What is power usage effectiveness?

Power usage effectiveness, usually shortened to PUE, measures how much of a data center's electricity actually does computing and how much goes to supporting it. It is the total facility energy divided by the energy delivered to the IT equipment. Because cooling and power conversion are costs that scale with every watt of computing, the ratio shapes both the energy bill and the carbon footprint.

How it works

The formula is PUE = total facility energy / IT equipment energy. A value of 1.0 would mean every unit of energy reaches the servers. A value of 1.5 means that for every unit used by servers, half a unit more goes to cooling, lighting and losses in power conversion.

There is a feedback loop to watch. Servers produce heat in proportion to the energy they use, and the cooling system must remove that heat, so overhead rises with load. But the relationship is not fixed: cooling plants are often least efficient when lightly loaded, and warmer outside air can raise the cooling share. Improving PUE lowers the cost of each unit of computing without changing the computing itself.

Example

Say a facility's servers draw 1,000 kW and the cooling, lighting and power conversion draw another 500 kW. The total is 1,500 kW, so PUE = 1,500 / 1,000 = 1.5. If an upgrade cuts the overhead to 200 kW, PUE falls to 1,200 / 1,000 = 1.2. At a price of 0.10 per kWh, running all year (8,760 hours) the total bill drops from 1,500 x 8,760 x 0.10, about 1.31 million, to 1,200 x 8,760 x 0.10, about 1.05 million, a saving of roughly 260,000 a year. These numbers are an illustration.

How to model it in Machinations

Use a Source for grid power, a Pool for total energy used, and a Converter that sends part of it to IT load and part to overhead. A Register computes PUE as total divided by IT energy, and a state connection can raise overhead as IT load rises. Make the overhead share a Lever, then Play and watch the energy bill and PUE chart change.