Renewable Energy 83 (2015) 1e12

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Performance enhancement of double skin facades in hot and dry

climates using wind parameters
Nazanin Nasrollahi*, Majid Salehi
Department of Architecture, Engineering and Technology Faculty, Ilam University, Ilam, Iran

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 1 March 2014
Accepted 7 April 2015
Available online 24 April 2015

This research aims to demonstrate a model of Double Skin Facade (DSF) that functions efciently in hot
and dry climates. Knowing that DSF performs well in winter at hot and dry climates as thermal mass is
needed. However, this method during summer evinces overheating between the two skin layers. This
paper introduces some modications in order to improve the functioning of DSF when overheating
occurs in the cavity between the two skin layers. Overheating in DSF can be prevented by using properties of wind pressure. Considering the condition of the wind ow, different geometric forms in the
upper part of DSF as well as its lower portion of window opening are evaluated. The numerical method
and Computational Fluid Dynamics (CFD) simulations are used in order to evaluate the hypotheses of this
study. The obtained results of this research suggest that dividing the cavity space into smaller parts
makes no signicant changes. Designing an additional channel in the northern part of the models,
directly impacts the functionality of DSF, which can be concluded to be very efcient. Finally, by
increasing airow velocity within the cavity, it is possible to decrease the problems of using DSF in hot
and dry climates.
2015 Elsevier Ltd. All rights reserved.

Keywords:
Building envelope
Double skin facades
CFD
Hot and dry climates
Airow
Overheating

1. Introduction
Glazing facades have recently been the center of attention [1,2],
and its popularity is increasing due to creating a desired facade
transparency along with improved acoustical quality and reduction
of energy consumption [3]. If they are designed properly it can
signicantly diminish energy consumption of HVAC systems in
buildings by obtaining solar energy during winter and controlling
overheating in summer [4]. It all can be reasoned and justied by
energy consumption perspective and the increased tendency towards new approaches such as passive solar systems [5]. As an
appropriate alternative, the DSF is composed of an external layer
(usually glass) and a centric space as well as an internal layer [6e9].
In addition to providing necessary transparency, the DSF can save
the absorbed solar radiation by an external glass facade in winter
and provide the appropriate ventilation during summer. As a result,
it improves the thermal comfort while reducing cooling and heating loads [5]. However, this technology performs better in countries
with moderate and cold climates [10e12]. As the temperature
* Corresponding author.
E-mail
addresses:
nazanin_n_a@yahoo.com,
(N. Nasrollahi), msalehi_88@yahoo.com (M. Salehi).
http://dx.doi.org/10.1016/j.renene.2015.04.019
0960-1481/ 2015 Elsevier Ltd. All rights reserved.

n.nasrollahi@mail.ilam.ac.ir

increases during the summer, especially in hot and dry climates, the
cavity between the facade skins experiences a severe increase in
temperature due to the greenhouse effect, and subsequently creates an uncomfortable condition for the occupants [13].
Recently, the technology of the double skin facade is being utilized in new buildings in regions that are very hot during the
summer and cold in the winter. This also can be related to the
recent growth of economic development [14]. As a result, the tendency towards implementing these facades is increasing and this
can be witnessed in Iran, where vast majority of the country is
located in hot and dry climates [5]. Although the double skin facades works well in winter of hot and dry climates, this application
may cause overheating between the skins layers during the summer. In order to solve the issue of cavity temperature, the appropriate dimensions of windows, suitable situation and position of
shading devices as well as the optimized space of the cavity are
suggested [15e19]. However the wind parameter, as an effective
factor for overheating ventilation in the cavity, is less emphasized.
There is a potential for using natural ventilation through the local
wind power in Iranian hot and dry climates. This paper studies the
direct effect of wind distribution on the air ow inside the cavity by
implementing some changes on the upper part of the DSF and
changing the position of windows. By dividing the cavity space into

