Solar Energy 81 (2007) 2030

www.elsevier.com/locate/solener

Wind-induced natural ventilation analysis

Panagiota Karava *, Ted Stathopoulos, Andreas K. Athienitis
Centre for Building Studies, Department of Building, Civil and Environmental Engineering, Faculty of Engineering and Computer Science,
Concordia University, 1455 Maisonneuve W, Montreal, Que., Canada H3G 1M8
Received 17 October 2005; received in revised form 14 June 2006; accepted 15 June 2006
Available online 28 August 2006
Communicated by: Associate Editor Matheos Santamouris

Abstract
The paper presents and discusses internal pressure and discharge coecients for a building with wind-driven cross-ventilation caused
by sliding window openings on two adjacent walls. The study found that both coecients vary considerably with the opening area
(porosity of wall(s)) and the inlet to outlet ratio. Comparisons with previous work were also carried out. Experimental results verify
the unsteady pressure and velocity eld, particularly in the case of cross-ventilation with large opening areas. For such cases, a simulation
sensitivity analysis of wind-induced building ventilation conrms that airow rates vary considerably when dierent discharge coecient
values are used.
 2006 Elsevier Ltd. All rights reserved.
Keywords: Wind-driven ow; Cross-ventilation; Windward wall porosity; Inlet to outlet ratio; Internal pressure coecient; Inlet discharge coecient

1. Introduction
In a recent survey carried out by Canada Mortgage and
Housing Corporation (CMHC, 2004) to verify the use and
utility of ventilation systems in new Ontario houses it was
found that Over 90% of new homeowners in Ontario do
open windows, with over 40% opening windows for periods
of the winter. In mid-summer, almost 10% do not open
windows at all, which may indicate continuous use of air
conditioning systems. These houses would benet from
mid-summer ventilation to provide fresh air. The window
opening data collected (CMHC, 2004) are summarized in
Fig. 1. Many complaints reported from the occupants are
due to noise and drafts from mechanical systems. Field
experiments carried out in an urban canyon have shown
that appreciable ventilation rates can be obtained with natural ventilation in residential buildings, especially when
*

Corresponding author. Tel.: +1 514 848 2424 (7080); fax: +1 514 848
7965.
E-mail address: p_karava@alcor.concordia.ca (P. Karava).
0038-092X/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2006.06.013

cross-ventilation with two or more windows is used (Niachou et al., 2005). Chang et al. (2004) investigated the eects
of outdoor air conditions on hybrid air conditioning with
natural and mechanical ventilation in oce buildings.
The study found that natural ventilation at temperatures
lower than the indoor temperature eectively covered the
lower indoor task zone through negative buoyancy, which
enabled energy-saving air conditioning in the task zone.
Natural ventilation a traditional method to eectively
remove solar heat gains and improve the indoor thermal
environment appears as a cost-eective and attractive
alternative to mechanical ventilation. Cross-ventilation,
which is the design type of choice, is now attracting considerable attention again as a measure for sustainable design
of a building; recent developments involve integration of
cross-ventilation with HVAC and solar technologies such
as double facades or Building Integrated Photovoltaic Systems (BIPV). However, while natural ventilation is conceptually simple, its detailed design can be a challenge; the
ventilation performance involves the buildings form, its
surroundings and climate. This complexity, particularly

P. Karava et al. / Solar Energy 81 (2007) 2030

21

Nomenclature
A
A1
A2
a
Aopening
Awall
Ac
b
Cc
CD
CD,inlet
Cp1

opening area
inlet opening area
outlet opening area
A1/A2 = inlet to outlet ratio
area of the opening
area of the wall
cross section area of the vena contracta
1/4 for uniform distribution of cracks
contraction coecient
discharge coecient
inlet discharge coecient
external pressure coecient in opening 1

for wind-driven ventilation, can make dicult the development of a successful design. Mochida et al. (2005) and Lee
et al. (2005) pointed out that careful consideration of wind
ow around a building when deciding the placement of
window openings is very important to fully utilize the
potential of cross-ventilation and improve thermal comfort. However, dierent window congurations result in
dierent ventilation eectiveness, indoor air quality (IAQ)
and impact on comfort conditions in the occupied zone
(Heiselberg et al., 2001, 2002). Currently, codes of practice
or ventilation standards do not provide any selection criteria for dierent window types (e.g., sliding, top, side and
bottom hung windows, windows with louvers) or their
placement on the facade for adequate ventilation.
Formulating guidelines for the design of ventilation
systems is indeed signicant. This includes selection of
appropriate window type and its location on the facade
aiming to enhance ventilation or summer cooling
performance. This paper investigates the main parameters
aecting natural ventilation design, i.e., number/type of
windows or openings, size and location on facade for
wind-driven cross-ventilation and determines appropriate
discharge coecient values. The unsteady pressure and
velocity eld in the vicinity of the openings are investigated

80
Never

Percent of season

70

Monthly

60

Weekly

Daily < 1 h

40

external pressure coecient in opening 2

internal pressure coecient
pressure coecient on windward facade
velocity coecient
pressure dierence across the opening
airow through an opening
local velocity in the opening
reference wind speed at the buildings height
air velocity in the vena contracta
theoretical air velocity
air density

and the validity of the orice equation for cross-ventilation

is examined. Results of a series of experiments carried out
in a Boundary Layer Wind Tunnel (BLWT) for the evaluation of the internal pressure coecient in a building with
cross-ventilation are provided and discharge coecient values for dierent opening congurations with dierent inlet
to outlet ratio are presented and compared with those of
previous studies. A sensitivity analysis by means of simulation is also performed in order to quantify the impact of
discharge coecients on airow through inlet openings
for dierent congurations.
2. Basic equations
Design of wind-driven or natural ventilation is usually
carried out by using the orice equation, which is based
on Bernoullis assumption of steady incompressible ow,
and requires that the discharge coecient, CD, and the
internal pressure coecient, Cpin, are known. Applying
the Bernoulli equation on a horizontal streamline between
a point in front of the building opening with stagnant air
and pressure, Po, and the vena contracta (minimum cross
section area, Ac, of the ow with parallel stream lines, uniform velocity and static pressure equal to the surrounding
air pressure) the following (Bernoulli) equation is derived
for the theoretical air velocity:

