New Content Updates
Educational Webcast Alerts
Building Products/Technology Notices
Access Exclusive Member Content
Part 1: PUE is One of Many Metrics for Data Center Energy Performance
Part 2: Free Air, Water-Cooled Servers Increase Data Center Energy Efficiency
Part 3: UPS Energy Saver Mode, High Voltage DC Power Cut Data Center Energy Use
Part 4: In Data Centers, Simplicity of Design, Commissioning Saves Energy
By James McEnteggart
February 2012 -
Data Centers Article Use Policy
The goal of the "PUE Arms Race" is to drive down power usage effectiveness (PUE) to 1.0, where the only energy used is the energy powering the computer. Some data center owners have even claimed net PUE of less than 1.0 using photovoltaics and similar on-site power-generation technology. This race for the lowest PUE is driving the computer industry to increase the energy efficiency of computer equipment, which is a boon to all those who design, engineer, own and operate data centers. Meanwhile, leading organizations are using new technologies and approaches to designing, engineering and operating data centers. Here's a look at the trends in mechanical and electrical systems that can increase energy efficiency.
First, a few words about PUE. Calculated by dividing total site load by IT load in kilowatts (kW), PUE is a good gauge of a facility's energy efficiency, but reliance on this number alone can be misleading. For example, a new data center designed and equipped for energy efficiency and future expansion — and which is not yet operating at full design load — initially will have a poor (i.e., high) PUE. Similarly, PUE will degrade if an owner installs new servers with more energy efficient power supplies in an existing data center. It is simply a function of the way PUE is calculated. PUE also changes over the course of the year as operating conditions change with the seasons.
Other valid energy efficiency metrics include kW per computing unit; total site kW; and data center infrastructure efficiency (DCiE), which is simply the inverse of PUE. It is also valid to measure the energy efficiency of individual equipment, for example, a chiller's efficiency rating based on kW per ton of cooling. Again, these metrics are valuable, but they only provide a snapshot of one point in time.
To get a full, accurate picture of energy performance over the full year, owners and operators need to continually recalculate and average the appropriate metrics. Building automation and control systems do this automatically, providing graphic reports that provide easily interpreted, reliable documentation of the facility's performance over time.
There are three fundamental ways to improve the energy efficiency of a data center. One way is to install new computer equipment with more efficient power supplies, and this is often done as owners periodically refresh their computing equipment. Another is to implement on-site power generation, for example, cogeneration or solar power. These grand-scheme approaches are not often implemented today, but they have increasing potential as the technologies improve and their capital cost decreases.
The third approach is to design, engineer and operate data centers to maximize the efficiency of the building infrastructure. Whatever else an organization is doing, this is fundamental to improving energy efficiency. Here is a look at some of the leading trends in mechanical and electrical systems. Many of the techniques being implemented by data center owners and engineers have an established track record in non-critical facilities. With data center energy costs escalating, these techniques are making their way into the mission-critical arena.