FORUM

Duct leakage and leakage testing

Ian Wills BE MEngSc PhD CPEng FIEAust M.AIRAH

Abstract
Although air leakage from ducts has previously been treated as insignificant and leakage testing as not cost effective, this paper
demonstrates that even minor leakage can have a significant effect on plant energy consumption and greenhouse gas emissions.
Recognising this, the recently published AS 4254.2:2012 reverses past practice and now mandates leakage testing of systems over
3000 L/s. Not only will reducing leakage reduce energy use and greenhouse gas emissions producing a better outcome for the building
owner and community, but attention to issues raised in this paper will benefit the contractor by reducing or eliminating the need for
costly rectification of leaks revealed by leakage testing. The paper examines the implications of the Standards testing requirements
and suggests extension of its scope.

1.Introduction
In the past, the effects of duct leakage have received little
attention in Australia. The last time AIRAH published a paper
on the subject was 1978 [1], and in 2002 AS 4254 [2] followed
earlier editions in recommending against duct leakage testing
on the grounds that leakage tests are an added expense in
system installation [and] generally not cost effective. With
the publication of AS 4254.2:2012 [3] that view has changed
and leakage testing is now required for each duct system over
3000 L/s. As the standard has been incorporated into the 2013
National Construction Code, leakage testing a will be a legal
requirement for all such systems after 1 May 2013. It is therefore
an appropriate time to look at duct leakage, duct sealing and
leakage testing in the light of the new standard.

2. Why worry about leaks?

2.1Effect of leakage on energy
and greenhouse gas emissions
Despite the view expressed in AS 4254:2002, a few simple
calculations suggest that there is reason for concern about
the impact of duct leakage.
Consider a typical air conditioning system in which the designer
follows AIRAH DA09 [4] and assumes a supply duct leakage rate
of 5%. To deliver the design air quantities to the spaces served, the
fan must handle 1/0.95 times the sum of the room air quantities
or 105.3% of the nominal air flow. Applying fan laws gives an
increase in fan power of 117%, so the widely accepted leakage
rate of 5% has added 17% to supply fan energy, for every hour
the plant operates. At 10% leakage the extra fan energy is 37%.
This is not the end of the story because leakage also affects
cooling and heating plant energy consumption. The size of
the effect depends on where the duct is located. If the duct is
in the conditioned space and the leakage percentage low, one
might argue that nothing need be done, that is, that the fan can
safely supply 100%, not 105.3% of design because the leaked air
produces useful cooling or heating effect. This is not the case if
the duct is in a ceiling return air plenum, as the leaked air will
travel around the system producing minimal useful cooling and
heating effect while increasing fan power and reducing return air
temperature slightly.
48

Eco l i b r i u m M AY 2 0 1 3

If the supply duct is outside the conditioned space, such

as in a ventilated roof space, the assumed leakage is simply lost
and the 17% increase in fan power is compounded by 5% waste
in cooling and heating effect and corresponding increase
in greenhouse gas emissions.
The analysis for return air ducts also depends on where the
return air duct is located. If the duct is in the conditioned space,
leakage has little or no effect since the air leaking into the duct
is the air that would have been returned anyway. If the return air
duct is outside the conditioned space, the effect is more serious.
Assume that under normal (non-economy cycle) operation
the plant handles 15% outside air, in which case return air will
be 85% of design supply air. Leakage at the rate of 5% into the
return air duct will thus be 5% of 85% or 4.3% of the design
supply air. If the air that leaks in is from outside the building, it
adds to the outside air load, the outside air percentage becoming
15% + 4.3% =19.3% of the supply air. Since the outside air load
is pro rata, the outside air load increases by 4.3% / 15% = 28%.
For a typical comfort cooling plant in Sydney, 15% outside air
would be about 18% of the peak cooling capacity so the leaked
outside air will add 28% * 18% = 5% to the peak cooling load.
In summary, a 5% leakage rate implies 17% increase in fan
power and fan energy on the supply side plus 5% additional
cooling and heating energy if the leakage is to outside the
conditioned space plus another 5% waste in heating and
cooling energy on the return side if it increases the outside air
percentage. The combined effects of these will depend on the
detail of the system. It will have less effect on a VAV system with
an economy cycle but more on a constant volume system with
a lower percentage of outside air. For the example discussed,
it is not unreasonable that a modest 5% leakage rate could add
10 or 15% to operating energy and greenhouse gas emissions.
We do not have published data for the effect of duct leakage in
Australian systems but there have been of a number of overseas
studies dealing with the issue. One [5] estimated the heating energy
wasted by duct leakage in Belgium at 15 GW.h (0.054 PJ) per annum
and 0.75 TW.h ((2.7 PJ) per annum for the rest of Europe (excluding
the former Soviet Union). Another study of VAV systems in large
commercial buildings in California [6] calculated that, compared to
tight duct systems (2.5% leakage), systems with 10% leakage had
annual HVAC system operating costs 9 to 18% higher, while those
with 5% leakage used 2 to 5% more energy.