0:5
2DP
V th
1
q
The air ow rate through an opening is:

0:5
2DP
Q Ac  V c C c  A  C v  V th C D  A 
q

3*Weekly

50

Cp2
Cpin
Cpw
Cv
DP
Q
u
V
Vc
Vth
q

Daily extended

30

Hence, CD is given by the following equation:

20

CD Cc  Cv

10

where

0
Spring or Fall

Summer

Winter

Fig. 1. Window opening data (CMHC, 2004).

Ac = cross section area of the vena contracta (Ac = Cc A);

Vc = air velocity in the vena contracta (Vc = Cv Vth);

22

P. Karava et al. / Solar Energy 81 (2007) 2030

A = opening area;
Cc = contraction coecient (61);
Cv = velocity coecient depending on the friction conditions (61);
DP = pressure dierence across the opening;
q = air density.
For low-rise buildings, the airow through ventilation
openings (i.e., openable windows) is mainly wind-driven,
especially during summer. For wind-driven ventilation,
the discharge coecient for an inlet can be determined by:
C D;inlet

AV 

Q
u
p
p
C pw  C pin V  C pw  C pin

in which
Q = airow through the opening;
u = local velocity in the opening;
V = reference wind speed at the buildings height;
Cpw = pressure coecient on windward facade;
Cpin = internal pressure coecient, determined for a building with two openings by:
C p2 a2  C p1
large openings
5
1 a2
e:g:; windows; doors turbulent flow
C 2 b  C p1
small openings
6
C pin
1b
e:g:; cracks laminar flow

C pin

Cp1 = external pressure coecient in opening 1;

Cp2 = external pressure coecient in opening 2;
a = A1/A2; and b = 1/4 for uniform distribution of cracks.
Generally, Cpin depends on the external pressure distribution, terrain, shape, area and distribution of openings
on the facade. More information on internal pressure coefcients for buildings with openings can be found in Liu
(1991), Stathopoulos et al. (1979) and Wu et al. (1998).
Current literature provides useful information for the
understanding of dierent aspects of cross-ventilation phenomena but there are still doubts about the accuracy of the
orice equation when applied to calculation of windinduced ow through openings. For wind-driven ventilation, the orice equation is valid under the assumption that
the ow is fully-developed turbulent, the pressure distribution on the building envelope is not aected by the presence
of openings (sealed body assumption), the kinetic energy is
dissipated downstream of the inlet opening and the pressure drop across the inow and outow opening is equal
to the static pressure dierence. However, for cross-ventilation with both inlets and outlets, a part of the turbulent
kinetic energy is preserved and directed outside without
interior dissipation (Murakami et al., 1991; Kato et al.,
1992; Kato, 2004). Recent research (e.g., Heiselberg

et al., 2002; Sandberg, 2004; Etheridge, 2004; Kato, 2004;

Axley and Chung, 2005) into the behavior of ow through
openings has highlighted the need to improve the mathematical formulation of this phenomenon. Dierent alternative theories have been proposed by Murakami et al.
(1991), Kato (2004), Hiyama and Kato (2005), Sandberg
(2004), Ohba et al. (2004), Kurabuchi et al. (2004, 2005),
and Axley and Chung (2005) but none can be established
as a potential replacement of the conventional orice equation. Karava et al. (2004) pointed out that modelling
factors such as scaling, upstream ow conditions and internal partitions should also be considered in wind tunnel
experiments of cross-ventilation. Prediction of the airow
through large openings (windows partly open) remains a
signicant source of uncertainty in building simulation
(Furbringer et al., 1999; Heiselberg, 2004a,b).
3. Description of the model and experimental set up
A 1:12 gable roof-sloped building model of rectangular
plan view 15.3 9.8 cm was tested with an eave height of
3 cm. This corresponds to a building 61 39 m and 12 m
high according to the 1:400 geometric scale in the BLWT
of Concordia University. The model provides variable
side-wall and windward wall openings and background
leakage of 0% and 0.5%. The opening area is expressed in
terms of wall porosity and is calculated by the following
equation:
Wall porosity

Aopening
Awall

The background leakage was achieved by a series of holes

which could be left open or closed. Five pairs (one for each
facade and roof) of closely-located internal and external
pressure taps have been selected for the measurements, as
indicated in Fig. 2. Plastic tubes connect each tap with a
Honeywell 163 PC pressure transducer. More details on
the building model can be found in Wu et al. (1998). A
standard wind prole over open country terrain was simulated with a power-law exponent equal to 0.15 and turbulence intensity at the eave height equal to 22%. In this
study only simple rectangular openings of the same height
(1.7 cm) were considered. Dierences in opening areas were
induced by dierent opening width (sliding windows). The
wall thickness is equal to 3 mm. A single-sided ventilation
conguration was considered rst for comparison proposes; experiments were performed for 0% and 0.5% background leakage. This was followed by cross-ventilation
experiments with 0.5% background leakage. Congurations with equal inlet and outlet opening area (A1 = A2)
were tested as well as congurations with inlet to outlet
ratio smaller or larger than 1. Fig. 2 shows the opening
congurations. The two rectangular openings were located
in the middle of long wall (windward wall) and short wall
(side wall). Measurements were carried out for a wind
angle, h, equal to 0. Congurations and range of variables
considered in this study are summarized in Table 1.

