Why Data Centers Are Ideal for Heat Recovery
Data centers reject enormous, continuous heat. Almost every watt of IT load becomes heat that must be removed, and the air leaving the computer-room air-handler is warm and clean year-round. Because the exhaust is steady, uncontaminated, and available 24/7, an air-to-air recovery device can pre-condition the equal make-up air with very high annual availability. That makes the data center one of the strongest heat-recovery business cases in the commercial building stock — the energy is there whether or not you capture it.
Air-Side vs. Water-Side Recovery
| Approach | What it recovers to | Best when |
|---|---|---|
| Air-side plate / wheel | Incoming outdoor air | Make-up air must be conditioned anyway |
| Water-side coil loop | Hydronic heating circuit | Heat is reused in another building |
| Heat pump upgrade | Higher-temperature water | District heating or process heat |
Where the Recovered Heat Goes
- Adjacent spaces — offices, amenities, or labs served by the same plant.
- District heating — a growing use where the campus or city has a network.
- Domestic water pre-heat — a simple, always-useful load.
- Free-cooling assist — in cool months the recovery pre-heats intake while an economiser handles the rest.
Device Selection
Fixed-plate exchangers dominate in data centers because the exhaust is clean and the units have no moving parts to fail. Enthalpy wheels add latent recovery where intake is humid, but their rotation and purge need maintenance that a 24/7 facility must plan for. Redundancy and ease of core replacement matter more than squeezing the last point of peak effectiveness.
Worked Example
A facility with a 2 MW IT load exhausts about 30,000 cfm at 24°C. A 70% sensible plate pre-conditions the make-up air, recovering on the order of 150–200 kW of heating value in winter that would otherwise be purchased as gas for the administrative block. Over a heating season that is a five-figure energy saving with no change to the IT cooling path.
Frequently Asked Questions
Does recovery hurt PUE? It can improve the effectively-used PUE when the captured heat offsets a real separate load; report it transparently.
Is the exhaust clean enough for a wheel? Usually yes, but keep supply-air filters to protect the core.
What is the biggest risk? A single point of failure — size recovery with the same redundancy philosophy as the cooling plant.
Practical Implementation Steps
Start by metering the actual exhaust flow and temperature over a representative week. Data-center airflow is usually well-instrumented, but confirm the recovery duty against live data rather than nameplate, because the make-up air — not the total IT exhaust — is what the recovery device must condition. Size the plate or wheel to that make-up flow, and integrate its control with the CRAH and economiser logic so the core modulates with outdoor air instead of fighting free cooling. Specify redundant cores or a clear bypass path so maintenance never takes the cooling offline, and instrument supply and exhaust temperatures on both sides so the recovered kilowatts can be reported next to PUE.
Avoiding the Common Pitfall
The frequent failure is over-specifying peak effectiveness while under-sizing access. In a 24/7 facility the core will need washing, and if it cannot be pulled in minutes the recovery gets bypassed "temporarily" for months, quietly erasing the saving. Design for the wrench, not just the datasheet: a slightly lower-effectiveness core that is actually maintained beats a higher one that is abandoned. With the device maintained and the control joined to the plant, the data center captures a steady, auditable fraction of its own rejected heat and turns a pure liability — server heat — into a useful, billable resource for the rest of the site.
Bottom line: a data center's rejected heat is too large and too steady to ignore; with a maintained plate or wheel joined to the plant control, recovery turns a pure cooling liability into a measurable, auditable resource for the rest of the site.
