in an older facility to a whopping 300 W/sq. ft. in a new, high-density data center. Do the math, as they say, and total power consumption might seem astronomical. But total power consumption does not matter as much as the efficiency of power usage, and, in most data centers, there is room for improvement. A power audit is the first step toward quantifying a data center's power usage effectiveness (PUE), which is a good gauge of a facility's efficiency. This knowledge will enable an owner or facility manager to identify opportunities to improve efficiency and reduce operating costs.
How To Take That First Step
The power audit focuses on the fundamental constituents of power usage in the data center: electricity used to power information technology (IT); electricity used to remove the heat from the space (HVAC); and electrical losses (EL) at transformers and uninterruptible power supply (UPS) systems.
The computing function can be defined as the "useful work" of the data center; for the purposes of the audit, it can be measured in kilowatts (kW) for point in time measurements or in kilowatt-hours (kWh) for usage over time. Efficiency is defined as useful work (kW) divided by the power required to produce the useful work. Measuring the components of power required to deliver the useful work will provide insight into the efficiency of the data center.
Begin by developing a data collection plan listing the equipment that comprises IT, HVAC and EL loads. Simply trace the path from the utility power source to the racks, and list everything on the path. Similarly, trace the heat transfer path from the data center to the external environment, and list all of the equipment.
Next, classify the loads. Classify equipment by load profile: steady (i.e., power is consumed whenever the device is energized and independent of amount of work performed); and variable (i.e., power use changes with amount of work performed). This distinction is important because steady loads can be measured at a single point in time, while variable loads must be measured over time and in conjunction with other system parameters.
Collecting energy-usage data takes time and costs money, so it is important to determine the relevance of various components to improving the efficiency of the data center. For example, lighting typically consumes 1 W/sq. ft., while the IT load consumes 25 to 300 W/sq. ft. So improving the efficiency of the lighting system is good, but measuring its power consumption down to the last kWH will have little relevance to improving the overall efficiency of the data center.