The characterization of water resources as scarce essential goods has resulted in refrigeration by evaporative cooling, according to different systems and technologies, and the consequent use of water in the sector data center, is no longer so well seen, and although this seems in itself something positive, sometimes it is not justified and there are numerous cases in which the non-use of local water supposes more global water consumption according to the energy destined to DC.

For many years, famous metrics such as PUE have prevailed in measuring the degree of energy efficiency in data center, leaving in the background ratios such as WUE, or Water Usage of Effectiveness, which quantifies the cost of water for each kWH IT cooled with said volume of water.

The relevance and impact on the sustainability of the use of water in the data center, leads us to delve a little deeper into the validity of the aforementioned WUE metric, and if it really makes sense to subordinate it to the self-defined border by the infrastructure of the data center to the case. That is to say, it forces to evaluate and include, or not, the indirect consumption of water that must be given by the energy provided to the data center.

In equivalent terms, ignoring the existence of other deferred water consumption that can be passed on to the DC consumption balance is like claiming that a highly competitive PUE (let's say less than 1.2) can characterize a DC as optimally sustainable, as long as the carbon embedded (or incorporated) in the electrical generation destined for the center is high: let us say, for example, that the electrical generation came from a coal-fired power plant; Then our low PUE would be of little use to us in strict terms of the impact on the carbon footprint.

Likewise, our WUE could be very low, or as so often, null, because there is no evaporative cooling, but if the electrical energy that feeds the DC has a high consumption of hydric resources, it will happen that our efforts to reduce the total consumption of water can cause paradoxes such that the total water consumption (between on-site and off-site) is greater in those cases in which on-site consumption is zero.

This notion of the global balance of water resources destined for a data center, led Wemfoff (Wemhoff and Chen, 2018) to redefine WUE as WSUE, that is: Water Scarcity Ussage Effectiveness, or efficiency in the use of water scarcity.

There may be cases in which using water for cooling (from a recycled water plant, WWTP, or recovered rainwater, for example) involves less overall water consumption than in those cases apparently free of water consumption where, however, the overall kWh efficiency is lower:

• Because in cases of zero water consumption, the PUE is higher (air-cooled chillers).
• Origin of company energy from an electrical mix where, depending on fossil origin or not, the water used in the process (combined cycle, thermal power plant, nuclear) is high.
• The m3/h intended for in-situ adiabatic cooling has a higher yield than the m3/h intended for electricity production, where the loss due to transport must be added.

Therefore, it is important to assess the origin of the energy destined to feed the DC: by system and in an unthinking way, discarding the use of water for adiabatic cooling can cause consumption to be higher in global terms. If the origin of the energy that feeds the DC is of renewable origin, without a doubt contemplating evaporative cooling practices is something pertinent and appropriate.

As defined by the WSUE, the scarcity of water in the location of the CD and the local origin of the production energy mix, means assessing the use of water at the local level for adiabatic cooling: because what is clear is that each local m3 of water that used to cool a DC can have two direct consequences:

• That the kWH of electrical origin with a carbonized production source be lowered: that is, in turn, the use of water decarbonizes in terms of total kWH from production with a range of energy of fossil origin.
• That the global consumption of water (local and production) is less than if the consumption of water is limited to only production.

© Nexitic 2022