P. Karava et al. / Solar Energy 81 (2007) 2030

23

Fig. 2. Exploded view of building model with pressure tap locations and opening congurations.

Table 1
Study considerations and range of variables
Ventilation strategy

Cross-ventilation with openings on two

adjacent walls

Building dimensions

61 39 12 (eave height) m
15.3 9.8 3 (eave height) cm
1:12 (gabled)
1:400
Open (a = 0.15, T.I. = 22% at
the eave height)
0 (perpendicular to long wall)
Sliding
0160 m2
010 cm2
022%
08
0% and 0.5%

Roof slope
Scaling
Upstream terrain
Wind direction
Window type
Opening area
Windward wall porosity
Ainlet/Aoutlet
Building envelope leakage

4. Results and discussion

The impact of a windward wall opening (single-sided

ventilation) on internal pressure was investigated for 0%
and 0.5% background leakage and results are presented
in Fig. 3 for dierent opening area or windward wall porosity (Ainlet/Awall). The opening is located in the middle of the
long facade. The internal pressure was measured at dierent internal taps and it was found to be uniform, as also
previously reported by Wu et al. (1998). The experimental
data is compared with the values obtained by Eq. (5) for
0% background leakage. For the case of a single opening,
Eq. (5) reduces to Cpin = Cp1, which is equal to 0.67.
Fig. 3 shows good agreement between the experimental
results and the theoretical values for 0% leakage. In addition, data obtained for 0.5% background leakage shows a
similar trend with that observed in previous studies, e.g.
Wu et al. (1998); however, application of Eq. (5) will be
arguable in this case due to the undetermined character
of the ow.

This section presents and discusses experimental results

regarding internal pressure coecients and discharge coefcients for dierent congurations.

0.8
0.6
0.4

The measured mean values of the external pressure coefcients (reference height = building height) are 0.67 for the
windward wall, 0.36 for the side wall, 0.25 for the leeward wall and 0.4 and 0.74 on taps 7 and 9 of the roof
(see Fig. 2). The measured mean internal pressure coecient for 0.5% background leakage (without openings) is
0.36 which is slightly dierent from the theoretical value,
0.23, obtained by Eq. (6) using Cp1 as the representative
of the positive external pressure and Cp2 as the area-averaged pressure on the rest of the building envelope.

Cpin

4.1. Internal pressure coecients

0.2
0
0

10

15

20

25

-0.2

Windward wall porosity (Ainlet/Awall) (%)

-0.4

BLWT, 0% leakage
BLWT, 0.5% leakage
Theoretical, 0% leakage

30

-0.6

Fig. 3. Internal pressure coecients for single-sided ventilation and

dierent windward wall porosity.

24

P. Karava et al. / Solar Energy 81 (2007) 2030

The impact of a windward and a side-wall opening

(cross-ventilation) of the same area (A1 = A2) on internal
pressure was investigated for 0.5% background leakage.
In the present study A1 = A2 (i.e., equal inlet and outlet
opening) does not correspond to equal windward (Ainlet/
Awall) and side wall porosity (Aoutlet/Awall), i.e., for the
same opening area, the side wall porosity is higher since
the side wall has a smaller facade area compared to the
windward wall. However, in other studies say for a cubical building A1 = A2 corresponds to equal inlet and outlet
porosity. The openings were located in the middle of the
long and short walls see Fig. 2. Measurements were carried out for opening areas up to 10.2 cm2 (or 22% windward wall porosity), which is typically the range in
naturally ventilated houses. The external pressure distribution was monitored and found unaected by the presence
of openings on the facade (sealed body assumption). The
internal pressure was measured on taps 2, 4, 6, 8 and 10.
Fig. 4 presents the mean value and standard deviation of
the internal pressure coecient as a function of the windward wall porosity. Wind tunnel results by Murakami
et al. (1991) and Sawachi et al. (2004), as well as CFD
results by Kurabuchi et al. (2004) and Hu et al. (2005)
are included for comparison purposes. The Cpin values
obtained by using Eq. (5) are also presented. For A1/
A2 = 1, Eq. (5) reduces to Cpin = (Cp1 + Cp2)/2. It should
be noted that the background leakage is not considered
when using Eq. (5). The experimental results show that
the average Cpin increases with the increase of the windward wall porosity, although Eq. (5) provides a constant
value. The standard deviation of the internal pressure coefcient is higher for opening area larger than about 5 cm2
(which corresponds to approximately 10% windward wall
porosity), indicating a substantial spatial variation of the
internal pressure. This is probably due to a virtual ow
tube that connects the inlet and the outlet see also Murakami et al. (1991). In fact, small Cpin values were measured

0.8
Present study

Previous studies

0.6

0.4

Cpin

0.2

0
0
-0.2

-0.4

10

15

20

25

30

Windward wall porosity (Ainlet /Awall) (%)

BLWT, 0.5% leakage
Theoretical, 0% leakage
Murakami et al. (1991), expt.
Sawachi et al. (2004), expt.
Kurabuchi et al. (2004), LES
Hu et al. (2005), RANS (SST k-w)

-0.6

Fig. 4. Internal pressure coecients for cross-ventilation, A1 = A2 (a = 1)

and 0.5% background leakage.

