How to Read an AHU Technical Data Sheet: A Consulting Engineer's Guide

An AHU technical data sheet is not a brochure — it is a set of claims, and almost every one of them can be checked with a calculator in under a minute. Most reviewers never do, because the document arrives as twelve dense pages with no obvious entry point. This is the reading order we would use, and the arithmetic that turns each page from an assertion into a verified number.

Read it in this order

1. The verdict lines first. ErP (EU) 1253/2014 pass/fail, the ECP-05-2026 energy class for both seasons, and the EN 1886 casing classes. These are the conclusions; everything else exists to support them.
2. The operating point. Airflow (supply and extract), external static pressure, altitude/air density assumption, and the design air conditions. Every downstream number depends on these four, and a data sheet that computes a coil at one airflow and a fan at another is not rare.
3. The section list, in airflow order. Each component with its own pressure drop and its own selection reference.
4. The arithmetic. Now go back and check that the sections add up to the totals.

If step 4 fails, the rest of the document is decoration.

Five cross-checks that take one minute each

1. SFP against fan total pressure

Specific fan power is electrical input power divided by volume flow, so for a single fan it collapses to a very useful identity:

SFP ≈ Δp_total ÷ η_total

where η_total is the combined impeller, motor and drive efficiency. A fan working against 850 Pa at 65 % overall efficiency lands at 850 / 0.65 ≈ 1 308 W/(m³/s). If a sheet claims 850 Pa and an SFP of 700 W/(m³/s), it is implicitly claiming 121 % efficiency — a physical impossibility, and the single fastest way to catch a fabricated or curve-fitted fan selection. The EN 16798 roll-up to unit SFP then has to reconcile with the per-fan figures.

2. Coil duty against mass flow

At 5 000 m³/h and 1.2 kg/m³, the mass flow is 5 000 / 3 600 × 1.2 = 1.67 kg/s. A heating coil lifting air from 2 °C to 22 °C must therefore deliver 1.67 × 1.005 × 20 = 33.5 kW. If the sheet says 28 kW for the same air-side states, either the density assumption differs (altitude? entering temperature?) or the number is wrong.

3. Water side against air side

The same duty has to appear on the water circuit. With an 80/60 °C regime (ΔT = 20 K), 33.5 kW needs 33.5 / (4.18 × 20) = 0.40 kg/s ≈ 1.44 m³/h. Check the stated flow rate and the water pressure drop that goes with it — a coil quoted with plausible capacity but implausible water flow usually means the circuiting was never actually sized.

4. Heat recovery efficiency from the temperatures

Dry thermal efficiency per EN 308 is the temperature rise achieved divided by the temperature difference available:

η_t = (t_{after HRS} − t_outdoor) / (t_extract − t_outdoor)

Outdoor −10 °C, extract 22 °C, supply leaving the exchanger at 12.4 °C gives 22.4 / 32 = 70 %. If the stated efficiency and the stated state points disagree, the psychrometric section and the compliance section were produced by two different calculations — which is exactly the failure ErP compliance depends on not happening.

5. Pressure budget closure

Sum the per-section pressure drops (filters at their design — not initial — condition, coils, HRS, dampers, silencers) and add the external static. The result must equal the fan total pressure used in check 1. Filters are where this quietly breaks: a sheet using clean-filter dP understates the fan duty and therefore the SFP and the energy class.

Check	Arithmetic	Typical failure it exposes
SFP vs fan pressure	SFP = Δp / η	Implied efficiency > 100 %
Coil duty	Q = qm · cp · ΔT	Wrong density or wrong airflow
Water flow	Q = qw · 4.18 · ΔTw	Circuiting never sized
HRS efficiency	ηt from state points	Compliance ≠ psychrometrics
Pressure budget	Σ section dP + external	Clean-filter dP used

Red flags that need no arithmetic at all

• A single-season energy class. ECP-05-2026 is evaluated winter and summer; a unit can pass in January and fail in July, which is why both fs-Pref values belong on the sheet.
• One dBA figure instead of a spectrum. Without octave bands there is no way to check an NR curve, and dBA totals hide the 125 Hz problem.
• No provenance line. A capacity with no vendor, product code or selection reference is an assertion, not a claim you can audit — the point of the consultant's checklist.
• Dimensions that contradict the drawing. If the section list and the general-arrangement drawing were generated from different models, one of them is stale.

Why this is a software problem, not a paperwork problem

Every check above fails for the same underlying reason: results stored separately from the inputs that produced them. When the data sheet, the drawing, the DXF export and the 3D geometry all derive from one model, the arithmetic closes by construction — there is no version of the unit that disagrees with another version of the unit. That is the difference between a document that survives review and one that comes back with questions.

Generate a data sheet whose numbers reconcile →