When talking about data centre cooling, to simply visualise it – a data centre is a large heat source.
In turn, it is of critical importance that the technological equipment of a data centre (primarily servers), especially in high capacity conditions, has a defined optimal operating temperature. And so, a significant cooling capacity must be provided, to be optimally used for maintaining the desired microclimate.
The appropriate selection of a cooling solution is very important, because it creates the largest portion of a data centre’s overall electrical consumption. In turn, electricity costs are the most significant component of resource costs and largely affect the overall cost-effectiveness of operating a data centre.
One of the industry’s significant inventions in the field of cooling is the so-called concept of corridors. This concept was invented several decades ago and is still relevant today. The operating basis of corridors is the principle that warm and cold are separated into zones, and mechanical constructions are used to achieve this. In this way, a larger efficiency for cooling a data centre is achieved.
When building a data centre at certain latitudes (air temperature, seasonality, humidity), technology must be chosen that is optimal for the specific climate conditions and provides the highest efficiency, i.e., least energy consumption for providing the optimal climate. An essential factor that must be kept in mind when choosing a cooling system is the dynamic of load increases.
Everything begins with an idea and business model for which the data centre is being built. Cooling systems are also adjusted according to these.
Usually, data centre cooling most often uses traditional and proven systems, including chillers or dry coolers in combination with high precision air conditioners with the free-cooling option.
A comparison to keep in mind when assessing the efficiency of a cooling solution is the PUE coefficient. It is an efficiency factor that indicates the proportion between the overall volume of consumed energy in relation to the volume of efficiently used energy during a whole year. The benchmark indicator in the industry is 1.3-1.4.
If the chosen solution provides this indicator, it is recognised as efficient. Of course, this is an average indicator, though it does allow to initially conclude whether the chosen solution is appropriate for the specific case.
Most often, known and proven technologies are chosen for data centre cooling. However, as in all aspects, time does not stand still, and innovations come into practice that help provide the best possible solution. Of relevance are the new free cooling technologies where air is let into the data centre from the outside. By itself, this novelty does not seem very interesting, but it must remembered that a specific microclimate is ensured within a data centre. Even cleaning work happens with technologies (vacuum cleaning equipment) that help avoid server damage. By using exterior air for cooling, appropriate filters must be provided to maintain the desired composition of air within the facilities.
If a data centre is located near a water reservoir (river, lake, etc.), it is possible to assess the use of this resource for cooling the data centre.
Another method that is more frequently used in geographical latitudes where the summer season lasts from four to six months (Spain, Italy) is the so-called evaporative/adiabatic cooling equipment. The basis for this system is installing micro orifice panels before cooling equipment collectors that as a result of the evaporative process allows lower temperature exterior air to flow into cooling equipment. In turn, this allows to provide equipment with air cooling to the desired temperature with notably less electrical consumption.
In any case, when preparing a project for a data centre cooling system, all technological solutions should be analysed in accordance with the location of the facility and power capacity dynamics in order to create a solution that is optimal for the specific situation, taking into account short-term and long-term goals.