Data centres support many critical services. We all know of their strategic importance and how they require large amounts of water, energy and land resource which continually comes under scrutiny. All economies are becoming increasingly digital and dependent on reliable infrastructures for essential and increasingly non-essential services.
Two of the most pressing concerns for data centres are the use of limited resources, water, and power, particularly harmonics can also have a detrimental effect on a system’s efficiency but are often overlooked. Harmonics (electrical pollution) sometimes underestimated, can require an increase in CAPEX and OPEX spending and attracting penalties when introduced to the grid.
Data centres are moving towards increasing internal temperatures in data centre halls in accordance with ASHRAE guidelines. Following a higher temperature strategy offers more opportunities for free cooling which can help to decrease energy consumption.
Other steps such as reducing the speed of electric motors in cooling systems can save energy and reduce emissions. Many data centre cooling systems are over specified and designed for peak loads under conditions such as the hottest day of the year which do not occur all year round. Variable speed drives (VSDs) used to address partial loads can save 20%-60% energy in cooling processes. EC fans are far more efficient than AC fans.
Some data centre providers are moving away from air-based cooling systems. They are investing in on-chip cooling technologies to meet the demands for higher rack densities, increasing efficiency and reducing whitespace m².
Using air cooling technologies with high density racks raises issues about energy consumption particularly when working with 80-100kW per rack for example.
PUE has been a long-standing metric for measuring data centres’ efficiency. For high density data centres sustainability is high on their agenda. PUE is not a holistic measure though as it does not address the overall picture of energy efficiency. PUE is measured by calculating UPS systems, air conditioning, chillers, CRAC and CRAH units, servers, and other network equipment, it does not consider what servers are idle and their duration. Data Centre Information Management (DCIM) systems can provide valuable real-time information.
Transitioning to a greener infrastructure requires solutions that go further than traditional thinking to rise to the sustainability challenge.
AI will have a notable impact. It could help data centres implement greater energy saving initiatives, through analysis and by monitoring energy usage in real time, thus identifying operational inefficiencies and improve overall performance.
Whereas it is anticipated that Generative AI will double or even triple energy consumption in data centres – all options to improve energy consumption and minimise waste need to be considered – rejected heat is already being regarded as energy that can be used to heat homes and offices.
Cooling plays a critical role in a data centre and is estimated to account for up to 40% of their energy consumption. As demand for services such as streaming and blockchain increase so does the cooling requirements.
Emerging trends such as immersion cooling and direct chip cooling are likely to increase, offering enhanced thermal management and greater efficiencies. Immersion cooling can reduce carbon emissions by up to 45% compared to more traditional cooling solutions. Direct chip cooling involves integrating the chip directly onto the heat exchanger and allowing the coolant to pass in close proximity removing the heat at its source.
The HVAC industry is undergoing a rapid transformation in the drive to deliver carbon negative cooling strategies. Customers are more environmentally aware and climate change ever more compelling. Companies that can display an authentic commitment to sustainability will hold the competitive edge.
