How to Size HVAC Ductwork: Metal vs. Flex Duct Sizing Guide

A duct is sized correctly when it passes two separate checks, not one. First, it has to be large enough that the design airflow can get through the whole run without using up more pressure than the blower has available: that's the friction-rate check, from ACCA Manual D. Second, it has to be small enough that the air moving through it doesn't generate noise: that's the velocity check. The section you actually install is whichever of the two sizes comes out larger. A calculator, chart, or slide rule that only checks one of these will hand you a duct that's correct on paper and audibly wrong in the house.

What sizing a duct actually means: two constraints, not one

Friction rate and velocity aren't two ways of asking the same question. They pull in opposite directions, and a duct sized for one alone will often fail the other.

The friction side asks: given the pressure the blower has left over after the coil, filter, and grilles each take their share, and given the length of the longest duct run, how much airway is needed to push the design CFM through without stalling the system? Undersize here and the blower struggles to hit its rated airflow, comfort suffers, and the compressor short-cycles against restricted flow.

The velocity side asks a different question: at that airflow, how fast is the air actually moving through the duct's cross-section? Air moving too fast through a duct generates turbulence noise at the register, and past a certain point that whistle or rush is what the homeowner calls back about, even though the system is otherwise working exactly as designed.

ACCA Manual D resolves the conflict directly: compute both the friction-rate size and the velocity-limit size for a given section, then install the larger one, dampering off the extra airflow if the velocity size turns out larger than the friction rate strictly requires. Skipping the velocity check because the friction-rate number looks fine is the shortcut that most often turns a correctly calculated system into a noisy one.

How to calculate duct size: the friction-rate method

ACCA Manual D §1-6 defines the design friction rate with one formula:

Friction Rate = (Available Static Pressure × 100) ÷ Total Effective Length

Available static pressure (ASP) is what's left of the blower's total external static pressure after subtracting the pressure drops of everything else in the airstream: the coil, the air filter, the supply outlet, the return grille, and the balancing damper. It is not the blower's full rated static. That number overstates what's actually available to move air through the ducts themselves.

Total effective length (TEL) is the length of the single longest airflow path in the system, measured from the farthest supply outlet, through the equipment, to the farthest return grille, with the straight duct footage added to the equivalent length of every fitting, elbow, and transition along that path. A 90-degree elbow doesn't add its physical length to this total; it adds whatever straight-duct length produces the same pressure drop, a number Manual D Appendix 3 tabulates per fitting type.

Divide ASP by TEL and multiply by 100, and the result is the friction rate: how much pressure the design can afford to lose per 100 feet of duct. ACCA Manual D Appendix 4 gives the acceptable range as 0.06 to 0.18 inches of water column per 100 feet. Land below 0.06 and the resulting ducts come out oversized for the building, usually a sign the blower doesn't have enough static pressure budget for the length of duct it's feeding. Land above 0.18 and the ducts come out small and fast enough that the fitting equivalent-length tables, which are only valid at or below 900 feet per minute, stop applying to your own design.

Once the friction rate is set, the diameter itself comes from the Darcy-Weisbach equation with the friction factor from the Colebrook equation, using the duct material's absolute roughness as an input. That roughness value is where metal and flex duct sizes diverge, covered in the next section. The diameter that comes out of the equation is then rounded up to the nearest manufactured size, never down, because rounding down reintroduces the exact restriction the friction-rate check was meant to rule out.

Worked example: sizing a 1,000 CFM trunk line

Take a system with a blower rated for 0.57 inches of water column of total external static pressure, feeding a 1,000 CFM supply trunk. The coil, supply outlet, return grille, and balancing damper together use up 0.27 inches of that static. The longest airflow path on this system, straight duct plus every fitting's equivalent length, measures 427 effective feet.

Step 1: Available static pressure. 0.57 minus 0.27 = 0.30 in.wg.

Step 2: Friction rate. (0.30 × 100) ÷ 427 = 0.070 in.wg per 100 ft. That falls inside the 0.06 to 0.18 acceptable range, so the design can proceed at this rate without redesigning the duct layout or upsizing the blower.

Step 3: Diameter. Solving Darcy-Weisbach and Colebrook for a galvanized trunk (absolute roughness 0.0003 ft) at 1,000 CFM and a 0.070 friction rate returns an exact diameter of 14.67 inches.

