Set a clear measurement boundary
Separate IT electricity from cooling and other facility loads before comparing performance. Power usage effectiveness can support facility-level tracking, but changes in IT load affect the ratio, and it does not describe every aspect of cooling or resilience. Combine whole-site indicators with equipment-level measurements and operating records. Compare consistent boundaries and periods so a fall in computing demand is not mistaken for a cooling improvement.
Fix airflow before increasing cooling capacity
Investigate rack inlet temperatures, hot-air recirculation, bypass airflow and local hot spots. Containment, blanking panels and cable-opening management may improve air delivery when appropriately designed. Return-air or room-average temperatures alone can conceal rack-level issues. Validate airflow changes against actual equipment requirements and fire-protection arrangements. A local thermal problem may require better distribution rather than colder supply air or additional cooling units.
Review chilled water and part-load performance
Evaluate chillers, pumps and heat rejection together across expected IT loads and outdoor conditions. Check temperature differences, control stability and plant sequencing. Redundant equipment operated unnecessarily can increase energy demand, but standby arrangements must be assessed against the agreed resilience design. Economiser or free-cooling operation depends on climate, system configuration and water or air-quality constraints; it is an option to evaluate rather than a universal solution.
Assess liquid cooling around the actual workload
Higher-density deployments may justify liquid cooling, but the decision requires consideration of server compatibility, distribution equipment, water temperatures, leak management and maintenance capability. Compare the complete cooling chain, including pumps and heat rejection, with the existing arrangement. Plan the interface between IT and facilities teams. Retrofit complexity and remaining asset life can materially change the business case even when the thermal approach is technically suitable.
Implement with operational safeguards
Use a documented change process with thermal limits, alarm review, rollback arrangements and competent operators. Stage adjustments and observe representative load conditions before extending them. Verify cooling energy alongside inlet temperatures, fault events and resilience requirements. Savings forecasts should make load growth, tariffs and redundancy assumptions explicit, and post-change measurement should distinguish efficiency gains from changes in IT demand.
Reference & further reading
Official technical resource. Confirm applicability to your facility and engagement.
Related insights
What a commercial energy audit should deliver