in taps 2 and 10 that may be aected by the ow tube, while

high Cpin values were recorded in taps 4, 6 and 8 that are
farther from the ow tube. These variations should be considered in the selection of measurement points for future
experiments. The non-uniformity of Cpin distribution in
the building was not observed in the case of single-sided
ventilation and it is not taken into consideration by the theory Eq. (5). In the study by Murakami et al. (1991) rectangular building models were tested with dierent wall
thickness and openings located in the middle of the long
or short walls (windward wall inlets and leeward wall outlets). For the building models and congurations considered in Fig. 4, the windward wall pressure coecient is in
the range of 0.570.67 and the leeward wall coecient is
ranging from 0.16 to 0.1. Recall that in the present
study the external pressure coecients are 0.67 for the
windward wall (inlet) and 0.36 for the side wall (outlet).
As shown in Fig. 4, the internal pressure coecient in the
study by Murakami et al. (1991) is in the range of 0.33
to 0.18 for dierent congurations with wall porosity in
the range of 2.820% (or opening areas ranging from 5 to
36 cm2). Examination of Fig. 4 reveals that some of the values given by Murakami et al. (1991) are outside the range
dened by the results of the other studies. However, it
should be noted that the internal pressure was measured
at only two points on the oor, in front of the inlet and
the outlet, which might not be sucient and additional
measurement points should have been considered. Results
of the present study show that for cross-ventilation the
internal pressure is uctuating particularly for high windward wall porosity.
The internal pressure coecient was also investigated
for inlet opening area equal to 2.6 cm2 (A1 = 2.6 cm2 or
5% windward wall porosity), 0.5% background leakage
and dierent inlet to outlet ratio (outlet opening area, A2,
varied from 0.9 to 12 cm2). The experiments were repeated
for inlet opening area, A1, equal to 5.2 cm2 (or 10% windward wall porosity). The internal pressure was measured
on taps 2, 4, 6, 8 and 10. The mean and standard deviation
of the internal pressure coecient as a function of inlet to
outlet ratio (A1/A2) for A1 = 2.6 cm2 (or 5% windward wall
porosity) is shown in Fig. 5. The results are compared with
wind tunnel data by Murakami et al. (1991) for the case of
a side wall outlet and a leeward wall outlet. The Cpin values
by using Eq. (5) are also presented. The experimental
results indicate that the average Cpin increases with the
increase of inlet to outlet ratio. The Cpin values predicted
by Eq. (5) are overestimated compared to the experimental
data, particularly for higher inlet to outlet ratios. This
might be due to the impact of the background porosity that
is more important for small opening areas and high A1/A2
ratios. Generally, there is relatively good agreement
between the results of the present study and those by
Murakami et al. (1991) particularly for the model with
the side wall outlet. Comparison of data with a 2.6 cm2
opening (5% windward wall porosity) and opening of
5.2 cm2 (10% windward wall porosity) shows that there

P. Karava et al. / Solar Energy 81 (2007) 2030

0.8
0.6
0.4

Cpin

0.2
0
0.1

10
A1/A2

-0.2
BLWT, 5% leakage, 10% windward wall porosity
Theoretical, 0% leakage
Murakami et al. (1991), side wall outlet
Murakami et al. (1991), leeward wall outlet

-0.4
-0.6

Fig. 5. Internal pressure coecients for cross-ventilation and dierent

inlet to outlet ratios (A1 = 2.6 cm2 or 5% windward wall porosity Ainlet/
Awall).

are no substantial dierences among average Cpin values

for the same A1/A2 ratio but dierent A1 and A2, as
opposed to Cpin values for congurations with A1 = A2
(Fig. 4). However, the internal pressure coecient varies
considerably (from 0.26 to 0.35) for congurations with
A1/A2 > 1 or A1/A2 < 1 compared to congurations with
A1/A2 = 1 (from 0.05 to 0.19, see Fig. 4). This might be
important to be considered in natural ventilation design.
Fig. 6 shows the variation of the standard deviation of
Cpin with the inlet to outlet ratio for A1 = 2.6 cm2 (5%
windward wall porosity) and A1 = 5.2 cm2 (10% windward
wall porosity). For A1/A2 < 1 (i.e., large outlet area, A2) the
standard deviation of Cpin is higher due to dierences
among the various internal pressure taps; these dierences
are even higher for larger inlet area, A1 = 5.2 cm2 (or 10%
windward wall porosity). This internal pressure coecient
variation is attributed to the ow tube connecting the inlet
and the outlet.

25

Fig. 7 shows the variation of the pressure drop across an

inlet (DCp = CpwCpin) for single-sided and cross-ventilation (A1 = A2). For small windward wall porosities (up to
3%) almost the same pressure drop is observed for singlesided and cross-ventilation; however, for higher wall porosities the pressure drop for cross-ventilation is larger
compared to single-sided ventilation. In a sealed building
with a windward opening, when the internal pressure overshoots the external pressure, the wind ow is reversed out
of the opening, which brings the internal pressure down.
Due to inertia eect, the pressure could drop down to a
level even lower than the external pressure at the opening
before bouncing back for another cycle of oscillation
(Wu et al., 1998). The existence of a leeward opening or
building envelope leakage reduces the ow reversal and
increases the ow rate into the inlet (larger pressure drop
across the inlet).
Fig. 8 shows the variation of the pressure drop across
the inlet (DCp = CpwCpin) and the outlet (DCp = Cpin
Cpside-wall) as a function of the inlet to outlet ratio for
A1 = 2.6 m2 (or 5% windward wall porosity). The DCp values by using Eq. (5) and measured values for the windward
(0.67) and side wall (0.36) pressure coecients are also
presented. For A1 > A2 most of the pressure drop occurs
across the outlet, while for A1 < A2 higher pressure drop
occurs across the inlet. Similar results were obtained for
A1 = 5.2 cm2 (or 10% windward wall porosity), for which
almost the same DCp observed across the inlet and the outlet when A1 = A2. However, for A1 = A2 = 2.6 cm2 the
pressure drop is somewhat higher at the inlet. Similar
trends were observed in the wind tunnel study by Sandberg
(2004), although the absolute DCp values were dierent due
to dierences in experimental set-up, building and opening
congurations used.
The internal pressure coecient variation with windward wall porosity seems to be more pronounced for

1.4

0.2
Windward wall porosity = 5%
Windward wall porosity = 10%

Cross ventilation, BLWT

Single-sided ventilation, BLWT

1.2

0.16

Cp

0.12

Cp

0.8

STDEV

0.6

0.08

0.4

0.04
0.2

0
0.1

1
A1/A2

10

Fig. 6. Standard deviation of the internal pressure coecient as a function

of A1/A2 for 5% and 10% windward wall porosity (Ainlet/Awall).