Step 4: Round up to a manufactured size. 14.67 inches rounds up to 15 inches, the nearest size a metal shop actually stocks and fabricates.

Step 5: Check velocity. At 1,000 CFM through a 15-inch round duct, the air is moving at 815 feet per minute. ACCA Manual D Table A1-1 sets a supply trunk's conservative velocity at 700 fpm and its maximum at 900 fpm. 815 fpm sits comfortably inside that ceiling, so the friction-rate size doesn't need to be bumped up further to satisfy the noise check.

The answer is a 15-inch round metal trunk, reached by computing both constraints rather than reading one number off a chart and assuming it covers both.

Metal vs. flex duct: why the same airflow needs a different size

Metal and flex duct are not interchangeable at the same labeled size, because the equation above is only as accurate as the roughness value it's fed, and the two materials have very different roughness.

ASHRAE Fundamentals Chapter 21, Table 1 classifies galvanized steel duct with longitudinal seams as "medium smooth," with an absolute roughness of 0.09 mm. It classifies flexible duct, fully extended, as "rough," with an absolute roughness of 3.0 mm, more than 30 times coarser than galvanized steel. That's not a manufacturing defect in flex; corrugated wire-helix duct is inherently rougher than smooth sheet metal, and the friction equation treats that roughness as a real physical input, not a rounding allowance.

Held at one airflow and one friction rate, the difference shows up directly in the diameter. At 600 CFM and a 0.10 in.wg per 100 ft friction rate, solving for the unrounded diameter in each material gives 11.27 inches for galvanized steel, 12.03 inches for rigid fibrous glass duct board, and 12.78 inches for fully extended flex. Rounded up to manufactured sizes, that's a 12-inch metal duct against a 13-inch flex duct for the identical airflow and friction budget: one full standard size larger, not a rule-of-thumb bump.

That duct-board number is worth separating out on its own, because it gets confused with flex in some published charts. Fibrous glass duct board has an absolute roughness of about 0.003 ft, roughly 0.9 mm, a different row of ASHRAE's table entirely from flex's 0.0098 ft, roughly 3.0 mm. Treating flex as though it had duct board's roughness understates its friction by close to 30 percent, and a duct sized on that mistake comes out one size too small.

That flex number also assumes the duct is pulled fully straight. Compressed flex performs substantially worse, and most installed flex is compressed to some degree: coiled slack, tight joist bays, and sagging between hangers all shorten the duct's effective length and increase its resistance. ASHRAE Fundamentals models this with a correction factor of roughly 1 + 9.86 times the compression ratio. Independent testing by Abushakra, Walker, and Sherman, measuring 6-, 8-, and 10-inch duct against ASHRAE Standard 120-1999, found the real penalty was size-dependent, and at 15 percent compression, a condition the researchers describe as typical for field installations, measured roughly 4.2 times the fully extended pressure drop on 8-inch duct against ASHRAE's own model predicting about 2.5 times. The two published models disagree with each other by a wide enough margin that neither should be treated as a precise correction. The safer approach is to size flex assuming it's pulled fully tight, then actually pull it fully tight on the install rather than sizing loose and hoping a compression factor covers the gap.

How to convert a rectangular duct to a round size

Rectangular and round ducts aren't sized on cross-sectional area. They're sized on the round duct that produces the same friction loss at the same airflow, a relationship called the circular equivalent. The formula, developed by Huebscher in 1948 and published as ASHRAE Equation 25, is:

D_e = 1.30 × (a × b)^0.625 ÷ (a + b)^0.25

where a and b are the rectangle's two side lengths in the same units.

Run the numbers on a 20-by-8-inch rectangular duct and the circular equivalent comes out to 13.48 inches. That's the diameter of the round duct that carries the same airflow with the same pressure loss, and a 13.48-inch circle only encloses about 143 square inches, against the rectangle's 160 square inches. The round duct does the same job with roughly 11 percent less material, because a circle has less wetted perimeter per unit of enclosed area than a flat rectangle does, and friction loss scales with the perimeter the air is dragging against, not just the area it's moving through. The flatter the rectangle, the bigger that gap gets.

