When it comes to cooling critical systems, conventional methods are simply proving to be increasingly cost-ineffective and inefficient. In March 2009, the AFCOM Data Center Institute reported that over next 5 years more than 90% of all companies will experience power failures and limits on power availability in their data operations.[1]
The recent economic crisis has further amplified the focus on the need to remove unnecessary costs from network infrastructure operations without affecting performance. The major expense and huge energy demand associated with data center cooling have moved the discussion from an afterthought into the strategic planning boardroom. When, where and how new facilities will be built, as well as retrofitting existing facilities are all under scrutiny.
In addition, the increasing awareness about eco-sustainability has added a sense of urgency to reduce the carbon footprint of enterprises and operators alike. Corporate responsibility concerns and eco-sustainable initiatives are impacting buying and operating decisions. A green business strategy is shaping the data center design.
Eco-sustainable solutions that maximize heat transfer and support the higher heat density equipment frames are available on the market today. These solutions can reduce energy consumption by as much as 90%, while simultaneously shrinking the carbon footprint, optimizing data center space utilization, and lowering operational expenditures (OPEX).
The key to understanding cooling savings and operational efficiency gains is to understand their coefficient of performance (COP). COP is the ratio of cooling capacity to the amount of power used.
(Click to continue)
Table 1 shows the operational efficiency of a refrigerant-based system compared to an air conditioning-based system.
Table 1: AC-based versus refrigerant-based COP
Based on the example in Table 1, the second system is 8 times more efficient.
Table 2 provides a comparative example of a data center with an environment in which 320 kilowatts must be cooled.
Table 2: Comparative example: AC-based versus refrigerant-based cooling
Even accounting for the more expensive initial cost of the refrigerant system, in this case US $20,000, it still provides payback in roughly 10 months. The AC system would require 4.9 years to achieve a full return-on-investment (ROI).
New solutions offer new options
Data center operators must figure out a way to keep the air cool within the operating environment where the heat-generating equipment resides. The challenge is that high heat density equipment racks cause “hot spots” and an irregular temperature distribution within the site. The best way to improve efficiency dramatically and avoid this issue is to remove heat right at the source.
Other actions include the improvement of AC-based cooling methods, such as:
- Cool aisle/warm aisle reconfiguration
- Hot aisle containment
- Cold aisle containment
- Dedicated cool air delivery and warm air return ducting
Chilled water plants offer another cooling option — although initially more costly to install than traditional AC-based systems — they offer a greater long-term ROI. This solution also provides flexibility because chilled water can be piped to almost any location in a building.
Chilled water plants allow improved efficiencies through the use of multiple cooling loops that take advantage of outdoor conditions and seasonal temperatures. Advantages of chilled water cooling include:
- Lower operating costs for the cooling system
- Reduced energy consumption required for cooling
- Ability to house more equipment in the data center because the cost and energy to cool the facility is reduced
Other alternatives include water or refrigerant fluid cooling systems, which are either aisle- or dedicated cabinet-based, provided by several HVAC and server companies entering the market. Yet another approach is fresh air or free air cooling.
Challenges of the past drive change
For years service provider executives accepted that data center cooling was an inevitable, expensive and unavoidable cost of doing business. The most common way that service providers and enterprises cooled their data centers was by using large AC units to flood the facility with cool air using large, very noisy air movers to provide circulation of the air. This cool air would then be pulled across the electronic systems by internal fans built into the servers. The heated exhaust would be forced out the back of the cabinet and re-collected by the air conditioning system. It would then be cooled by either the chilled water system or the internal compressor cooling stage. While this approach has adequately performed the task in the past, it presented a number of limitations.
- Air is a poor medium for heat transport, which means this method takes a significant amount of energy to continuously recycle the cold and warm air.
- It is very difficult to segregate the cold and warm air, which means the cool air can mix with warm air — even before being drawn into the servers.
- Data center operators must limit the amount of server equipment in each cabinet to manage the heat load of the facilities that house their systems.
Typically, data centers have had to restrict the total heat load to between 3 to 4 kilowatts of heat per equipment frame. This meant poor utilization of very expensive and finite data center space.
A better way forward
The power consumed by data center demands will continue to rise. The industry requires an efficient solution to reduce the power consumption and costs associated with cooling.
Telco requirements such as the Network Equipment Building System (NEBS) standards provide information about the environment of a typical United States RBOC Central Office. The European Telecommunications Standardization Institute (ETSI) performs a similar function for Europe and the rest of the world. Because of the mission-critical nature of the information handled in data centers, it is anticipated that similar hardware requirements will be adopted for this environment as well. Extensive experience in meeting the stringent NEBS and ETSI requirements provides a distinct advantage. This experience is designed, for instance, into the Alcatel-Lucent Modular Cooling solution.
Modular cooling is a highly efficient, low-pressure, oil-free refrigerant-based solution that can reduce costs compared to conventional cooling methods. And it can be deployed in a customer data center with no physical changes to the present data center configuration.
The solution transports waste heat generated by computer servers from each cabinet or rack within a data center to the location’s heat sink, typically a chilled water system. This method of cooling provides better space utilization by allowing more equipment per individual cabinet and more cabinets per data center with minimal impact on ambient air temperature.
The modular cooling solution uses a phase-changing refrigerant process, removing heat as it exits servers instead of discharging waste heat to the ambient air for further transport. The refrigerant is non-carcinogenic and converts to a gas at standard temperature and pressure. To support this solution, consulting and deployment services and material delivery processes are available to produce a turnkey solution for customers.
While the physics of transferring heat remain constant, there have been some innovative developments in how data center operators optimize components in the heat transfer path to enable higher efficiencies. Today enterprises and operators can choose a strategy that allows them to meet their business objectives and provide eco-sustainable data cooling solutions.
To contact the authors or request additional information, please send email to enrich.editor@alcatel-lucent.com.
Footnotes
Loading ...
Loading ...
Comments
Post a Comment
Comments are moderated and will be published/addressed upon review.