10
20
30
Windward wall porosity (Ainlet/Awall) (%)

40

Fig. 7. Pressure dierence coecient across the inlet as a function of the

windward wall porosity (Ainlet/Awall) for equal inlet and outlet area.

26

P. Karava et al. / Solar Energy 81 (2007) 2030

BLWT, inlet
Theoretical, inlet
Sandberg (2004), inlet

BLWT, outlet
Theoretical, outlet
Sandberg (2004), outlet

1.8
Previous studies

Present study

1.5
1.2

Cp

0.9
0.6
0.3
0
0.01

0.1

A1/A2

10

100

-0.3

Fig. 8. Pressure dierence across the inlet and outlet as a function of inlet
to outlet ratio (A1 = 2.6 cm2 or 5% windward wall porosity Ainlet/Awall).

congurations with unequal inlet to outlet ratios compared

to congurations with equal inlet and outlet opening areas,
as Figs. 4 and 5 indicate. Also there is better agreement
between the present study and the theoretical prediction
or other literature sources for unequal inlet to outlet ratios.
This might be due to the fact that in the present study
A1 = A2 (i.e., equal inlet and outlet opening) does not correspond to equal windward (Ainlet/Awall) and side wall
porosity (Aoutlet/Awall); however, in other studies say for
a cubical building A1 = A2 corresponds to equal inlet
and outlet porosity. Also in the present study the outlet
is located on the side wall as opposed to leeward wall outlets in most of the other studies recall that leeward and
side wall pressure coecients are dierent. In addition,
likely reasons for the dierences among various studies
include modelling factors such as scaling, upstream ow
conditions in wind tunnel experiments and/or possible
errors contribute to discrepancies among these results.
The best way to identify and explain these dierences
among the various studies might be to conduct eld measurements but, as far as cross-ventilation is concerned,
Katayama et al. (1992) did not monitor internal pressures,
whereas Mochida et al. (2005) mainly focused on the
indoor thermal comfort impact of cross-ventilation.
4.2. Inlet discharge coecient
4.2.1. Theoretical background
The discharge coecient (Eq. (4)) is determined by
applying the Bernoulli equation on a horizontal stream line
between a point in front of the building opening with stagnant air and pressure and the vena contracta (minimum
cross section area of the ow with parallel stream lines, uniform velocity and static pressure equal to the surrounding
air pressure) see also Andersen (1996) and Karava et al.
(2004). Hence, in case of an opening separating two regions
with still air boundary conditions, the discharge coecient

can be assumed to be constant ranging from 0.6 to 1,

depending primarily on the shape (geometry) of the opening. Bernoulli equation assumes two points that are not
aected by the opening and the pressure is uniform. Thus,
selection of the measurement points should not be an issue.
For wind-driven cross-ventilation the velocity and pressure elds at the inlet are unsteady, creating diculties with
the selection of CD and, more fundamentally, its denition
(Etheridge, 2004). Experimental results conrm the previous statement, particularly in the case of cross-ventilation
with large opening areas (see Figs. 46). Thus, care should
be taken in the selection of the measurement point. For
wind-driven cross-ventilation, the discharge coecient
depends on the geometry of the opening but also the external (surrounding and building) and the internal ow eld
(background leakage, leeward/side wall openings). In fact,
for large opening area, there is signicant air movement in
the room due to the dimensions of ow tube and the ow
might no longer be considered as pressure-driven. Therefore, discharge coecient values may be outside the standard range of 0.61.
4.2.2. Experimental
In order to evaluate the inlet discharge coecient experimentally, the velocity ratio u/V see Eq. (4) was evaluated using a hot lm anemometer. The results are presented
in Table 2 along with data from other similar wind tunnel
studies reported in the literature. The ratio u/V is about
0.50.63 in the present study and it varies between 0.45
and 1 in other literature sources. In the present study, the
velocity in the opening is assumed equal to the velocity in
front of the opening, i.e., for two inlet openings with dierent areas the velocity ratio (u/V) is assumed to be the same.
Consequently, the dierence in the discharge coecient
between the two openings is due to Cpin variation with
the opening area.
Fig. 9 shows the variation of the inlet discharge coecient with the windward wall porosity for cross-ventilation
with two rectangular openings of the same area (A1 = A2)
located on the windward and side wall of the building
model with 0.5% background leakage. Wind tunnel data
by Murakami et al. (1991) (for cross-ventilation with openings located in the long facade of the building), Jensen et al.
(2002) (for the case of two circular openings placed on top
and bottom of a cylinder and wind direction along the axis
of the cylinder cross-ventilation), as well as CFD results
by Hu et al. (2005) are included. Most previous studies conTable 2
Velocity ratios (u/V) for dierent studies
Reference

u/V

Vref (m/s)

Present study
Hu et al. (2005)
Etheridge (2004)
Sawachi et al. (2004)
Murakami et al. (1991)

