Holding data center PUE under 1.2 with telemetry you already have
Chiller loops, CRAC set points and weather-aware pre-cooling: the three levers that move PUE, and how to watch them without a separate DCIM.
Power usage effectiveness is a ratio, so it moves for two reasons: IT load changes, or cooling and distribution overhead changes. You control the second.
Lever 1 — chiller loop temperature
Every degree of chilled-water set point you can raise safely is roughly 2–3% less chiller energy. Watch supply and return temperature as live channels and set a warning threshold on return temperature, not on the room.
Lever 2 — CRAC and airflow
Hot-aisle/cold-aisle containment fails quietly: a missing blanking panel shows up as a temperature spread across racks. Track differential pressure and rack inlet temperature per zone, and let anomaly detection flag spread rather than absolute values.
Lever 3 — weather-aware pre-cooling
Outside air temperature predicts your afternoon cooling load. A facility map with a weather overlay lets operators see the next peak-tariff window and pre-cool before 17:00 instead of fighting it at 18:30.
| Metric | Target | Where to see it |
|---|---|---|
| PUE (rolling 24 h) | < 1.20 | Energy & Utilities |
| Return water ΔT | 6–8 °C | Live Telemetry |
| Inlet temperature spread | < 3 °C | Anomaly radar |
Make it stick with digests
A daily digest with the energy curve, held alerts and overdue services is the cheapest way to keep PUE on the agenda. Pick the hour it lands — 07:00 local for the operations lead works well — and compare week over week.
Ready to run every shift from one command center?
Start on the free Starter plan — one site, your whole team, every monitoring feature. Upgrade when the fleet grows.