Insights · Standards
Every duct system leaks. The question a specification answers is how much is allowed, and the answer is written as a leakage class. This is a practical guide to the SMACNA leakage class and the F = CL x P0.65 formula behind it, how the seal class you build to sets the leakage class you can expect, the EN 1507 and DW/144 air-tightness classes used elsewhere, and how duct is leak tested on site.
Air leakage is the air that escapes through the joints, seams and penetrations of a duct instead of reaching the terminal it was sized for. It is never zero. A leaky supply system makes the fan move more air than the design called for, which shows up as higher fan energy every hour the system runs, rooms that never reach setpoint at the end of long runs, and balancing reports that will not close. On a negative-pressure return or a fume system, leakage pulls unconditioned or contaminated air in rather than pushing it out, which is a safety question rather than an energy one.
Leakage is set at the shop, not on site. Once the duct is hung, sealing a joint you can no longer reach is slow and unreliable. That is why a specification fixes an allowable leakage before fabrication, and why the number it uses, the leakage class, is worth understanding before you cut steel.
SMACNA, the Sheet Metal and Air Conditioning Contractors' National Association, expresses allowable leakage as a leakage class, written CL. The class is the leakage rate in cubic feet per minute per 100 square feet of duct surface area, measured at a reference static pressure of 1 in. w.g. A lower CL number is a tighter duct.
Because duct rarely runs at exactly 1 in. w.g., the class is scaled to the real pressure with a power law:
F = CL × P0.65
Here F is the leakage in cfm per 100 sq ft of duct surface, CL is the leakage class, and P is the static pressure in inches of water gauge. The 0.65 exponent is the value SMACNA uses to classify duct. The real exponent for a specific leakage path can sit anywhere between about 0.5 and 0.9, but 0.65 is the number the class system is built on. The classes you will see quoted are a discrete set:
| Leakage class (CL) | Relative tightness | Where it typically applies |
|---|---|---|
| CL 3 | Tightest commonly published | Sealed round and flat-oval duct at high pressure, Seal Class A |
| CL 6 | Tight | Sealed rectangular duct at high pressure, Seal Class A |
| CL 12 | Moderate | Sealed round duct at low pressure, or rectangular at Seal Class B |
| CL 24 | Loose | Sealed rectangular duct at low pressure, Seal Class C |
| CL 48 | Loosest | Unsealed rectangular metal duct |
The single most common mistake is to quote one leakage class for a duct type without naming the pressure and seal class it belongs to. Sealed rectangular duct is not one number. It is CL 24, 12 or 6 depending on how tightly it is sealed, which is the subject of the next section.
The leakage class is an outcome. The thing the shop actually controls is the seal class, which says which parts of the duct get sealed: Seal Class C seals the transverse joints, Seal Class B adds the longitudinal seams, and Seal Class A adds every duct-wall penetration as well. Tighter sealing gives a lower expected leakage class. SMACNA correlates the two like this:
| Construction | Seal Class C | Seal Class B | Seal Class A |
|---|---|---|---|
| Rectangular metal duct | CL 24 | CL 12 | CL 6 |
| Round and flat-oval metal duct | CL 12 | CL 6 | CL 3 |
Two things fall out of that table. First, round and flat-oval duct sits about one leakage class tighter than rectangular at the same seal class, because a roll-formed spiral lockseam has far less joint length to leak through than a four-corner rectangular section. Second, the seal class is itself tied to the pressure class. The 2005 SMACNA HVAC Duct Construction Standards assign Seal Class C to 1/2, 1 and 2 in. w.g., Seal Class B to 3 in. w.g., and Seal Class A to 4, 6 and 10 in. w.g., though the 2 in. w.g. boundary sits at Seal Class B in some presentations, and ASHRAE 90.1 now asks for Seal Class A regardless of pressure class. Either way, a high-pressure spec pulls you down to the tightest leakage class through the seal class, without it ever being written explicitly.