0.50.63
0.45
0.60
0.50
0.641.0

7.2
7.0
4.0
3.0
Not reported

P. Karava et al. / Solar Energy 81 (2007) 2030

Inlet CD

0.8

0.6

0.4

BLWT, u/V = 0.63

BLWT, u/V = 0.5
Murakami et al (1991), u/V = 0.64-1
Jensen et al (2002)
Sawachi et al (2004), u/V = 0.5
Hu et al (2005), u/V = 0.45

0.2

0
0

5
10
15
20
25
Windward wall porosity (Ainlet/Awall) (%)

30

Fig. 9. Inlet CD as a function of the windward wall porosity (Ainlet/Awall)

for A1/A2 = 1.

clude that the discharge coecient increases with the

increase of the opening area. However, Sawachi et al.
(2004) reported that CD is not aected by the opening area;
while Heiselberg et al. (2002) found that CD might
decrease, increase or remain almost constant depending
on the conguration, regardless of opening area. Karava
et al. (2004) reviewed the current literature and concluded
that no clear trend can be established. The results of the
present study show that for windward wall porosity from
2% to 22% the inlet discharge coecient varies from 0.74
to 0.9 for u/V = 0.63 and from 0.6 to 0.71 for u/V = 0.5.
Therefore, CD = 0.65 might be a good approximation for
velocity ratio u/V = 0.5. For the opening and building congurations considered in Fig. 9, inlet discharge coecient
values for congurations with large wall porosities are
higher than the typical values given in textbooks, i.e.,
0.610.65 for sharp-edged orices, Etheridge and Sandberg
(1996). This shows the decrease of ow resistance for crossventilation, particularly in the case of a ow tube connecting the inlet and the outlet, and indicates the need for
appropriate discharge coecients to account for this phenomenon. It is likely that dierent CD values observed in
the various studies may be due to dierent external ow
conditions, opening types and building congurations considered, which in turn may explain the dierences in internal pressure coecient values described in Section 4.1. Also
a 0.5% background leakage was considered in the present
study, as opposed to impermeable models used previously.
Stathopoulos et al. (1979), Aynsley (1999) and Heiselberg
et al. (2001, 2002) pointed out that building envelope leakage could have a considerable impact on pressure distribution and discharge coecient values, particularly for
congurations with small opening areas. However, further
experimental work is required considering dierent building and opening congurations, upstream ow conditions,
internal ows, scaling and wind directions other than 0
before any generalization is to be made. A more precise
method might be to consider velocity eld measurements,

Particle Image Velocimetry (PIV) technique, as opposed

to point measurement techniques in order to map properly
the unsteady pressure and velocity ow eld in the case of
cross-ventilation.
Fig. 10 shows the variation of inlet discharge coecient
with the inlet to outlet ratio (A1/A2) for A1 = 2.6 cm2 (5%
windward wall porosity). Wind tunnel data by Murakami
et al. (1991) were also considered for the case of crossventilation with a windward wall inlet and a side wall outlet
(placed near the corner with the windward wall). Generally, the discharge coecient varies from 0.65 to 1.15 for
u/V equal to 0.63 and from 0.52 to 0.9 for u/V equal to
0.5. For A1/A2 < 1 there is smaller variation of CD compared to that for A1/A2 > 1. A similar observation was
made by Sandberg (2004) for the catchment area Ac,
dened as Q/u for relatively large opening area. Unconventional discharge coecients (i.e., CD > 1) are observed
for A1/A2  1. This can be justied by the unsteady external and internal ow eld in the case of cross-ventilation
particularly for large opening areas, as previously
discussed.
4.2.3. Sensitivity analysis: case study
Considering the variation of discharge coecients, it is
important to quantify the impact of their values on airow
through inlet openings and identify the possible dierences
between various congurations. A sensitivity analysis was
performed by means of simulation. An airow network
model was developed using ESP-r (ESRU, 2002) for a typical oor of a 22-story high residential building located in
Ottawa. Extensive experimental and modeling work was
carried out for this building in order to investigate the airow patterns, air leakage and related energy consumption.
A detailed description of the building, experimental and
simulation results (including validation) can be found in
Reardon et al. (2003) and Karava et al. (2006). A plan of
a typical oor is illustrated in Fig. 11. For the present work
1.2

CD = 1
1

0.8
Inlet CD

1.2

27

CD = 0.65
0.6

0.4

0.2

0
0.1

BLWT, u/V = 0.63

BLWT, u/V = 0.5
Murakami et al. (1991), u/V = 0.64

1
A1/A2

10

Fig. 10. Inlet CD as a function of inlet to outlet ratio for A1 = 2.6 cm2 or
5% windward wall porosity (Ainlet/Awall).

28

P. Karava et al. / Solar Energy 81 (2007) 2030

Fig. 11. Typical oor plan of case study building.

a single oor, namely the 4th oor for which experimental

data are available, is simulated during summer assuming
wind-driven ow. Weather data (wind speed and direction)
measured on site, as well as measured values for the leakage characteristics (ow exponent and ow coecient) of
the exterior walls were used as inputs. Initially, only the
air leakage was modeled because experimental data were
available for validation. The measured 1 h (between 12:00
and 13:00 that the tracer gas test was performed) mean
envelope pressure across the suite 409 (see Fig. 11) is equal
to 3.4 Pa and the average air exchange rate 0.85 ach. The
calculated values are 3.8 Pa and 0.80 ach, respectively,
i.e., very close to the measured data. Consequently, the
external pressure coecient distribution assumed for the
calculation was used for the evaluation of the impact of
CD on the airow through inlet openings.
A window was considered to be open on the West facade
of the 412 suite (see Fig. 11). Air enters the suite through
the window opening (inlet opening, area A1) and it is
exhausted to the corridor through the apartment door
undercut (outlet opening, area A2). The discharge coecient of the door undercut evaluated to be equal to 0.5
using experimental data for the pressure drop and the airow through the opening. The opening congurations considered are presented in Table 3. The simulations run for
29 h (July 1213, 2002) with a time step equal to 4 min.
During this period of time the wind was almost perpendicular to the West facade. Table 3 shows the mean and
STDEV airow rate for the inlet opening (window). The
airow rate varies considerably when dierent CD values
are used especially for congurations with large opening
area and Ainlet/Aoutlet = 1.0, while this variation is less signicant for congurations with A1/A2 > 1. In particular,
with the opening area or the discharge coecient increased,
the pressure drop across the opening decreased and the
resulting airow is less than what is expected if this

