From a sustainability perspective, data centre (DC) growth comes with several challenges and a
few opportunities. Thise report brings these dimensions together, analysing DC sustainability from a circular economy perspective, revealing existing synergies and trade-offs.
For energy consumption, the widely used PUE metric has several fundamental limitations: Being
a relative metric, it does not inform on the overall energy consumption of the DC. It considers the two main types of DC infrastructures (cooling and power provisioning) jointly, mixing their individual efficiencies, and is thus too coarse. It is also skewed: Due to data limitations, it attributes both consumption of server fans and transformation losses of the power supply units (PSUs) to the IT energy instead of the non-IT energy.
The PUE does not measure compute efficiency but infrastructure efficiency. The latter has already
reached very good values and progress will only be incremental. Compute energy, however, is exploding, so a metric reflecting compute efficiency across heterogeneous loads, while challenging, is needed.
To cope with the increased DC power densities and minimise losses, power provisioning is shift-
ing: from alternating current to direct current and towards higher voltages. This is the perfect opportunity to define a new metric for the power provisioning efficiency, which should be both comprehensive (including PSU losses) and disjunct from the cooling overhead. A new set of energy metrics for DCs could thus comprise i) compute efficiency, ii) transformation efficiency, and iii) cooling overhead.
On-site water consumption depends on the cooling technologies deployed. Waterside economisers and water-cooled chillers (via cooling towers) have a high water consumption, while dry coolers, air-cooled chillers, and airside economisers do not. However, these last technologies are often adiabatically
supported, yielding them at least temporarily water consuming as well. Upstream water consumption is mainly due to electricity production, and its main source is the behind-the-dam evaporation for hydro power plants.
Two trade-offs emerge between energy and water used in cooling as well as between on-site and
upstream water consumption. The two are related: Using a water-consuming technique generally
lowers the energy required in cooling, which also lowers the water consumed upstream in power generation (if any). For very ‘wet’ electricity, there is little competition, and it is generally worth spending some more on-site energy to save both electricity and the related upstream water. For ‘dry’ electricity, however, there is a trade-off between the two.
Energy circularity can be achieved through waste heat recovery. There is a trade-off between high
heat reuse (close to populated areas) and low energy consumption (far North). And the waste heat has limited uses and is not the same energy quality, a fact not reflected by current metrics. A more relevant metric would consider the avoided energy through heat recovery instead of the amount recovered.
The production impact of microelectronics is poorly understood. From the three production phases (mining, refining, and fabrication), the mineral refining / purification is the least understood in terms of sustainability. Prolonging lifespans is always beneficial.
Material circularity can be achieved by interpreting general circularity principles such as the 9R framework in the context of DCs. Possible circularity-enhancing measures exist for both the DC itself and, more importantly, for its microelectronics. The measures can be categorised into product design, process design and business models, choice of materials, and operating conditions. Together, they have effects across all circularity levels.
The relation between DCs and the power grid is complex and multifaceted. Modern DCs possess
three crucial and novel features that present new challenges for the grid. Various measures are used to mitigate these challenges, the most important of which are workload shifting, battery storage and on-site generation. These measures have, in turn, further consequences, which are both beneficial and detrimental. They can bring offer grid flexibility as well as innovations in the field of energy. But they also bring noise, pollution, and GHGs, and compete with the energy sector for resources.