Data centres cannot switch off. Servers run around the clock, and nearly every watt they draw turns into heat. If the chilled water loop that cools those servers stops flowing, rack temperatures climb within minutes and thermal throttling or hardware failure follows. For operators working to tight power usage effectiveness (PUE) targets and strict service level agreements, planned shutdowns on the cooling loop are rarely an option.
Cooling is also a big slice of the energy bill. The UK government report on the impact of growth of data centres on energy consumption sets out how central thermal management has become to UK digital infrastructure. When pipework needs attention, the loop has to stay live.
So how do you cut into a pipe that cannot be drained? Three live pipeline techniques carry most of the load.
Why Full Shutdowns Are Off the Table
Shutting down a chilled water loop shifts the thermal load onto backup systems sized for emergencies, not planned works. Even a short window risks hotspots around high-density racks, SLA breaches and knock-on damage to sensitive hardware. Draining and refilling a large closed loop also means re-dosing glycol, purging air and testing water chemistry before the system is back in spec.
Live pipeline work avoids all of that. Flow keeps moving, dosing stays intact and engineers work around the active system.
The Three Core Techniques
Hot Tapping
Hot tapping, sometimes called under-pressure drilling, is how you add a new branch to a live pipe. A fitting is welded or clamped to the pipe wall, an isolation valve is bolted on, and a specialist cutter bores through the wall while the loop stays pressurised. The coupon is retained, the valve closes, and the new branch is ready to tie in.
In a data centre, this is typically the method of choice for tying in a new chiller, a CRAH unit for a fresh aisle or an additional pump set without touching the wider loop.
Line Stopping
Line stopping takes isolation a step further. After an initial hot tap, a temporary plugging head is inserted inside the pipe to block flow at a precise point. Pair two line stops with a bypass and you can isolate a whole section while the rest of the loop keeps running. When the works are complete, the heads retract and the fittings are capped off.
This is the route to take when a butterfly valve has failed in service, when a corroded section needs cutting out, or when a tie-in point sits between two valves that no longer seal.
Pipe Freezing
Pipe freezing uses liquid nitrogen inside a jacket clamped around the pipe to form a solid ice plug. That plug acts as a temporary valve, rated to the full system pressure for the duration of the works. Once the downstream job is done, the jacket is removed and the plug thaws back into flow.
Freezing is the tool of choice on smaller branches, on copper and carbon steel risers and where no upstream isolation valve exists. It suits glycol-water mixes common in data centre loops, although freeze times extend as glycol content rises and the technique needs careful planning at higher concentrations.
When to Use Each Method
The right choice depends on pipe size, material, what the job involves and how much of the loop can be taken out of service.
| Scenario | Best method | Why it works |
|---|---|---|
| Tying in a new chiller or CRAH unit | Hot tapping | Creates a live connection without draining the loop |
| Replacing a failed isolation valve on a main header | Line stopping | Blocks flow at two points so the valve can be swapped |
| Fixing a pinhole leak on a branch riser | Pipe freezing | Temporary ice plug removes the need for a drain-down |
| Cutting out a corroded section of pipe | Line stopping | Gives a clean, depressurised section to work on |
| Adding a branch for a new server aisle | Hot tapping | Connects the feed with the rest of the loop still live |
The Uptime Case
The headline benefit is obvious: the servers stay up. Underneath that sit a few others worth flagging. There is no forced reliance on backup cooling, no water chemistry reset after a drain-down, and planned works can run in daytime hours with the facility still trading. For a colocation operator, that is often the difference between a routine maintenance window and a chargeable SLA breach.
Live work does need proper planning. Pipe material, wall thickness, flow rate, temperature and glycol content all shape the method chosen. Getting a specialist on site early, ideally at the design stage of a new tie-in, avoids surprises later.
Talk to RDS Pipeline
RDS Pipeline has more than 20 years of experience in live pipeline work across the UK, including chilled water loops on mission-critical sites. From a planned tie-in to an urgent leak repair, we can plan and deliver the job with the cooling still flowing.
See the full range of pipeline services we offer, or call 01277 500510 to talk through your site.
