DCN August 2016 | Page 17

cooling revisit the data centre cooling process. Twenty years ago that was what the cooling infrastructure did: it cooled the IT space. In general DX CRAC units were used, continuously running its compressors to push cold air of about 12°C into the raised floor, ensuring the temperature in the room at no point exceeded 22°C. Whenever a chilled water system was used, the chillers would provide 6°C chilled water all year round. The cooling process Today, the term ‘cooling’ does not give the process the credit it deserves. We are actually talking about heat removal. One could argue that the only cooling process required in the data centre occurs within the IT equipment itself, where the hot processing cores are kept at their correct operating temperatures by heat sinks supported by fans. Normally, by using air, the excess heat is then rejected into the data centre white space to be removed. Within this realm a system needs to be in place that removes this heat and provides the space in front of the servers with air of the correct conditions. Until around 2005 the requirements for data centre temperatures changed focus from the room temperature to the server inlet temperature. This opened up great opportunities for smart designers and cooling vendors to think up new and clever ways to remove the IT heat from the white space. Suddenly, there was not only the choice between DX or CHW. We had kyoto cooling, (other) indirect fresh air cooling, direct fresh air cooling, storage of cold in underground aquifers. And to the surprise of many, chilled water systems could be designed to great efficiencies too, now that an air temperature set point of 25°C was sufficient. Today, many heat removal systems and technologies are available that can all achieve partial PUE values of 1.1-1.2, at least in most of the temperate world. A nominal data centre design should see the energy consumption of the ‘cooling system’ equal to that of the electrical system. But the call for social responsibility actions does not end here, and data centre stakeholders and the industry are constantly working on producing even better designs and facilities. Recently, data centre waste heat reuse has become a hot topic and the author has personally been involved in several initiatives to this end. Is there a need to focus on PUE? Interestingly, the current focus on PUE is not good news for heat reuse. Harvesting the energy inevitably costs additional energy, which increases the PUE which is not necessarily bad news. Nearly every bit of electrical energy consumed by the IT equipment is transferred into thermal energy. When this thermal energy is utilised in a useful way, the ‘cooling’ system is finally serving society in a direct way. Who then really cares about the PUE when you get to spend your kWh twice? What are the opportunities for data centre heat reuse? So why is it not mainstream yet? Typical issues encountered are concerned with the impact on the infrastructure (availability and reliability), the heat collection effectiveness and the value of the heat (the business case). To start with the latter, although sustainability ambitions play an important role, the venture still needs a solid business case. Compared to your current state-of-the-art data centre, you need additional investments, for instance for heat exchangers, water buffers, storage systems, etc. There must be a return on these investments. There are several ways that heat can be reused. In a paper from 2012, Zimmermann et al investigates the economic value of recovered heat. Next to the obvious use for space heating, there are also other possible The cooling infrastructure is a vital element in any data centre, providing essential support to the business applications running in them. 17