The Real Cost of Oversizing an AHU: Capex, Footprint, Part-Load SFP

One casing size up pays for itself in under two years on a continuously-run unit — that argument is easy to make and correct. The mistake is treating "bigger is cheaper to run" as a rule with no ceiling. Push face velocity down far enough and the relationship inverts: capex and footprint keep climbing linearly with frontal area while the SFP saving flattens out, because filter and coil pressure drop don't fall as fast as velocity squared once you're already in the low-velocity range. Then the same oversized fan gets punished twice more at part load, which is where most AHUs actually spend their operating hours.

Where the SFP curve stops paying for itself

Frontal area is airflow divided by face velocity — A = q_v / v. At 5 000 m³/h (1.39 m³/s), the four common casing steps look like this:

Face velocity	Frontal area	Δ area vs 2.5 m/s	Typical internal Δp
2.5 m/s	0.56 m²	—	~690 Pa
1.8 m/s	0.77 m²	+39 %	~420 Pa
1.5 m/s	0.93 m²	+67 %	~330 Pa
1.2 m/s	1.16 m²	+108 %	~260 Pa

The first step (2.5 → 1.8 m/s) is the one worked in the SFP-to-euros article: +39 % frontal area buys a 39 % drop in internal Δp, because at this velocity range coil and filter losses still track close to v^1.8–2.0. Casing steel, coil face, and filter cells all price roughly with frontal area, so capex and SFP move together and the trade is clearly worth it.

Go further, to 1.2 m/s, and the picture changes. Frontal area is now up 108 % against the base case — more than double the casing, coil and filter cost — but internal Δp has fallen only from 420 to roughly 260 Pa, because filter media pressure drop is closer to v^1.0–1.3 and stops shrinking proportionally. You are paying for area the airstream barely uses to save pressure that was already mostly gone. Take the actual dP from the manufacturer's selection data at each face velocity rather than a square-law estimate — see face velocity and casing sizing for why the exponent varies by component.

The footprint cost nobody schedules

Frontal area doesn't stay in the AHU — it sets the plant room opening, the riser, and often the roof curb. A unit that grows from 0.56 to 1.16 m² of face area needs a proportionally larger service corridor for filter withdrawal (EN 13053 requires clear access on the downstream face of every filter and coil section), and on a retrofit that space usually doesn't exist without demolishing something. Oversizing a new-build plant room by one step is a drawing revision; oversizing an existing one by two steps is often the reason the "cheaper to run" unit doesn't fit through the door.

Part-load SFP: the second penalty

Most AHUs don't run at design flow most of the time — VAV terminals, night setback and demand-controlled ventilation mean an office unit commonly spends the bulk of its hours between 40 % and 75 % of nominal airflow. Fan affinity laws say power falls with the cube of flow, but total efficiency also falls as the fan moves away from its best-efficiency point, and an oversized fan selected for a low-velocity design point is already sitting further from BEP at 100 % load — so it drifts even further off the curve as flow drops.

Load fraction	Cube-law power (ideal)	Typical efficiency droop	Effective power
100 %	100 %	reference	100 %
75 %	42 %	−3 to −5 pts	~45 %
50 %	12.5 %	−6 to −10 pts	~15 %
25 %	1.6 %	−15 to −20 pts	~2–3 %

The absolute kW at low load is small, but a fan chosen two casing sizes bigger than needed spends most of its life in the region where the affinity-law estimate is least accurate — which is exactly where quotes stop checking it, because the headline SFP figure is calculated at 100 % design flow.

Finding the actual sweet spot

The rule from the euros article still holds — one casing size up is usually worth it on any unit running more than a few thousand hours a year. What changes here is the stopping point: check the marginal capex of the next size against the marginal Δp it actually removes at that specific duty, using real selection curves, not a velocity-squared shortcut applied past the range it holds in.

In AirSelect3D, SFP_int and internal Δp recalculate from the real fan and coil selection engines on every casing resize, so the diminishing-returns point is visible before the quote goes out — not discovered when the plant room drawing comes back.

Compare casing sizes on a real duty →