FORUM

Static pressure limit

Positive

Negative

Maximum air
velocity

Pa

Pa

m/s

L/s/m duct surface area

Low pressure Class A

500

500

10

0.027 p0.65

Medium pressure Class B

1000

750

20

0.009 p0.65

High pressure Class C

2000

750

40

0.003 p0.65

Duct Pressure Class

Air leakage limit C in equation (2)

Figure 1: Shows equation (2) plotted for the pressure ranges above.

Maximum permissable leakage (L/s/m2 duct surface area)

1.700
1.600
1.500
1.400
1.300

Class A

1.200

Class B

1.100

Class C

1.000
0.900
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
0

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200

Pressure difference between inside and outside of duct (Pa)

Figure 2: Permissible duct leakage as a function of Class and pressure.

If duct leakage has such significant effects we might ask why

it has not been seen as a problem in the past. One answer is
that conventional plant design techniques assume leakage is
minimal so the fan is selected to handle the sum of the outlet air
quantities. In practice leakage does not show itself as a problem
because is compensated for by increasing fan speed or pitch
angle, a much cheaper solution for the contractor (but not the
building owner) than making the duct more airtight.

2.2Duct leakage, tightness, surface area

and pressure
Leakage is dependent on three factors: the tightness
of the construction and installation; the duct surface area,
and the pressure differential across the duct wall. This can
be characterised by a power relationship (for a discussion
of the derivation of this, see [7]):

Qleak = C As pn

(1)

Where Qleak is the quantity of leaked air, C is a constant related

to the duct tightness (which relates to the details of manufacture,
installation and sealing), As is the duct surface area, p is the
pressure difference between the inside and outside of the duct
and n is an exponent related to the geometry of the holes through
which the air leaks. For leaks through round holes, n is 0.5, while
for long openings such as longitudinal seams, it is 0.6 or higher.
Eurovent [8] and most other sources use a value of 0.65 for n.
Substituting this into equation (1) gives:

Qleak = C As p0.65 (2)

For Qleak in L/s, As in m and p in Pa, HVCA [9 Table 1] sets
the limits shown in figure 1. The values in the right hand column are
the same as for the same Leakage Class (A, B or C) in Eurovent [8].
may 2 0 1 3 Eco l i b r i u m

49

FORUM

3.Leakage testing
Test methods such as HVCA DW/143 [9] and SMACNA [10]
require a section of duct to be temporarily sealed by covering
the inlet end, incomplete branches, spigots and the like, to make
it air tight (Figure 2). The test apparatus consists of a relatively
low volume fan or blower fitted either with an inlet damper
or variable speed drive to permit the flow to be varied. This is
temporarily connected to the duct system via a flow measuring
device such as an orifice plate or anemometer. The blower speed
or damper is then adjusted to achieve the nominated static test
pressure in the duct, the air volume supplied by the blower being
the leakage from the duct at that pressure. If it is less than the
permitted leakage, the system passes. If not, the leaks must be
found, sealed and the system retested. Leak detection can employ
a variety of methods including visual inspection, listening for
leaks, feeling for air movement, application of soap and water
or use of tracers such as a smoke pellet in the duct or tracer
gasintroduced at the blower. When leaks are detected, the source
should be recorded and, since AS 4254.2:2012 requires only a
type-test, the information used to rectify leaks in the parts
of the system not tested.
Duct system under test

Flexible duct
Office plate
and manometer
to measure
leakage rate

Temporary caps
on duct ends,
branches and spigots
Manometer to measure
test pressure in duct

Variable speed blower

Figure 2: Typical leakage testing set up.

While leakage is clearly a cause for concern, leakage testing

has its own issues. Apart from the time involved, cost and
complication of temporarily closing up the duct, testing disrupts
efficient work flow on site. It is also essentially a contractual
hold point but may be difficult to program because the whole of
the section to be tested must be complete but not yet externally
insulated or enclosed in ceilings or risers. The contractor will
need to weigh up the cost of careful sealing of ducts at the factory
and as they are erected against the potentially greater cost and
complication of identifying and rectifying leaks found when the
system fails under leakage testing.
The demise in the 1970s of high-pressure, high-velocity
duct systems that were routinely leak tested has meant that
few duct manufacturers and installers have had their normal
duct construction and sealing methods tested. Except in extreme
cases where leakage is so great that system air quantities cannot
be achieved during commissioning, we simply do not know how
little, or how much, ductwork constructed to current industry
practice leaks. Some traditional approaches (Figure 3) are clearly
not leak tight. American research [11] measuring duct leakage
in existing buildings found rates between 0% to 30%, with most
between 10% and 20%. Since these American systems would
most likely have been constructed to SMACNA standards
which are very similar to the AS 4254 series, similar results
are likely in Australia.
50

Eco l i b r i u m M AY 2 0 1 3

Figure 3: Despite having been painted, this duct still shows potential
leakage points (arrowed).