Velocity works differently, and this is where people trip: velocity is always computed on the duct's real physical area, never the circular equivalent. That same 20-by-8-inch duct carrying 800 CFM has an actual area of 160 square inches (1.11 square feet), which puts the air velocity at 720 feet per minute. Using the equivalent diameter for a velocity check instead would understate how fast the air is actually moving.

What is the maximum air velocity for a duct?

The ceiling depends on which section of the system it is, because ACCA Manual D Table A1-1 treats velocity as a noise limit specific to where a duct sits relative to the living space, not a single number that applies everywhere:

• Supply trunk: 700 fpm conservative, 900 fpm maximum, for both rigid and flex.
• Rigid supply branch: 600 fpm conservative, 900 fpm maximum.
• Flex supply branch: 700 fpm conservative, 900 fpm maximum. Flex is allowed a higher conservative figure because its wall absorbs sound instead of transmitting it.
• Return trunk: 600 fpm conservative, 700 fpm maximum.
• Rigid return branch: 500 fpm conservative, 700 fpm maximum.

Return-side limits sit lower across the board because return grilles are almost always mounted in occupied space, close to where people sit and sleep, so there's less tolerance for airflow noise there than at a supply register tucked into a ceiling or a floor.

There's a second, easy-to-miss reason not to exceed 900 fpm anywhere in the system. Manual D's own fitting equivalent-length tables, the ones used to build total effective length in the friction-rate calculation, are only tabulated for airflow at or below 900 fpm. Run a duct faster than that and the TEL used to compute the friction rate in the first place is no longer built on valid numbers. The velocity ceiling and the friction-rate method aren't independent rules; they depend on each other.

Why does this differ from a ductulator or a slide rule?

Because the slide rule is deliberately more conservative than the raw equations, and the gap between the two is consistent and measurable. The ACCA Duct Sizing Slide Rule that Manual D's own worked examples are read from returns 0.066 in.wg per 100 ft where Darcy-Weisbach with ASHRAE's medium-smooth roughness computes 0.060 for 1,000 CFM of round sheet metal, and at a 14-inch duct size, the slide rule reads 0.10 against the equation's 0.089. That's a 10 to 13 percent spread, and it runs in the same direction every time: the slide rule sizes ducts slightly larger than the pure equations would.

That's not the slide rule being wrong. It's modeling duct with real installed joint irregularity, closer to ASHRAE's "average" roughness category than the "medium smooth" category the underlying friction chart assumes. Abushakra, Walker, and Sherman measured the same bias independently and found the ACCA chart over-predicted pressure drop, relative to their measurements of fully stretched duct, by an average of 21 percent across every size they tested. The practical takeaway: a section that lands within one manufactured size of its limit under either method is worth bumping up one size rather than treating the calculated number as exact to the decimal.

This is also why elevation matters less than people assume, and why most published charts don't bother correcting for it. They're built at sea level, 14.696 psia, 70°F air at 0.075 lbm per cubic foot. ASHRAE notes the standard friction chart needs no correction for elevations up to roughly 1,600 feet, temperature swings of about ±27°F, or duct pressures within ±20 in.wg of ambient, because all of those keep the resulting error under 5 percent. Above that elevation, the air is less dense and the true friction loss runs below what a sea-level chart returns, so the sizing error, where one exists, lands on the safe side: slightly larger duct than strictly required, not smaller.

Getting the load right before you size a single duct

None of the arithmetic above means anything if the airflow number feeding it is wrong. Manual D takes a room-by-room CFM requirement as its input; it doesn't generate one. That number comes from a proper Manual J load calculation, not from a square-footage rule of thumb, and the two produce different results often enough that it's worth reading why square-footage HVAC sizing rules miss by 40 to 100 percent before locking in the CFM this whole calculation depends on. If you're starting from an existing system and just need to confirm what's already installed, decoding the tonnage off the equipment's data plate with an AC tonnage calculator is the fastest way to back into a rough CFM target before refining it room by room.

The easy way

Doing this by hand means solving Darcy-Weisbach and Colebrook for diameter, converting rectangular runs through the Huebscher equation, and cross-checking two separate velocity tables, for every section, on every job. Our duct size calculator runs all of it from the airflow and a design friction rate: it solves the exact diameter, snaps to a manufactured size, applies the correct roughness for metal, duct board, or flex, and returns the Manual D verdict on whichever of the friction-rate size or velocity-limit size comes out larger, the check most hand calculations and simple ductulators skip.