A1: Power Usage Effectiveness (PUE)


This metric is the ratio of the total data center energy use to total IT energy use. The total energy use includes all energy types supplied to the datacenter (electricity, fuel, district chilled water, etc.). All the energy data values in the ratio are converted to common units. For all-electric data centers, PUE can be calculated using site energy. For data centers that also have other forms of energy (e.g. fuel, district chilled water) PUE should be calculated based on source energy.

Units: Dimensionless

For an all-electric data center:
PUE = dA1 dA2
dA1: Annual Electrical Energy Use (kWh)
dA2: Annual IT Electrical Energy Use (kWh)
For data centers that also have other types of energy:
PUE = (dA1 + dA3 + dA4 + dA5) dA2
dA1: Annual Electrical Energy Use (Source MMBTU)
dA2: Annual IT Electrical Energy Use (Source MMBTU)
dA3: Annual Fuel Energy Use (Source MMBTU)
dA4: Annual District Steam Energy Use (Source MMBTU)
dA5: Annual District Chilled Water Energy Use (Source MMBTU)
Note: To calculate source energy, site energy should be multiplied by the source factor for each energy type. For national average source factors for different energy types, see "ENERGY STAR Performance Ratings Methodology for Incorporating Source Energy Use".

See Required Data for more information on the data items.


Standard Good Better
2.0 1.4 1.1

This metric can be benchmarked relative to other facilities in the LBNL database, although the LBNL database contains PUE based on power rather than annual energy. For the LBNL dataset, the average PUE value is 1.83.

Figure. Power Usage Effectiveness metric for data centers in the LBNL database. Note that these PUE values are based on power, not energy.

Actions Inferred:

This metric provides an overall measure of the infrastructure efficiency i.e. higher values relative to the peer group suggest higher potential to improve the efficiency of the infrastructure systems (HVAC, power distribution, lights) and vice versa. Note that it is not a measure of IT efficiency.

Special Considerations:

Since this metric does not account for the efficiency of the IT itself, it is important to note that if a data center has a low PUE, there may still be major opportunities to reduce overall energy use through IT efficiency measures such as virtualization, etc. The ability to decrease PUE is also affected by climate (e.g. free cooling offers much greater potential in cooler climates).