G-XR8P2XJ088

Data Centre Water: Ask What Happens in a Dry Week

Imagine two data centre operators report the same annual water use per unit of computing equipment electricity. One draws most heavily on its local supply during a few hot, dry weeks. The other spreads its demand more evenly and can switch cooling modes during a shortage. Their annual figures could look alike. For the people responsible for keeping water available in either place, those two patterns tell different stories.

That is why it helps to ask two questions whenever an infrastructure supplier presents a water efficiency figure: what exactly went into the calculation, and when did the demand fall on the supply? The first question makes two numbers more comparable. The second starts to tell you what those numbers mean for a particular place. An annual average alone cannot tell you how the site will operate when water is least available.

Read the numerator before comparing the number

The 2024 United States Data Center Energy Usage Report from Lawrence Berkeley National Laboratory describes site water usage effectiveness, or site WUE, in litres of on-site water consumption per kilowatt-hour of electricity used by IT equipment. It distinguishes that on-site consumption, much of which relates to cooling, from water consumed in generating the electricity used by a facility. This is a study of the United States and includes modelled estimates. Its definitions can help readers ask questions elsewhere; its national estimates cannot tell you what happens at a specific site.

If the supplier gives you one number, ask for the volume and the electricity figure behind it, for the same reporting period and the same boundary. A site can improve the ratio and still use more water in total if it grows. In a hypothetical case, a site drops from one litre per kilowatt-hour to half a litre, but its IT electricity use triples. Under those assumptions, its reported water volume still rises by half. Each number can be correct; they answer different questions.

Separate water taken from water consumed

The US Geological Survey’s explanation of water-use concepts distinguishes water withdrawn from the portion consumed, including water evaporated or otherwise removed from the immediate water environment. Water that returns to the local environment is counted separately as return flow. Withdrawal and consumption describe different parts of the same process; treating them as interchangeable can make a comparison meaningless.

Suppose a project draws water from a public supply and returns some to a sewer. Ask how much comes in, what portion is consumed and where the remainder goes. A return somewhere downstream is not a promise that the original source will have more water available that afternoon. Nor does the use of reclaimed water establish on its own how much fresh water the site will need in an emergency. The useful evidence is a description of the actual routes, backed by quantities and assumptions.

Ask for the hardest week, not just the average year

Cooling design and climate are among the factors in the Berkeley Lab report’s modelling. For a project decision, ask for the month with the highest demand and the assumptions behind a hot, dry week. A new site may only have a design estimate; a running site may have meter readings. Both can be useful, but label them clearly and ask how a forecast will be checked once the site is operating.

Consider a fictional project that can switch from water-based cooling to a mode that draws less local water but uses more electricity. That could help in a local shortage. It raises a separate question about the additional electricity demand and where that electricity comes from. An honest comparison states the operating conditions of each mode, treats local water and power as different constraints, and explains who decides when to switch.

  • Ask for monthly site withdrawal and consumption where available, along with a peak-day or peak-week estimate. Check which water sources are included and whether the figures are measured or forecast.
  • Ask how the cooling plan changes during restrictions or unusually hot periods. Look for an action, the condition that triggers it and who is responsible for carrying it out.
  • Ask which quantities will be reported after the site opens. A plan for comparing actual readings with forecasts is more useful than a promise to publish an annual ratio alone.

Match the question to your decision

An independent business buying cloud services may have little influence over a facility’s design. You can still ask whether the figures refer to a named site, a region or an entire fleet. If a supplier cannot give site-level information, record what it can substantiate and leave the local question open. Compare like with like when information is available. A precise number covering the wrong geography is less useful than a clear account of what is known.

Water is also only one part of infrastructure stewardship. Our article on deciding which retired servers deserve a second life takes the same approach to a different stage: follow the materials and get enough information to understand what happens beyond the headline. For water, that means following the source, the use and the return route. A credible annual figure is a good start. A plan for the dry week tells you what the operator intends to do when the average is least reassuring.

Sources

One thought on “Data Centre Water: Ask What Happens in a Dry Week”

Leave a Reply

Discover more from Marychuks.com AI, Psychology, Business & CreativeVerse

Subscribe now to keep reading and get access to the full archive.

Continue reading