Not surprisingly, computer servers have different demands than people. This is also true when it comes to designing climate and cooling solutions for IT equipment and data centers; one of the major differences being that IT equipment generates no humidity so removing sensible heat is the main target.
The optimal temperature for server spaces has changed over the years with the evolution of processors and servers. For today’s equipment the optimal temperature range is in the range of 18 and 27 degrees Celsius, according to ASHRAE. But – again, unlike the demands for people – the server space will need to keep this ambient temperature 24 hours per day, 365 days per year. The collected data centers around the world account for a sizable chunk of the global energy consumption today. This means that energy saving and efficient solutions, with a maximum of availability, often are essential.
Since the server spaces generate sensible heat constantly and will require cooling all year round, a free cooling solution that takes advantage from low external temperatures during nighttime or the colder seasons might be worth looking into.
Small data centers
Consists of up to a handful of racked servers, for example in a small business. Space is often precious, and required for the servers themselves, which makes it necessary to minimize everything else – including cooling. Here a compressor-based solution is often best. It is easily scalable to fit increased or decreased demand, making it easy to optimize the system. One possible – and effective – solution is to install so-called “in-row” cooling units, which efficiently remove large amounts of sensible heat.
Medium/Large data centers
More servers in a larger, dedicated space – possibly a municipal or commercial distributed data center. Here efficiency is even more important, and savings on energy consumption is a main driver. Often the best cooling solution is a combination of a water-based chilling system with a free cooling chiller. The water-based solutions are harder to upgrade and expand, however, so the cooling system should be dimensioned for a projected maximum capacity demand from the very start. Uptime is often critical, particularly in Tier 4 certified data centers, so redundancy in both power supply and cooling systems might very well be necessary.
Enormous data centers
These gargantuan, hyper-scale data centers (think Google, Facebook or Amazon) often need highly customized solutions. They might for example be located in colder regions of the globe to reduce power consumption. Typically, though, these server parks use high capacity cooling systems specifically designed to the building layout.
To sum up, to design an efficient IT cooling system you need to analyze location, size and future expansion plans for the data center carefully to come up with a solution that maximize the removal of sensible with a minimum amount of energy usage. This can be done in a smaller data center with a compressor based solution, in large data centers a combination of water and free cooling is recommend and in the enormous centers a customized solution may be the best solution. However, it is a complex task to design an IT cooling system and many variables such as compactness, ease of maintenance, and control system must also be put into the equation for optimal efficiency.
For the Data Center Cooling industry as a whole, the real game changer, in terms of efficiency and sustainability, would be to focus more on designing systems that could reuse excess heat from the server farms in a smart way so that nearby housing and industrial processes could make use it. That would be energy efficiency.