Data Center Cooling: Keep Your Cool
Best practices for energy efficiency and reliability in data center cooling.
According to the U.S. General Services Administration (GSA), less than half the power used by a typical data center powers its IT equipment. The other half is used to support infrastructure that includes cooling systems, uninterruptible power supply inefficiencies, power distribution losses, and lighting.
The complex technology housed in data centers and the continuous uptime demands on the facilities make reliability of heating, ventilation, and air conditioning (HVAC) systems critical. Choosing energy efficient and sustainable cooling technologies that address rising energy costs and heat load densities is a priority for many data centers.
There are many HVAC solutions available that provide greater energy efficiency and sustainability, while also offering a high level of reliability. Keeping in mind a few key factors and best practices when designing and selecting a chilled water cooling solution will help ensure reliable and efficient performance.
KEY ISSUES FOR DATA CENTERS
Several key issues and trends impact mission critical cooling system selection and design. As server technology advances, more powerful equipment is being used in data centers. This often results in more heat being produced per square foot, causing the heat load density in server rooms to increase. Over the last 10 years electrical and mechanical system design and operating procedures have evolved, which has allowed for the HVAC industry to provide higher system reliability.
As data center equipment evolves and is able to operate at higher temperatures, some data centers are increasing building temperatures in response to changing industry guidelines. As part of ongoing data center efficiency improvements, the GSA recommended raising data center temperatures from an average of 72°F to 80°F. Based on industry best practices, the GSA estimated it could save 4% to 5% in energy costs for every 1° increase in server inlet temperature, according to the association’s 2011 Data Center Consolidation Plan.
The American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) has also raised the upper end of its recommended operating temperature range for data center servers, from 77°F to 80.6°F.
Finding ways to improve efficiency also helps data centers achieve a lower power usage efficiency (PUE, also referred to as power usage effectiveness). PUE is an industry metric that measures how effective a data center is in using input power. It is determined by dividing the total amount of facility energy by the amount of energy used to run the IT equipment within it. The larger the number, the less efficient the energy utilization — with overall efficiency improving as the number decreases toward 1.0.
For example, according to the GSA, improving PUE from 2.0 to 1.6 for a data center with a 2.5 MW IT load yields a 20% energy savings — or more than $800,000 in annual savings at $0.08/kilowatt hour. Considering these potential savings, data centers want equipment and solutions that will result in the lowest possible PUE.
WAYS TO IMPROVE EFFICIENCY
There are various approaches to data center cooling. Keeping key critical factors in mind when selecting and designing cooling systems will help improve overall data center efficiency.
One option to consider is to include an airside economizer in the air handler design. In various climate regions there are times of the year when a system can use outdoor conditions to cool the process using the standard cooling components to distribute its cooling effect. The most prevalent technique is an air economizer, which reduces or eliminates mechanical cooling for much of the year in many climates.
In climates or building applications where an air economizer is not practical, a waterside economizer can reduce compressor run hours and energy use. With a waterside economizer, the supply air of a cooling system is cooled indirectly with water that is itself cooled by heat or mass transfer to the environment without the use of mechanical cooling.
In addition, some equipment is designed to react quickly and adapt to system changes or to operate at broader temperature ranges. These factors are important when considering chiller technology for data center applications.
IMPORTANT CONSIDERATIONS IN CHILLER SYSTEM DESIGN
When selecting and designing a chiller system, reliability and energy efficiency are the two of the most important factors. This is especially true in tier three and four data centers where 100% reliability is critical.
A simple chiller design contributes to greater reliability and less risk for unplanned downtime. Chiller plant efficiency is improved by using lower flow, equals higher delta T, and chilled water setpoints that range between 60°F to 65°F.
Simplicity in chiller design also helps save time and money on maintenance. Direct-drive compressor technology is an example of simplified design that can provide these benefits. With this technology, the motor is directly coupled to the centrifugal compressor, resulting in only one moving part. Direct-drive technology also offers higher efficiency because it is optimized for variable speed operation.
Rapid restart technology is another key advancement that contributes to chiller performance. After a power interruption, a chiller designed with rapid restart can quickly regain full operational capacity. This allows mission critical applications to continue with minimal interruption.
Design and modeling tools are available to help determine which technology and solutions are best suited for data center applications. Energy modeling software and CAD software help predict outcomes and reliability — aiding in the chiller selection process.
BENEFITS OF AIR COOLED OPTIONS
In climates where conditions are favorable, the least expensive way to cool a data center is with outside air that doesn’t require mechanical cooling. However, this is not always feasible in warmer climates where air pollution or other factors are a concern or where outside air is not readily available.
In these situations, compressor technology may be the preferred solution. For data center applications that use compressor technology, choosing an air cooled chiller provides benefits for energy efficiency while still offering the high reliability necessary for the industry.
On an annual basis, air cooled chillers can be equally or even more efficient than a water cooled system depending on the geographic climate. In most climates air cooled chillers are more efficient at night and in the non-summer months because the outside air cools the condenser. On days when the temperatures are hotter, water cooled systems are more efficient because the compressor energy is reduced. Because air cooled chillers do not have a condenser water pump, the difference in total compressor and heat-of-rejection energy compared to a water cooled system makes air cooled chillers more efficient when the ambient temperatures are below 68°F.
Additionally, air cooled systems do not require portable water, water treatment, or sewage costs, since no water evaporation is being used for cooling. Water cooled systems may require a 72-hr supply of water stored onsite for operation and regular water treatment. The increased number of moving parts and control points all introduce a reliability factor into the system.
The resulting total life cycle cost benefits may favor air cooled systems in some applications.
Efficiency improvements can also result from incorporating advanced system controls. Web-enabled, scalable building automation systems (BAS) help optimize energy efficiency and deliver reliable and sustainable system performance over the life of the data center.
Installing properly sequenced controls is not the end of optimization. Regular and ongoing monitoring of equipment is the key to keeping a cooling system maintained and performing as designed.
A few manufacturers have proven unit control algorithms built into the chillers to ensure operation under distressed conditions. These unit operations are designed to adapt during a stressful condition to provide as much chilling capacity as possible while also producing external alarm notification. This adaptive control technique is preferred over total unit shutdown or failure that could occur when controls do not use adaptive technology.
In addition, consider a manufacturer or service provider that offers secure, 24/7 system monitoring, tech support, and response capabilities to help keep mission critical systems optimized.
Selecting the most appropriate chiller system — and considering technologies like rapid restart, adaptive controls, and waterside economizers — helps optimize data center operation and improve efficiency. In addition, ongoing system monitoring by product experts can identify operating trends that help uncover additional energy optimization opportunities or characteristics of impending issues to ensure continued efficient operation with maximum uptime.