With energy consumption still rising, improving data center efficiency is more important than ever for reducing costs. Based on a 600 kilowatt (kW) load, each percentage point in efficiency improvement translates into $6,000 in annual savings in operating expenses. Additionally, efficiency is viewed as a strategy to manage capacity (possibly saving capital costs) and promote environmental responsibility.

The data center offers a number of routes from grid to chip to lower energy usage, but the cooling system is the major roadway. In a typical 5,000-sq-ft data center, cooling has been found to account for 38% of total energy consumption. This means the cooling system is where the biggest efficiency impact can be made.

Unfortunately, the typical practice of making cooling unit improvements one at a time misses the opportunity to optimize the entire cooling system, and in turn, maximize savings from all efficiency opportunities. In comparison, taking a holistic approach to improving cooling system efficiency enables changes that boost overall performance, availability and capacity; achieve better return on investment (ROI); improve decision making for future expansion; and save money. In fact, with a holistic approach, cooling system energy consumption can be reduced by up to 70% .

There’s no one magic formula for getting to this maximum level of efficiency. Rather, it’s a combination of system adjustments and/or technology upgrades, which will be discussed in this article.


A variety of technologies are now available with vastly improved efficiency profiles. Before evaluating these technologies, however, it’s necessary to have a clear picture of your data center’s current state. You can’t change what you don’t know, so the first step toward significant cooling system energy savings is a cooling assessment. A cooling assessment will reveal the information needed to understand a data center’s cooling system performance and will help you make informed decisions on how to improve it. For example, a professional cooling assessment will identify hot spots, airflow issues, and inefficiencies in existing room conditions and practices.

There are two types of assessments. A periodic cooling assessment reveals information about the data center at a specific point in time and is used as a baseline for comparing the next cooling assessment, which may not be conducted for another year or more. Compared to a periodic assessment, a dynamic cooling assessment continuously collects and analyzes information to help manage change and maintain an optimal environment in the ever-evolving data center. A dynamic assessment plays a valuable role in a holistic strategy to improve system efficiency.

Continuous thermal study of the data center through a dynamic cooling assessment provides real-time engineering analysis of the cooling infrastructure. Because temperatures are tracked continuously, the resulting temperature trending can act as an early warning system by offering visibility to future problems that could lead to a data center outage. Examples include identifying future hot spots caused by increasing heat loads and/or cooling system capacity issues or the discovery of over cooled areas that could provide energy-saving opportunities. Solutions are recommended to address any concerning environmental conditions that are found.

A dynamic cooling assessment provides the same performance and planning benefits as a periodic cooling assessment. However, dynamic cooling assessments also evaluate seasonal changes that impact the cooling system and incorporate real-time temperature variations in the heat load caused by server load fluctuations or by adding new information technology (IT) equipment.

Cooling experts play an important role in a dynamic cooling assessment. They collect, review, and analyze environmental data and assess cooling effectiveness and efficiency. Quarterly reports prepared by a cooling expert include trending analysis, identification of potential cooling problems, and specific recommendations to improve the availability and efficiency of the cooling infrastructure.

When a problem is identified through ongoing evaluation, cooling experts can perform on-site analysis of cooling infrastructure to identify the root cause of hot spots and inefficiencies. Identifying the root cause increases the chance that a problem can be taken care of once and for all.

Dynamic cooling assessments offer a long-term view of the changing data center environment. With this perspective, availability and efficiency problems can be identified and prioritized to ensure the cooling infrastructure and strategies keep pace with change. Cooling experts analyze the impact of upcoming equipment or infrastructure changes before they are implemented, which makes a dynamic cooling assessment a valuable tool for future planning. Equipment can be replaced or upgraded with the knowledge that the cooling infrastructure can handle changes in IT loads. Airflow and temperature changes can be easily managed when data center inventory ages, is moved, or is enhanced with new technologies. Additionally, existing or potential cooling problems can be identified in advance to support critical system availability and energy efficiency, as well as prevent costly problems before they occur.


Whether before or after cooling system changes are made, adhering to best practices improves cooling efficiency, resulting in money saved and better data center performance. These best practices include:

• Sealing moisture out of the data center with consistent vapor barrier.

• Optimizing airflow (including arranging server racks in a hot aisle/cold aisle configuration).

• Raising supply air temperatures without increasing server fan speeds — ideally staying within recommendations of the American Society of Heating, Refrigerating and Air Conditioning Engineers (ASHRAE), which is from 64.4ºF to 80.6º/18ºC to 27º.

• Adding blanking panels to any open U-slot.

• Supplying conditioned air to the inlet of the server or communication equipment only.