Table 3
Mean and STDEV air change rate for dierent opening congurations
(July 1213, 2002)
A1 (m2)

A2 (m2)

A1/A2

Inlet CD

Outlet CD

Air change rate

(ach)
Mean

STDEV

0.02
0.02
0.1
0.1
0.1
0.1

0.02
0.02
0.02
0.02
0.1
0.1

1
1
5
5
1
1

0.5
1
0.5
1
0.5
1

0.5
0.5
0.5
0.5
0.5
1

0.38
0.54
0.60
0.62
1.99
3.76

0.18
0.26
0.29
0.30
1.02
2.03

decrease of pressure drop is not taken into account. Therefore, opening area, discharge coecient, airow and pressure drop are coupled and these interactions may have
important implications on control of openings. Although
these are data for a test building, and results cannot be generalized, it is clear that signicant dierences stem from
inaccurate determination of discharge coecients. A sensitivity analysis in terms of internal pressure coecient is
partially covered since the latter aects the evaluation of
CD. However, this work should be extended to validate
(using experimental results) predictions in airow modeling
with respect to Cpin since experimental results show that the
internal pressure coecient variation is not predicted accurately by the theory particularly for A1 = A2.

5. Concluding remarks
The paper presents and compares with other literature
sources the results of a series of experiments carried out
in a Boundary Layer Wind Tunnel (BLWT) for the evaluation of the internal pressure coecient and the discharge
coecient in a building with cross-ventilation caused by

P. Karava et al. / Solar Energy 81 (2007) 2030

variable (sliding) window openings on two adjacent walls.

Experimental results conrm the unsteady pressure and
velocity eld, particularly in the case of cross-ventilation
with large opening areas. The study found that the internal
pressure coecient for cross-ventilation varies considerably with the wall porosity and inlet to outlet ratio and it
is not uniform for windward wall porosity higher that
10%. This non-uniformity of Cpin spatial distribution in
the room is not observed in the case of single-sided ventilation and it is not predicted by the theory Eq. (5). However, this nding is important to be considered in the
selection of measurement points.
The inlet discharge coecient varies with the opening
area (or wall porosity) and inlet to outlet ratio; this variation is more pronounced for A1/A2 > 1. A simulation-based
sensitivity analysis was performed in order to quantify the
impact of discharge coecients on airow through inlet
openings and identify possible dierences between various
congurations. The study conrmed that airow rates vary
considerably when dierent CD values are used, particularly for congurations with large opening area and
Ainlet/Aoutlet = 1.0. Regardless, the results of the study are
particularly important for natural ventilation design. Opening area, discharge coecient, airow and pressure drop are
coupled and these interactions might have important implications on control of openings. However, further experimental work is required considering dierent building and
opening congurations, as well as more precise measurement techniques before a denite generalization is made.
Acknowledgements
This study is carried out with partial support from the
Center for Energy Technology of Natural Resources Canada (CANMET) through a University Research Network
Grant for which the authors are grateful. The authors
would also like to thank Dr. James Reardon of the Institute for Research in Construction (IRC) for the provision
of the experimental data for the high-rise residential building case study.
References
Andersen, K.T. 1996. Inlet and outlet coecients: a theoretical analysis.
In: Proceedings of Roomvent 1, pp. 379390.
Axley, J.W., Chung, H., 2005. POWBAM0 mechanical power balances for
multi-zone building airow analysis. International Journal of Ventilation 4 (2), 95112.
Aynsley, R. 1999. Unresolved issues in natural ventilation for thermal
comfort. In: Per Heiselberg (Ed.), Proceedings of 1st International One
Day Forum on Natural and Hybrid Ventilation, Sydney, Australia,
International Energy Agency Annex 35 project Technical Paper, IN
Annex 35 CD, Aalborg University, Denmark.
Chang, H., Kato, S., Chikamoto, T., 2004. Eects of outdoor air
conditions on hybrid air conditioning based on task/ambient strategy
with natural and mechanical ventilation in oce buildings. Building
and Environment 39, 153164.
CMHC 2004. Analysis of ventilation system performance in new Ontario
houses. Technical Series 04117.