3.1AS 4254.2:2012 Ductwork for air-handling

systems in buildings Rigid duct
As noted previously, previous editions of the AS 4254 series
recommended against leakage testing but AS 4254.2:2012
now requires testing (Clause 2.2.4). In summary, the standard
requires that each duct system of 3000 L/s or more be tested
to at least 1.25 times its operating pressure, and not leak more
than 5% of the design air quantity of the duct system. It is
notable that the standard does not require the whole of the
system to be tested. Instead it calls for type-testing of at least
10% of the system, including longitudinal seams, circumferential
joints, floor distribution, riser and plant room duct, and each
type of seam, joint and sealing construction. The standard
does not indicate whether the 10% relates to total duct length,
surface area or length to seams but overseas standards use duct
surface area. Designers and contractors should agree as to which
sections are to be tested having regard to design intent and
practicability on site.
Several points should be noted about these requirements.
Firstly, testing is done under static, rather than operating,
conditions and, as the test pressure is higher than the operating
pressure, the leakage rate in operation will be less. As the
actual leakage in operation cannot be calculated from the static
leakage, the approach recommended in [1] and [8] is to use the
arithmetic average of the pressure at the start and end of the
duct system. For example, consider a rigid supply duct system
with 250 Pa at the fan end and 40 Pa at the spigot to the last
flexible duct. For this system, the arithmetic average pressure
is 145 Pa. The test pressure to AS 4254.2:2012 for such a system
is 312 Pa (1.25 times 250 Pa). In this case, the ratio of leakage
rates, from equation (2) will be (145/312) 0.65 or about 61%. That
is, if the system passes the leak test with less than 5% leakage,
in operation leakage can be expected to be less than 3%. This
is an improvement on the design norm of 5% discussed above.
It is also a significant improvement on leakage rates reported in
overseas research [5] [6] and comparable with the value of 2.5%
assumed for a tight system in [6].
Secondly, type-testing of 10% of each system rather than whole
system should simplify the task and reduce the impact of leakage
testing on construction cost and site programing.

FORUM

This however raises a third issue: how much of the system should
be tested. Eurovent [8] uses a staged strategy in which the first
stage of testing uses 10% of the total surface area for round ducts
or 20% for rectangular. If the air leakage rate on test is greater
than the permitted rate, the test is repeated with double the
duct area. If it still fails, Eurovent requires that the whole of the
system be tested. This staged strategy compels the contractor to
fix most leaks, not just those in the test section. AS 4254.2:2012
does not include such a strategy and, since in Australia most
duct systems are rectangular, requires only half the amount
to be tested (10%) compared to Eurovent.
The fourth issue is that the permissible leakage in
AS 4254.2:2012 is expressed as a percentage of the design flow.
This criterion has the merits of being simple and directly
related to the effect the standard wants to achieve, reducing the
impact of total leakage on the system. However, it differs from
the practice in Europe and the US where permissible leakage
is expressed as a function of system pressure and duct surface
area. In 1985, SMACNA [12] concluded that this approach is
far superior to the arbitrary assignment of a percentage of fan
flow rate as a leakage criteria. The single percentage used in AS
4254.2:2012 produces anomalous results as can be illustrated by
reference to equation (2). Consider two systems each handling
3000 L/s. One has a duct surface area of 50 m and operates at
200 Pa, the other has a surface area of 150 m and operates at
500 Pa. For each system the permissible leakage is 5% of 3000
L/s or 150 L/s. Substituting these values into equation (2) gives

values for the constant, C (representing duct tightness)

of 0.096 for the first system and 0.018 for the second. That is,
the second system must be over five times tighter than the first
to meet the same leakage criteria. This is likely to create uncertainty
for contractors since they will be unable to confidently predict (and
hence price) what needs to be done to meet the AS 4254.2:2012
leakage requirements. They could perhaps mitigate this by using
the surface area of the system to determine the required leakage
class and sealing using overseas standards.
Finally, there is the question of what will and will not be tested.
Since AS 4254.2:2012 covers only rigid mild and stainless steel
sheet metal ducts, its testing requirements do not apply to:
Rigid ducts constructed from aluminium, PVC u and other
materials not covered by AS 4254.2:2012.
Flexible ducts, which are covered by AS 4254.1:2012.
Air handling units.
Non-duct components installed in the duct system such
as VAV boxes, active chilled beams, diffusers and grilles.
Builders work items such as ceiling plenums and risers.
Experience has shown that these can be very leaky, in part
because of the division of responsibility on site.
Although the standard does not include these, designers and
specifiers should consider including them in the parts to be tested,
having regard to the projects performance and energy requirements.

