Power usage effectiveness is the ratio a data centre is judged by: total facility energy divided by IT energy. We measure it accurately, find where the overhead sits, and reduce it through engineering.
8760h
per year of overhead
1.0
the theoretical minimum pue
Seasonal
verification at the meter
What is PUE Optimization?
PUE divides everything a facility consumes by what the IT equipment consumes. The remainder is overhead, mostly cooling. A facility at 2.0 spends a full watt on infrastructure for every watt of computing, and the difference between that and a well-run facility is an operating cost paid every hour of the year.
Reducing it is detailed engineering work: raising supply temperatures to the limit of what the equipment tolerates, extending the free-cooling hours the climate offers, closing containment bypasses, correcting fan and pump operation, staging plant so it runs at its efficient points. Each measure is small on its own. Together they are the number the board sees.
Measurement comes first, because the figure is only useful if it is accurate. Metering boundaries, seasonal variation and load normalisation are settled before any improvement is claimed.
the process
How PUE Optimization works
step 01
Measurement
Metering boundaries established and the baseline PUE measured across representative conditions.
step 02
Decomposition
The overhead broken into its parts, cooling, distribution losses and ancillaries, each quantified.
step 03
Measures
Improvements engineered and ranked: setpoints, containment, free cooling, plant staging, with CFD where airflow is the question.
step 04
Verification
Measures implemented in sequence, with the improvement verified at the meter, season against season.
from our portfolio
PUE Optimization projects we worked on
Certified floor area across Serbia, Italy, Switzerland and beyond, from public infrastructure to commercial high-rise.
PUE improves in decimals, and the first question is usually whether the number being reported is the number being run.
Metering boundaries and seasonal variation move the figure considerably, so an accurate baseline is worth establishing before improvements are claimed or budgets set against them.
01 measurement
Accurate baseline
Metering boundaries and seasonal variation are settled before the baseline is recorded. Without that, an improvement can appear or disappear depending on which month was measured and where the meter was placed.
02 Evidence
Simulation-backed setpoints
Raising supply temperatures is one of the largest PUE levers and the one operators are most cautious about. Our CFD work establishes the thermal margins, so the decision rests on evidence rather than on nerve.
03 ENGINEERING
Climate-matched cooling
Economiser strategies are tuned to the hours the local climate provides, with adiabatic cooling where the water case supports it. The available hours vary considerably by site, and so does the right strategy.
04 operations
Delivered without downtime
Measures are staged according to the facility's tolerance for risk and verified one at a time, with production untouched throughout.
Related services
CFD-DC
CFD for Data Centers
Airflow and cooling simulation for white space and plant. Containment, free-cooling hours and failure scenarios tested in the model first.
Mechanical, electrical and plumbing design from concept to construction issue, engineered against a modelled performance target and verified after handover.
Building-systems verification and testing across handover, occupancy and the first operating year. Fundamental and enhanced Cx, with the correspondence held in-house.
