US Liquid Cooling editorial · Reviewed · 4 min read
Retrofitting liquid cooling into an operating data center is an exercise in working with existing constraints. The building, electrical infrastructure, air system, access routes, and operating procedures already have a history. Before selecting equipment, establish which parts of that environment can support the proposed change and which need adaptation.
Schneider Electric publishes reference designs that include high-density clusters beside traditional IT. Chatsworth’s retrofit guidance highlights physical and infrastructure constraints in existing facilities. These examples support a site-specific approach; a reference design is useful context, not proof that a particular building is ready.
Survey the actual installation
Begin with current drawings, equipment schedules, operating information, and a physical survey by qualified personnel. Record the routes available for piping, the spaces available for equipment, service clearances, and relevant structural limitations. Compare documents with what is actually installed.
Include maintenance staff in the survey. They may identify access problems or operating dependencies that are not obvious from drawings. Create a list of assumptions that still require confirmation rather than allowing uncertain details to become accepted design inputs.
Define the first deployment precisely
Describe the initial servers or racks, the planned workload, and the manufacturer’s cooling requirements. Separate that first phase from future expansion ambitions. Equipment sized for an undefined ultimate build can be difficult to evaluate against the immediate project’s cost and operating needs.
Agree on an expansion strategy without pretending that every future configuration is already known. Reserve or plan interfaces where appropriate, and record which decisions must be revisited when the next phase is specified.
Account for the remaining air load
Direct liquid cooling may remove heat from selected components while other equipment continues to depend on room air. ASHRAE’s retrofit guidance explicitly discusses retaining air systems for residual loads in hybrid arrangements. The exact split depends on the installed IT configuration.
Have the engineering team assess the combined air and liquid arrangement. Do not remove air capacity based only on the number of liquid-cooled racks. Consider how the surrounding legacy equipment operates and how airflow changes during installation and normal service.
Examine the facility interface
Determine whether a suitable facility-water connection exists and whether it can support the proposed equipment under the required conditions. Where it does not, investigate alternatives with the project designer. The choice affects where heat is delivered and which infrastructure must handle it.
Identify power, controls, alarm routing, and monitoring requirements as well as the mechanical connection. A complete retrofit plan explains how the new cooling equipment joins the existing operational system, including who responds when an alarm involves both facilities and IT.
Plan work around live operations
Divide the installation into controlled stages. Define what must be isolated, what remains in service, and what evidence is required before each transition. Establish contingency arrangements with the operations team rather than relying on informal assurances that the work will be quick.
Review delivery and replacement access. A unit that fits in its final location still needs a safe route into the building. Keep construction protection, cleanliness, and handover requirements visible in the schedule so they are not lost when installation dates become tight.
Accept the phase and document the next one
At completion, update drawings, equipment records, maintenance instructions, and capacity assumptions. Record commissioning results at identifiable conditions. Give the operating team a clear view of the new system boundaries and unresolved expansion dependencies.
A successful first phase creates an evidence base for later decisions. It does not automatically validate every future rack density or server generation. Reassess the thermal path, electrical capacity, physical arrangement, and operating procedures when the next deployment changes the assumptions that supported the original retrofit.
Sources & further reading
- Schneider Electric reference design with retrofit scenarios ↗
- Chatsworth existing data center liquid cooling guidance ↗
- ASHRAE retrofit and modernization strategies ↗
Explore the company directory → · All liquid-cooling guides →