Designing Data Center Leak Detection by Zone
Key Takeaways
- Zone boundaries should limit coolant release and cooling loss without creating unnecessary operating complexity.
- Each zone should be independently detectable and isolatable at a practical piping boundary.
- Smaller zones can reduce potential discharge but add valves, controls, testing, pressure loss, and maintenance.
- Each zone should document the cooling capacity, racks, and workloads affected by isolation.
In data center liquid cooling systems, leaks vary widely in severity. A slow drip beneath one rack presents a different risk from a pressurized release in a distribution main. Shutting down too much of the system can disrupt more capacity than the leak threatens. Zone size should reflect how much coolant can escape, which equipment loses cooling after isolation, and how reliably operators can test and restore the affected area.
Zone boundaries need to limit leak spread and lost compute capacity without making the cooling system difficult to install, maintain, or operate. The added valves, controls, and operating work are a modest investment compared with the potentially catastrophic cost of downtime and damaged capital equipment. They contain the release, preserve cooling to unaffected equipment, and keep a local fault from becoming a broader outage.
Why Is One Facility-Wide Leak Strategy Insufficient?
One facility-wide strategy is insufficient because leak size, exposed equipment, personnel risk, and lost cooling capacity vary across the system. Without predefined zones and response thresholds, operators must assess the leak, decide which systems to isolate, and protect unrelated cooling loads during the incident.
Zoning lets teams set severity thresholds and shutdown actions during design, along with detection, containment, and isolation. Without these zone-specific thresholds, data centers expose themselves to significant financial risk. Uptime Institute’s 2026 outage analysis found that 57% of respondents said their most recent major outage cost more than $100,000, while one in five placed the cost of their most recent impactful outage above $1 million. These figures cover outages broadly, not coolant leaks specifically, but show the financial exposure when a local fault disrupts a larger portion of the facility.
To reduce that risk, teams need to estimate how much coolant could escape, how much cooling would be lost, and how the zone would return safely to service.
What Defines a Useful Leak-Protection Zone?
A useful leak-protection zone is the smallest practical section of the cooling system that operators can detect, identify, and isolate without causing a broader cooling failure or disruptingd be defined by the amount of coolant it can release and the equipment and cooling capacity affected when it is isolated. Teams should consider the highest expected flow rate, coolant trapped between valves, where a leak could travel, what equipment it could reach, how it would be detected, and how much cooling would be lost. Expected discharge includes both the coolant already contained between isolation points and the flow that continues while the system detects the leak and closes the valves. Equipment in the same room may require separate zones if it is connected to different piping branches.
How Can Local Isolation Protect Uptime?
Local isolation can protect uptime by shutting off coolant only to the leaking zone while the rest of the liquid cooling system remains in service. This works only if backup cooling is available, workloads can be moved or shut down safely, and the affected equipment can tolerate the interruption.
Closing an isolation valve stops additional coolant flow to the leak, but it also removes cooling from the equipment served by that branch and may redirect flow elsewhere in the system. Engineers must verify that affected workloads can migrate or shut down safely and that adjacent branches can absorb the resulting hydraulic changes. Even a branch containing little coolant may support enough compute capacity to make isolation operationally significant.
Where Should Sensors and Valves Be Placed?
Place leak sensors where leaking coolant is likely to collect and auto-shutoff valves where they can stop flow to a defined zone. Floor slope, containment, raised-floor openings, overhead piping, equipment pans, and cable penetrations all affect where coolant will travel. Pay particular attention to connections, quick disconnects, manifolds, CDUs, serviceable fittings, and transitions because leaks are more likely where components join or require service.
ASHRAE guidance for water-cooled servers covers pressure testing, couplings, flexible hoses, filtration, material compatibility, and systems with or without CDUs. These are the same design details teams need to review when locating sensors and isolation valves. Valves should keep the trapped coolant volume low without creating excessive pressure drop or making the system difficult to service.
What Tradeoffs Come With Smaller Zones?
While smaller zones reduce the amount of coolant that can escape and limit the racks or cooling branches affected by isolation, they also require more valves, sensors, wiring, control logic, testing, and maintenance. Every added device creates another component that must work during an incident.
Add a zone when it clearly reduces the expected release, limits the cooling capacity taken offline, or makes the leak easier to locate. The design review should confirm that those benefits justify the additional valves, controls, testing, and maintenance.
How Does OCP Guidance Support Zoned Protection?
OCP guidance supports zoned protection by treating leak detection and response as one system. It covers cold plates, manifolds, hoses, couplings, CDUs, and associated piping, along with direct and indirect detection, manual and automatic intervention, shutoff valves, de-energization, pressure effects, and nuisance alarms. The guidance does not prescribe one sensor or response method; it directs teams to assess how the complete cooling system will detect and contain a leak.
In practice, each detection zone must match a section of piping that operators can isolate and return to service. The design must connect the sensor location, coolant volume, valve response, cooling impact, operator actions, and recovery procedure.
How Should Teams Specify and Evaluate Leak-Protection Zones?
Teams should specify and evaluate each zone using clear design and operating requirements. The specification should show the zone boundaries, sensor coverage, isolation valves, the position each valve takes if power or control is lost, alarm routing, response time, pressure effects, manual override, proof-testing procedure, and restoration sequence. It should also define how teams will calculate trapped coolant volume, maximum credible release, expected discharge before isolation, cooling capacity removed, racks affected, thermal ride-through, and testing frequency.
Design review should compare several possible zone boundaries. For each option, estimate how much coolant could escape, how long flow would continue before isolation, which racks would lose cooling, and what operators would need to do to test and restore the zone. Select the boundary that provides a meaningful reduction in risk and remains practical to operate.
The plan must also work during routine operation. Operators need to test the zones regularly and see their status clearly in the BMS. If the design is too complex to test or understand, simplify it and keep the isolation points, detection, and response procedures that provide the most protection.
What Leak-Zoning Decision Should Teams Make Early?
Teams should set leak-zone boundaries before finalizing the piping, controls, and compute dependencies. Before approving the design, document what each valve isolates, which cooling and compute loads are affected, how unaffected equipment will remain in service, and how the zone will be restored.
For every zone, answer four questions:
- How much coolant can escape?
- What cooling capacity is lost?
- Which racks and workloads are affected?
- How will operators test, override, and restore the zone?
Conclusion
Effective leak protection depends on knowing how far coolant can spread and which equipment loses cooling when a valve closes. Well-designed zones contain the leak and keep unaffected equipment operating. Before approving the design, simulate a release in each zone and confirm the expected discharge, cooling impact, control response, and recovery steps.
About the Author
Keith Zander
Senior Product Manager, Commercial Applications & Solutions - Leak Defense®, a Watts Brand
Keith Zander has decades of experience in the application of elegant solutions to complex problems in the energy and sustainability market. His experience ranges from renewable energy, data center climate control, emergency power systems, HVAC for light aircraft, heavy truck & offroad equipment. In his current role, he enjoys working with all the influencers in developing and applying cost-effective solutions for protecting all types of structures against the financial perils of catastrophic water damage.