One note on editions. The values above are the long-standing SMACNA figures carried in the HVAC Duct Construction Standards. A later revision of the SMACNA leakage test manual tightened the expected classes, so if your specification cites a specific manual edition, read the class table in that edition rather than assuming these numbers.
A leakage test is simple in principle. A section of the finished system is isolated and blanked off at both ends. A blower pressurises it to the test pressure, and the airflow the blower has to keep supplying to hold that pressure is measured through a calibrated orifice. That make-up flow is the leakage out of the section. Divide it by the surface area tested and you have the leakage in the same cfm-per-100-sq-ft units as the class.
Specifications rarely ask for the whole system to be tested. They name a representative fraction, a test pressure equal to the section's design pressure, and a pass criterion. That criterion is often written two ways at once: a leakage class the section must beat, and a cap on total leakage as a percentage of system airflow. The ASHRAE Handbook puts acceptable whole-system leakage in the range of roughly 1 to 5 percent of design airflow at operating pressure, and project specs commonly land within that band. The designer of record sets the exact figure.
A tight leakage class is won or lost at three points in fabrication. The transverse joint is the biggest single leakage path, so a TDF flange rolled square on the duct end, closed with a factory-applied gasket and clipped tight, is doing most of the work on a Seal Class A rectangular job. The longitudinal seam is the second: a Pittsburgh lock has to be fully rolled and, above the lower seal classes, sealed along its length. The third is simply squareness. A duct end that is out of square will not pull a gasketed flange up evenly no matter how much sealant goes on, so a line that holds length and width to tight tolerances is quietly buying leakage performance.
This is also where duct shape earns its place in the decision. Because round and spiral duct leaks about half of what rectangular does at the same seal class, moving a high-leakage-sensitivity run to spiral can hit a tight class with less sealing labour than forcing rectangular duct down to the same number. Matching the machinery to the classes you build, rectangular flange-and-seam work, spiral lockseam, or both, is a spec-time decision worth making deliberately rather than discovering on the test day.
Ask Taokron which line suits your leakage classes →
Outside North America the same physics appears under different labels.
For a fuller side-by-side of the construction standards these leakage classes live inside, see International duct standards compared.
A mechanical specification will usually pin down leakage in one of two styles. A SMACNA-based spec reads like: "Supply ductwork shall be sealed to Seal Class A and shall not exceed Leakage Class 6. A minimum of 25 percent of the system shall be leak tested at 2 in. w.g." A European or Australian spec instead reads in percentages: "Ductwork shall achieve air-tightness Class C to EN 1507," or "leakage shall not exceed 5 percent of design air quantity when tested to AS 4254.2." Both are saying the same thing in different units: how tight, at what pressure, proven by what test.
A duct leakage class, CL, is the allowable leakage in cfm per 100 sq ft of duct surface at a reference pressure of 1 in. w.g. A lower CL is tighter. Common classes are CL 3, 6, 12, 24 and 48, and leakage at other pressures follows F = CL x P to the 0.65 power.
F = CL x P0.65, where F is leakage in cfm per 100 sq ft of duct surface, CL is the leakage class, and P is static pressure in inches w.g. SMACNA uses the 0.65 exponent for classification; real leakage paths can range from about 0.5 to 0.9.
Tighter sealing gives a lower leakage class. For rectangular metal duct, Seal Class C, B and A correlate to about CL 24, 12 and 6. Round and flat-oval duct is roughly one class tighter, near CL 12, 6 and 3. Unsealed rectangular duct is about CL 48.
A duct section is isolated and blanked, a blower pressurises it to the test pressure, and the make-up airflow needed to hold that pressure is measured through a calibrated orifice. That flow is the leakage, which is then compared to the specified class or percentage limit.
EN 1507 and EN 12237 define classes A to D, Class A leakiest and Class D tightest, with f_max = C x p^0.65 and the constant stepping 0.027, 0.009, 0.003 and 0.001. Each class is three times tighter than the last. DW/144 uses the same four limits.
Got a spec question on this guide? A Taokron mechanical engineer replies within 12 hours, not a salesperson.
Ask an engineer