may 2 0 1 3 Eco l i b r i u m

51

FORUM

4. Conclusion and
recommendations

[3

Duct leakage can have a significant effect on HVAC system

performance, energy consumption and greenhouse gas emissions
in the form of CO2 from power stations. The leakage testing
requirements of AS 4254.2:2012 are an important step towards
addressing this problem and can be expected to improve the
quality of duct construction and installation in Australia. This
is an important first step but there is potential for expanding the
content of the standard:

[4] AIRAH; DA09 : Load estimation and psychrometrics.

Melbourne: Australian Institute of Refrigeration,
Air Conditioning and Heating, (1998).

1. At present leakage test criteria are divorced from the

design process. Ideally, the designer should determine the
permissible leakage rate (or Leakage Class, for simplicity),
and take this into account in selecting plant and include it in
the contract specification.
2. Research is required to correlate duct sealing methods to
achievable leakage rates. Interrelating these would give the
contractor some certainty that if the prescriptive sealing
methods are followed, the duct system would be likely to pass
the leakage test.
3. The present single value percentage leakage rate is not related
to duct pressure class, duct construction or duct surface
area. It would be preferable if the standard followed overseas
practice and adopted leakage criteria based on equation (2).
Pending a change to the standard, the designer could specify
leakage class and sealing based on overseas practice. This
would also increase designer and contractor certainty.
4. As presently written, AS 4254.2:2012 covers only testing
of rigid duct only but there are many other potential leak
sources including air handling plant, builders work items,
flexible ducts and terminal equipment. Leakage testing
could be extended to cover these. Despite the desirability
of reducing the leakage in builders work items, there are
significant practical problems in extending AS 4254.2:2012 to
cover their sealing.

5.References
[1] Gilberg, J.; The consequences of air leakage from ducts
on operating economy, Australian Refrigeration, Air
Conditioning, and Heating, Vol. 32, pp. 45-51, May (1978).
[2] Australian Standard: AS 4254:2002 : Ductwork for
air-handling systems in buildings. Sydney: Standards
Australia, (2002).

52

Eco l i b r i u m M AY 2 0 1 3

Australian Standard: AS 4254.2:2012 : Ductwork for airhandling systems in buildings Rigid duct. Sydney: Standards
Australia, (2012).

[5] Carri, F. R., Bossaer, A., Andersson, J. V., Wouters, P., and
Liddament, M. W.; Duct leakage in European buildings:
Status and perspectives, Energy and Buildings, Vol. 32,
pp. 235243, (2000).
[6] Wray, C. P. and Matson, N. E.: Duct leakage impacts on VAV
system performance in California large commercial buildings,
Berkeley: Lawrence Berkeley National Laboratory, (2003).
[7] Aydin, C. and Ozerdem, B.; Air leakage measurement and
analysis in duct systems, Energy and Buildings, Vol. 38,
pp. 207213, (2006).
[8] Eurovent Standard: Eurovent 2/2 : Air leakage rate in sheet
metal air distribution systems. Brussels: Eurovent Association,
(1996).
[9] HVCA Standard: DW/143 A practical guide to ductwork
leakage testing. London: Heating and Ventilating Contractors
Association, (2000).
[10] SMACNA Standard: HVAC air duct leakage test manual,
2nd edition. Washington: SMACNA, (2012).
[11] Fisk, W. J., Delp, W., Diamond, R., Dickerhoff, D., Levinson,
R., Modera, M., Nematollahi, M., and Wang, D.; Duct systems
in large commercial buildings: Physical characterization, air
leakage, and heat conduction gains, Energy and Buildings,
Vol. 32, pp. 109119, (2000).
[12] SMACNA Standard: HVAC air duct leakage test manual.
Washington: SMACNA, (1985).

About the author

Ian Wills is currently Engineering Services Editor
for NATSPEC, the National Building Specification.
He has worked in the HVAC industry in Australia
for over 40 years and was a director of D Rudd and
Partners (Australia) Pty Limited where he was
responsible for design and construction of building
services in a wide range of commercial, health,
education, defence and other projects.