• Locating perforated tiles and discharge ducts in the cold aisle. Putting them in the hot aisle or elsewhere will reduce overall capacity and efficiency.

• Implementing system monitoring as part of a preventive maintenance program.

• Having periodic maintenance performed on cooling systems to maintain factory performance.

• Periodically assessing cooling effectiveness.

• Matching cooling and airflow to server requirements.


Once inefficiencies are identified, there are a variety of technologies and approaches that can be implemented to improve efficiency and reduce operational expenditures. The fans in the cooling units are often a good place to begin making changes because they account for up to 76% of unit energy consumption on chilled water units (for fixed speed centrifugal fans traditionally used in precision cooling units). This is important because fans that move air and pressurize the data center’s raised floor are a significant component of cooling system energy use. In fact, on chilled water systems, they are the biggest energy consumer.

The most efficient cooling system is one that matches needs to requirements. This is a challenge in the data center because cooling units are sized for peak demand, which rarely occurs in most applications. Fortunately, cooling units can be equipped with variable speed fan technologies that enable them to operate more efficiently when not at peak load.

Variable-speed drives (VSDs), also known as variable-frequency drives (VFDs), can be added to centrifugal fans, enabling the fan speed to be adjusted based on operating conditions and significantly improving energy efficiency. For example, adding a VSD to the fan motor of a chilled water precision cooling unit allows the fan’s speed and power draw to be reduced as load decreases, resulting in a dramatic impact on fan energy consumption. A 20% reduction in fan speed provides almost 50% savings in fan power consumption.

Like VSDs, electrically commutated (EC) fans are more efficient at partial loads because less fan power is needed. However, unlike VSDs, EC fans save energy even when the cooling unit is at full load. EC fans use a brushless EC motor in a backward curved motorized impeller (plug fan).

EC fans are inherently more efficient than traditional centrifugal fans because of the fan design and the fact that they are direct drive with no belts. This means there are no belt losses, which constitute approximately 5% of the fan’s total energy consumption. EC fans also reduce system losses associated with how air is directed through the cooling unit cabinet and discharged out of the unit (airflow).

Independent testing of EC fan energy consumption compared to VSDs found that EC fans mounted inside the cooling unit created an 18% savings. When EC fans were mounted outside the unit below the raised floor, savings increased to 30% due to the higher volume of underfloor airflow.

Both VSDs and EC fans can be installed on existing chilled water and direct expansion (DX) cooling units with appropriate controls in place, or specified in new units. A factory-grade installation is a must when fans on existing units are upgraded. Whether it is paired with a legacy control or the newest controller, the unit will be able to constantly adjust the fan speed to match airflow, providing the most efficient amount of cooling to the servers. All set points (thermostat, pressure, transducer, and others) should be checked to ensure everything works together to deliver maximum efficiency.

Working with service professionals to install these cooling fan technologies — and modify them based on conditions in the data center — creates the conditions for a data center in which efficiency can be optimized in ways that simply weren’t possible five years ago.


Designing intelligence into the infrastructure is one of the most effective ways to improve data center efficiency. Using intelligent controls alone ensures the optimal combination of compressor/chiller capacity and airflow based on monitoring temperature at the servers. This optimization may allow inlet temperature set points to be between the typical nameplate temperature (75ºF/24ºC) and the ASHRAE-recommended limit (80.6º F/27ºC) for class A1-A4 data centers, assuming all legacy IT hardware is compliant. Additionally, intelligent controls allow all the cooling units to run at a reduced load by allowing them to operate together as a system utilizing teamwork. With aisle containment in place, this reduction in load increases total operating efficiency by an additional 15% on average.


Data center infrastructure management (DCIM) is perhaps the approach that most enables efficiency in cooling systems, and subsequently, entire data centers. Where an on-site efficiency assessment examines the physical components of your infrastructure, DCIM combines facilities management with IT for more holistic insight. DCIM treats critical infrastructure as a symbiotic ecosystem. For example, computing demand is reported in real time, therefore space, power, and cooling can be allocated based on actual demand — eliminating the need to over-provision. This increases efficiency and prevents wasting much needed capacity.

With the ability to see infrastructure performance in real time, there’s no need to guess at capacity or efficiency thus enabling the most informed decision making possible when selecting new, highly efficient cooling technologies.


A holistic approach to improving cooling system efficiency includes knowing the data center’s existing inefficiencies, and observing best practices such as using variable fan technology and implementing intelligent controls. Ultimately, adding DCIM will provide real-time insight into how the cooling system interacts with the rest of the data center ecosystem, to realize the greatest efficiency. Taking this holistic approach will result in better performance, availability, capacity, and ROI, as well as improved decision making and lower operating costs.