Data centers play a key role in our modern digital society, so it’s no surprise that people are paying more attention to how data centers consume vital resources such as water and energy. Data center operators like Equinix are dedicated to using these resources responsibly. In order to do that, we must be able to measure and report that use accurately. This is where water usage effectiveness (WUE) can help, especially if there’s better understanding of the context and limitations of its use.
WUE is a metric used to quantify how efficiently a data center uses water. It compares the amount of water used for cooling IT equipment with the amount of energy used to power IT equipment.
In a previous blog post, my colleague Alex Setmajer described how the choice of cooling technology during the initial design phase determines a data center’s water efficiency over its entire lifecycle. Unlike power usage effectiveness (PUE), a comparable metric for energy efficiency, there are few opportunities to improve WUE after a facility is operational.
Evaporative cooling consumes less energy, but significantly more water. Air cooling consumes no water, but significantly more energy. Therefore, data center operators must pick a cooling technology that is symbiotic to the local environmental conditions. For instance, they might use air cooling in water-stressed areas to limit water consumption and use evaporative cooling in other areas to capitalize on the energy-efficiency benefits, particularly where the carbon intensity of the power grids is high.
How is WUE measured?
WUE is measured in cubic meters of water per megawatt hour of energy (m3/MWh). A facility that uses no water for cooling—one that uses air cooling exclusively—would typically report a WUE of 0. In contrast, one that uses evaporative cooling exclusively could report a WUE as high as 2.5, depending on the type of evaporative cooling equipment, local climate and the choice of server cooling technology.
In reality, many data centers use a combination of different cooling methods. For example, they may switch to air cooling during the winter months, when the air is naturally colder and better suited for cooling IT equipment. The WUE for these hybrid-cooled facilities will fall somewhere in the range between 0 and 2.5. Logically, the exact figure will depend on the proportion of evaporative cooling the facility performs over the course of the year.
Data center operators around the world are looking to optimize water efficiency, primarily because we recognize and respect that water is a valuable resource, but increasingly because of more stringent regulatory targets and reporting requirements. That’s why it’s very important to measure and report WUE regularly and consistently. It’s equally important for all parties involved to have an accurate understanding of what WUE really means and how this impacts other data center performance metrics.
How do WUE and PUE metrics relate to each other?
To properly account for the tradeoff between water and energy use, WUE should be measured and reported alongside PUE. The chart below indicates how different cooling choices at the building level and server level can impact both WUE and PUE.
WUE alone doesn’t tell the full story. For instance, using air cooling instead of evaporative cooling will decrease WUE, but it will also increase PUE. The ability to simultaneously minimize both WUE and PUE is limited and very dependent on the local climate and environmental conditions.
Regulations that set aggressive WUE targets without linking those targets to PUE and local environmental conditions may have unintended effects. Operators may avoid evaporative cooling altogether and revert to air-cooled solutions, driving up energy consumption. Similarly, if PUE targets are too aggressive, data center operators may have no alternative but to increase water use in order to meet those targets, an issue that’s particularly important in water-stressed regions.
At Equinix, we recognize the need for such targets. In some cases, they’ve even inspired us to find innovative efficiency solutions that help decrease PUE without driving an equivalent increase in WUE. However, there’s a limit to how much energy a data center can save without using evaporative cooling. Therefore, it’s best for regulators to consider the link between WUE and PUE when they set targets.
WUE typically overlooks the embedded water usage in power
In some cases, the compromise of increasing power consumption to minimize a data center’s on-site water use may indirectly increase withdrawal from a water basin. This is because water use is embedded into the power generation process, and different power sources require significantly more water than others. In addition, power companies may not have the same commitment to responsible water use that the data center operators have themselves.
Imagine that a data center operator is planning a new facility. They estimate that air-cooling the facility will require 5 MW of power, while evaporative cooling will require only 4 MW. The operator must decide whether saving 1 MW of power is worth the increased on-site water use. Conversely, they should also account for the water needed to generate that extra 1 MW of power.
Since the data center operator doesn’t use this water directly, it won’t show up in typical WUE reporting. This illustrates one key shortcoming of the metric: Ideally, it would account for both direct and indirect water use in data centers. Directly using more water inside the data center may sometimes be the more responsible choice, even though it will drive higher WUE.
WUE doesn’t differentiate by water source
Another limitation of WUE is that it treats every liter of water equally, despite the fact that data centers pull water from many different sources. Potable water taken from domestic supplies requires much more processing than reclaimed greywater or water taken directly from large lakes or rivers. Demand for that potable water is also much higher.
If two data centers are using the same amount of water, but one of them is using significantly more potable water, then that one will have a much higher impact on local infrastructure and other water users in the community. However, WUE alone wouldn’t be able to communicate that.
Not all operators report WUE the same way
Although I mentioned previously that WUE is based on water used for cooling, there are multiple established ways to report WUE set out in the ISO/IEC 30134-9 WUE standard[1], ranging from basic to advanced.
Most operators—including Equinix—align to the basic WUE1 method. This uses meter readings from a single incoming water meter to represent the water withdrawal of the entire facility. Therefore, it’s significantly easier to measure than other categories of WUE.
Installing additional submeters throughout the site would provide more granular data and enable operators to report more advanced categories of WUE. These categories help operators account for the proportion of on-site water withdrawal that isn’t used primarily for cooling, thus improving their WUE. However, since the WUE improvement achieved with submetering would likely be no more than a few percentage points, many operators have decided it’s not worth the additional complexity. Valuable water circularity initiatives, such as wastewater recycling to third parties, are also not credited and incentivized through this approach.
WUE3, the most advanced WUE reporting approach, does account for embedded water in power consumption. However, reporting it would require reliable data for embedded water in power. Since very few utility providers reliably report this information, the opportunity to report WUE3 is limited.
The potential for WUE reporting variance arises when a single company-wide WUE figure is reported without context. Specifically, there’s no way of knowing whether that figure includes all the company’s data centers, including the air-cooled facilities, or just the ones that use evaporative cooling.
Equinix provides valuable context around water use
At Equinix, we’re dedicated to increasing transparency around our water use. This is why we report two separate WUE metrics: one across our entire data center portfolio and one exclusively for data centers that use evaporative cooling. During 2023, these figures were 1.07 and 1.63, respectively.
Going forward, we’ll continue to report both metrics annually, so that our stakeholders can compare our performance against our peers that use either of these reporting methodologies. We also provide transparency on other valuable water metrics such as the amount of water we withdraw, consume and discharge and our proportion of potable water consumption.
We’re also helping to drive greater consistency in water reporting best practices in the data center industry. Equinix employees are leading a refresh of the ISO/IEC 30134-9 WUE standard.
To learn more about the Equinix approach to responsible resource consumption, including both water and energy consumption, access our interactive sustainability report today.
[1] ISO/IEC 30134-9:2022(en), Information technology — Data centres key performance indicators — Part 9: Water usage effectiveness (WUE)
Subscribe to the Equinix Blog