Airedale regional manager Stuart Kay examines how integrated, high efficiency cooling solutions can reduce the data centre PUE.
The EC estimates the 56TWh of power used by data centres in Western Europe in 2007 is likely to double by 2020. Given the constraints of power supply, the more energy efficient the data centre, the greater the power capacity available for computing. And the facility user will pay for less energy.

The Power Usage Effectiveness (PUE), developed by the Green Grid consortium, has been broadly adopted as a benchmark to enable data centre operators to estimate energy efficiency and make positive changes. The more efficient the data centre, the closer the PUE is to 1, indicating a greater proportion of the power required is used to drive IT equipment.

Since cooling accounts for around 58 per cent of a typical legacy data centre power usage, cooling is an area where the data centre manager can substantially drive down the PUE.


Traditional on/off cooling is inefficient. Power use in the data centre is not static and most servers are under-utilised. An intelligent, efficient cooling system can vary the amount of cooling to match changing heat loads.

Fan efficiency is pivotal to reducing cooling power. In a 24/7 application, just 1 kW of power saved per hour during a year, is equivalent to around 4,020 kg CO₂.

When heat exchangers, airside systems and/or outdoor units are engineered with low air flow resistance to accommodate EC (electronically commutated) fans, there is enhanced fan efficiency particularly at part load. By applying intelligent controls, an EC fan is up to 70 per cent more efficient under part load than an AC (alternating current) fan.

By reducing air volume in response to changes in room demand, the fan power use is significantly cut. Typically a 50 per cent drop in air volume will result in an 83 per cent reduction in fan power input.

The variable speed control of latest 30 - 90 Hz inverter driven scroll compressors featured in some high efficiency airside and liquid chillers, allows fully modulated cooling from 25 per cent - 100 per cent when operating at part load, to match the cooling demand exactly.

Revolutionary centrifugal compressors present near silent, oil-free operation and ultra efficient, infinitely variable speed control. Chillers incorporating this technology can offer significantly low energy mechanical cooling because they are able to match load requirements exactly. Typically, centrifugal compressors will reduce running power by 20 per cent compared with a high efficiency screw chiller.

Both these compressor technologies have low starting currents, removing starting spikes.

Mechanical input

In a free-cooling system, cooler ambient air is moved across a large surface air-to-water heat exchanger to pre-cool the secondary cooling medium. Only if additional cooling is needed will a mechanical input be required. Since free-cooling operates with as little as 1ºC differential between ambient and return fluid temperatures, the continuous system operation and the high temperatures of a server environment in relation to ambient air, mean a free-cooling chiller or dry cooler will use little mechanical cooling.

By dealing with a higher grade heat and subsequent higher processor air off temperatures, a larger proportion of the year can be spent in free-cooling mode. A chiller designed with integrated, simultaneous mechanical/free-cooling achieves 0-100 per cent free-cooling and will always use free-cooling first before initiating mechanical cooling. Such a chiller typically saves more than 45 per cent¹ of the energy consumed by a conventional chiller.

Taking these technologies and by using advanced controls, many rack-based heat exchangers, airside units and chillers can be integrated to create an intelligent solution in which cooling products interact and communicate.

The solution is managed from a single controls platform, with two-way communication, alarm handling and an IP address that can sit on the Ethernet and be browsed remotely.

Significant power can be saved by managing air volume to the space across multiple airside units by sharing airflow duty between run units and standby units; all units run at reduced capacity and deliver greater efficiency. The more units, the greater is the benefit. In the graph above, running three airside units instead of two, gives a 70 per cent saving in power of 5.3 kW. During a year this represents a saving of 46,428 kW/hs of energy. Unit airflow is managed by either measuring the airflow across all units and sharing the total required for the space, or by measuring the floor void pressure and averaging between units. Either option relies using the standby unit.

By monitoring the room load and automatically adjusting chilled water setpoints to match this load, we see a significant increase in plant energy efficiency at the higher water temperatures and increase the free-cooling threshold. For a 1°C increase in fluid temperature we see an 8.5 per cent² increase in chiller energy efficiency, so by introducing the standby airside plant as a means of reducing fan power usage, we also see a secondary advantage by having additional heat exchanger surface which allows room heat removal at a closer temperature difference.

¹ Met Office average ambient figures for London, UK at 10/15ºC, 20 per cent ethylene glycol

² Met Office average ambient figures for London, UK at 6/12ºC on wet systems using free-cooling chillers