The impact of raising data centre temperatures

Despite an ongoing industry effort to raise data centre operating temperatures, many data centre managers have been slow to follow this guidance and their caution could be well founded according to a new white paper published by Schneider Electric, the global specialist in energy management and automation.

White Paper 221, “The Unexpected Impact of Raising Data Center Temperatures” describes in detail the cooling mechanisms of a typical data centre and evaluates the trade-offs and consequences associated with particular cooling strategies. Despite the high cost associated with data centre cooling, it is of note that the potential savings on offer has so far not captured the imagination of an industry which is notoriously conservative by nature.

“Data centre operators are struggling to make decisions about raising temperatures in the white space,” said Kevin Brown, Vice President of Data Center Global Solutions and Strategy at Schneider Electric. “They want answers to questions like, is it safe to do so? What is the right temperature? Is it worth the increased risk? White Paper 221 helps to explain the implications of making the choice to raise IT temperatures.”

The idea that higher ambient temperatures inside a data centre can reduce energy costs is enshrined in the revised TC9.9 standard concerning best practices in data centres, issued by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) in 2011. By permitting higher ambient temperatures inside a data centre, it said, operators could avail of more “Free Cooling” hours in which the chillers which cool the overall data centre could be switched off.

However, the notion that Free Cooling guarantees reduced energy consumption is overly simplistic thanks to the complex dynamics of a data centre. Running at higher temperatures can produce energy savings but often it does not. For example, raising the IT temperature set point allows the chillers to avail of economizer modes for a greater part of the year which produces immediate energy savings. However, these can be offset by the greater burden placed on other parts of the cooling infrastructure. 

Schneider Electric analysed data centres in three US cities with different climates: Chicago, Seattle and Miami. In each case it compared the energy consumption and total cost of ownership of the data centre in three operating temperature scenarios. In the first case, to establish a baseline, it assumed a fixed IT inlet temperature of 20C (68F); in the second it allowed temperatures to float between 15.6 and 25.7C (60 and 80F); and in the third case it fixed temperatures at the higher level of 26.7C (80F).

The analyses of the White Paper demonstrate that there are many variables that influence cost savings (or cost penalties), and that raising temperatures is not always a good thing. Before making temperature changes to a data centre, it is important to have a solid understanding of the design conditions, system attributes, load, and so on.

The White Paper also shows that both cooling architecture and local climate conditions have a significant impact on the optimal IT temperature set point. Other factors such as server fans and the airflow CFM curve are key drivers. Overall, the research carried out to write “The Unexpected Impact of Raising Data Center Temperatures” revealed that the facilities utilising direct and indirect air economizer modes performed better than those with packaged chiller architecture.

The White Paper 221 entitled “The Unexpected Impact of Raising Data Center Temperatures” is available for free download from www.schneider-electric.com/whitepapers