Data centre cooling towers: water treatment without risking uptime

How data centres can stay on top of cooling tower water treatment, cleaning and Legionella control without putting uptime at risk.

Call 020 3633 6064

Data centres have a slightly awkward relationship with cooling towers. The towers are essential to keeping the halls cool, so nobody wants to touch them. But leaving them alone is exactly what you can't do. ACOP L8 and HSG274 Part 1 apply to a data centre's towers in just the same way as an office block's, and the cleaning, monitoring and treatment still have to happen.

The good news is that keeping on top of water treatment and keeping the lights on aren't in conflict. It just takes more planning than most sites. Here's how to approach it, with summer heat loads just around the corner.

Why data centres are a bit different

  • Continuous load. Towers often run around the clock, all year. There's no natural winter shutdown to fit a clean into.
  • Change control. Any work on cooling plant usually needs a method statement, a risk assessment, a change request and an agreed window.
  • Access and security. Contractors need inductions, escorts and pre-booked access. A "quick visit" isn't really a thing.
  • Equipment protection. Scale and corrosion don't just waste energy, they threaten heat exchangers and condensers that the whole facility relies on.
  • Reputational risk. A Legionella problem at a well-known facility is a story nobody wants to be in.

None of these are reasons to do less. They're reasons to plan earlier.

Use the redundancy you've already got

Most data centres are designed with spare cooling capacity, whether that's N+1 towers, multi-cell towers or standby chillers. That redundancy is your friend when it comes to cleaning. Typical approaches:

  • Clean one cell at a time while the others carry the load.
  • Take one tower out of service while the standby runs, then swap.
  • Time cleans for spring and autumn, when outside temperatures give you more headroom.

Bear in mind that a standby tower sitting full of water and never running is effectively a stagnant system. Rotating duty and standby, or circulating the standby regularly with treatment, keeps it ready and under control. We wrote more about that in bringing a tower back after a shutdown.

Make the dosing as reliable as the power

Data centres put a lot of thought into backup power. Dosing deserves some of the same thinking, because a quiet dosing failure is one of the most common routes to a high count. Things worth considering:

  • Alarms for low drum level, no flow and high conductivity, linked into the BMS so someone actually sees them
  • Spare chemical stock on site, stored safely in line with the safety data sheets
  • Regular calibration of conductivity probes and controllers
  • Clear labelling of drums and lines, so a drum change can't connect the wrong product

Our post on spotting dosing failures early goes through the common causes in more detail.

Get the monitoring routine right

Monitoring on a data centre tower is the same in principle as anywhere else: dip slides or TVCs at least weekly, chemistry checks as your scheme sets out, and Legionella sampling at least quarterly to a UKAS-accredited lab. The difference is making it fit the site's way of working:

  • Agree standing access for routine monitoring visits so they don't need a fresh change request each time.
  • Put sample points where they can be reached safely without entering restricted areas unnecessarily.
  • Make sure results reach the facilities team quickly, with clear action levels, so a high result isn't sitting in someone's inbox.
A typical scenario

Imagine a facility with three towers in an N+1 arrangement. Cleans keep slipping because nobody wants to raise the change request in summer, and by autumn the certificates are overdue. A better pattern is to fix the cleaning dates a year ahead, one tower at a time in spring and autumn, with the change requests drafted well in advance. The work becomes routine instead of a recurring argument.

Plan for a positive result before you get one

On a site where you can't easily shut the cooling down, the time to decide what happens after a high Legionella result is now, not when the certificate arrives. Your written scheme should cover:

  • Who gets told, including out of hours
  • How an online disinfection would be done without disrupting cooling
  • How a tower could be isolated if the risk called for it, and what that means for capacity
  • Pre-approved method statements, so emergency work doesn't wait on paperwork

Our first 48 hours guide explains the response in general terms.

What about adiabatic and hybrid coolers?

Plenty of newer data centres use adiabatic or hybrid coolers instead of, or alongside, open cooling towers. They use far less water, but where a unit sprays or wets surfaces it can still create aerosol. HSG274 Part 1 expects the Legionella risk from any unit that can produce spray to be assessed, so don't assume they're automatically out of scope. Whether they need notifying to the local authority depends on the design, and it's worth getting a clear answer for your units.

The short version

Use your redundancy to clean on schedule, make dosing as resilient as everything else on site, agree access for routine monitoring and plan your response to a high result in advance. Our water treatment and cleaning and disinfection services are built around sites that can't switch off. If you'd like to talk through how that could work on yours, we're happy to.

Running cooling towers on a critical site?

Tell us about your plant and how much redundancy you've got. We'll suggest a treatment and cleaning plan that works around your uptime.

Call 020 3633 6064

Need a treatment contractor who explains things properly?

Tell us about your system and we'll come back with a clear, itemised quote. No hard sell, just a straight answer on what your site needs.

Call 020 3633 6064