29

ESRU, 2002. The ESP-r system for building energy simulationUser Guide
Version 10 Series. University of Strathclyde, Glasgow, UK.
Etheridge, D.W., 2004. Natural ventilation through large openings
measurements at model scale and envelope theory. International
Journal of Ventilation 2 (4), 325342.
Etheridge, D., Sandberg, M., 1996. Building ventilation: Theory and
measurement. John Willey & Sons, London.
Furbringer, J.M., Roulet, C.A., Borchiellini, R., 1999. An overview of the
evaluation activities of IEA ECBCS Annex 23. Energy and Buildings
30, 1933.
Heiselberg, P. 2004a. Building integrated ventilation systems-modelling
and design challenges. In: Proceedings of CIB World Congress,
Toronto, Canada.
Heiselberg, P., 2004b. Natural ventilation design. International Journal of
Ventilation 2 (4), 295312.
Heiselberg, P., Svidt, K., Nielsen, P.V., 2001. Characteristics of airow
from open windows. Building and Environment 36, 859869.
Heiselberg, P., Bjorn, E., Nielsen, P.V., 2002. Impact of open windows on
room air ow and thermal comfort. International Journal of Ventilation 1 (2), 91100.
Hiyama, K., Kato, S. 2005. Study on new prediction procedure of crossventilation performance based on power balance. In: Choi, C.K., Kim,
Y.D., Kwak, H.G. (Eds.), Proceedings of 6th Asian-Pacic Conference
on Wind Engineering (APCWE VI), Seoul, Korea, September 1214,
pp. 14851495.
Hu, H.C., Kurabuchi, T., Ohba, M. 2005. Numerical study of windinduced ventilation with the local dynamic similarity model. In:
Naprstek, J., Fischer, C. (Eds.), Proceedings of 4th European &
African Wind Engineering Conference (EACWE4), Prague, July 11
15.
Jensen, J.T., Sandberg, M., Heiselberg, P., Nielsen, P.V., 2002. Wind
driven cross-ow analyzed as a catchment problem and as a pressure
driven ow. International Journal of Ventilation, HybVent-Hybrid
Ventilation Special Edition 1, 88101.
Karava, P., Stathopoulos, T., Athienitis, A.K., 2004. Wind-driven ow
through openings: A Review of discharge coecients. International
Journal of Ventilation 3 (3), 255266.
Karava, P. Athienitis, A.K., Stathopoulos, T. Reardon, J.T. 2006.
Modelling of indoor-outdoor air-exchange in a multi-unit residential
building. Submitted. In: Proceedings of International Building Physics
Conference, Montreal, August 2731.
Katayama, T., Tsutsumi, J., Ishii, A., 1992. Full-scale measurements and
wind tunnel tests on cross-ventilation. Journal of Wind Engineering
and Industrial Aerodynamics 4144, 25532562.
Kato, S., 2004. Flow network model based on power balance as applied to
cross ventilation. International Journal of Ventilation 2 (4), 395408.
Kato, S., Murakami, S., Mochida, A., Akabashi, S., Tominaga, Y., 1992.
Velocity-pressure eld of cross ventilation with open windows
analyzed by wind tunnel and numerical simulation. Journal of Wind
Engineering and Industrial Aerodynamics 4144, 25752586.
Kurabuchi, T., Ohba, M., Endo, T., Akamine, Y., Nakayama, F., 2004.
Local dynamic similarity model of cross ventilation: Part 1
theoretical framework. International Journal of Ventilation 2 (4),
371382.
Kurabuchi, T., Endo, T., Ohba, M., Goto, T., Akamine, Y. 2005. Local
dynamic similarity concept as applied to evaluation of discharge
coecients of cross-ventilated buildings. Part 1: Basic idea and
underlying wind tunnel tests. In: Santamouris, M. (Ed.), Proceedings
of PALENC, Santorini, Greece, May 1921, pp. 409414.
Lee, J.Y., Kang, J.H., Jang, Y.G., Zhang, W., Lee, S.J. 2005. PIV analysis
of the ventilation ow inside a large-scale factory building. In: Choi,
C.K., Kim, Y.D., Kwak, H.G. (Eds.), Proceedings of 6th AsianPacic
Conference on Wind Engineering (APCWE VI), Seoul, Korea,
September 1214, pp. 427436.
Liu, H., 1991. Wind engineering A handbook for structural engineers.
Prentice-Hall, New Jersey.
Mochida, A., Yoshimo, H., Takeda, T., Kakegawa, T., Miyauchi, S.,
2005. Methods for controlling airow in and around a building under

30

P. Karava et al. / Solar Energy 81 (2007) 2030

cross-ventilation to improve indoor thermal comfort. Journal of Wind

Engineering and Industrial Aerodynamics 93, 437449.
Murakami, S., Kato, S., Akabash, I.S., Mizutani, K., Kim, Y.-D., 1991.
Wind tunnel test on velocity-pressure eld of cross-ventilation with
open windows. ASHRAE Transactions 97 (1), 525538.
Niachou, K., Hassid, S., Santamouris, M., Livada, I., 2005. Comparative
monitoring of natural, hybrid and mechanical ventilation systems in
urban canyons. Energy and Buildings 37, 503513.
Ohba, M., Kurabuchi, T., Endo, T., Akamine, Y., Kamata, M.,
Kurahashi, A., 2004. Local dynamic similarity model of crossventilation: Part 2 application of Local dynamic similarity model.
International Journal of Ventilation 2 (4), 383393.
Reardon, J.T., Won, D., MacDonald, R.A., 2003. Characterisation of
Air Leakage, Pressure Regimes and Resultant Air Movement in High-

rise Residential Buildings. Institute for Research in Construction,

Indoor Environment Program, National Research Council of Canada.
Sandberg, M., 2004. An alternative View on Theory of Cross-Ventilation.
International Journal of Ventilation 2 (4), 400418.
Sawachi, T., Narita, K., Kiyota, N., Seto, H., Nishizawa, S., Ishikawa, Y.,
2004. Wind pressure and airow in a full-scale building model under
cross ventilation. International Journal of Ventilation 2 (4), 343357.
Stathopoulos, T., Surry, D., Davenport, A.G. 1979. Internal pressure
characteristics of low-rise buildings due to wind action. In: Proceedings
of the 5th International Wind Engineering Conference, 1, Fort Collins,
Colorado USA, pp. 451463.
Wu, H., Stathopoulos, T., Saatho, P. 1998. Wind-induced internal
pressures revisited: low-rise buildings. In: Proceedings of Structural
Engineers World Congress, San Francisco, CA, USA.