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

smaller parts as well as adding a channel upon the arrival of wind

ow, functionality of different DSF layouts is compared.
1.1. Function of the double skin facades
As an important architectural element, the double skin facades
are widely used in ofce buildings. These facades are composed of
two layers, one exterior and the other interior, which are of
different glass materials that are separated by a cavity. The exterior
layer protects the structure and signicantly minimizes noise
pollution.
This advantage allows users of this technology to enjoy opening
their windows without facing the problems of mono-skin facades
which are the effects of wind pressure and speed, direct sunlight,
and environmental pollution. The possibility of installing shading
devices inside a cavity between two skins is another feature of the
DSF. The shading devices protect the rooms from direct light of the
sun while they decrease the cooling load in summer. The DSF
functions as a heat exchanger in cold weather, meaning that the
radiant energy is retained between the two layers and its temperature equals that of the building. This reduces the heat exchange
between outer and inner space. The method of ventilation inside
the DSF can be natural, mechanical, or a combination of both.
Natural ventilation is produced by the buoyancy resulting from the
difference in temperature (which causes pressure difference inside
the cavity) between inside and outside of the cavity. Also, factors
including pressure, velocity, and direction of the wind, building
orientation, and dimensions of the openings are effective in
creating appropriate natural ventilation in a building [14,20].
Considering the aforementioned points, the advantages of DSF
are as follows [21]: suitable insulator of noise, thermal insulator in
winter and summer, night ventilation (night ventilation can be
used for cooling the thermal mass of structure and decreasing
mechanical load in regions with high temperature difference between day and night as well as dominant wind during the summer), saving energy and reducing environmental damages, suitable
protection of shading devices, deduction of wind pressure, transparent designing, natural ventilation and reduction of heat
transfer.
However, the system has its own particular drawbacks,
including: Increasing the primary cost of building, conagration
crisis (if there are no modications in designing, the DSF can
intensify the re due to the high suction), the issue of increasing
temperature in summer (due to lack of suitable ventilation and
shading devices can convert this facade into a thermal furnace),
increasing air velocity and interfering the system as well as
increasing the total weight of the structure [21].
This study focuses to investigate both the behavior of airow
and reduction of temperature within the cavity of DSF during the
summer.
2. Method
In order to examine and evaluate the DSF behavior while
proving the hypothesis of this project, meticulous studies are
conducted to forecast the created conditions. Solutions that include
both analytical and experimental, laboratory measurements and
computer simulation are among common methods used for this
particular matter. However, since analytical and experimental solutions are not highly accurate and due to high cost of laboratory
measurements, computer simulation is used. Instead, the high
credibility of numerical model and Computational Fluid Dynamics
(CFD) simulation are considered as an appropriate method in order
to meet the objectives of this study. The necessity of using CFD
simulation is well emphasized as one of the most powerful

techniques in the design phase of a DSF [1]. Fluent and Gambit

software (Geometry and Mesh Building Intelligent Toolkit), the
commercial Toolkits of CFD are used for the design and analysis of
the models and the hypothesis respectively.
3. Simulation procedure
Three different basic models are precisely analyzed for purpose
of studying the effects of DSF upper part and the location of its
windows. This analysis includes 16 different states based on the
change in wind direction and positions of the windows (Fig. 1). At a
constant speed of 3 m/s models are analyzed once the DSF is on
leeward and also when the DSF is on windward at angles of 0 and
180
respectively (the dominant wind velocity in Isfahan from 1955
to 2005 during warm months is about 3 m/s) (Table 1). The wind
ow in DSF is investigated and analyzed based on the urban and
buildings structure and orientation in Iran which is predominantly
northesouth. However due to the complexity of wind ow model,
the direct wind ow is considered whereas the turbulence is
omitted. After designing the models using Gambit software, their
networking is made by organized networks. The boundary conditions of the models are dened based on Fig. 2; although these
conditions can be altered by Fluent software in the next stages of
the analysis. The boundary condition of the windows is considered
to determine the correct function of the windows in states of high
and low pressure.
In the second stage, the models are utilized in Fluent software
(Fluent 6.3.26), the solution condition is described in Table 2 and as
followings:
Solver: Pressure based.
Space: 3D.
Formulation: Implicit.
Time: Steady.
Operating pressure: 101325P.
Pressureevelocity coupling: SIMPLE.
Energy equation is inactive at this stage, since the study of air
ow speed and pressure is sufcient and the air is described as the
concerned uid (not as an ideal gas). With the entrance speed of
3 m/s, the ow uid is non-density. Depending on the chosen
viscose model for conditions of the ow, turbulent model of Kepsilon [22,23], is used. The ow is dominated by continuity
equations, momentum, K and epsilon while the tolerance for
solving them is considered to be 1e06. These equations become
convergent after a repetition range of 1000e1500.
Validity of the data is determined by wall Y plus criterion. It is
necessary to evaluate the sensitivity of the network near the wall
which is done by Y plus, since turbulent ows are highly inuenced by the walls. Using Keepsilon modeling, the network
arrangement near the walls should be such that Y plus < 5 or Y
plus > 30. The black dots represent the amount of Y plus along the
vertical axis of the northern wall of the DSF volume, as Fig. 2
demonstrates. In the mentioned region, according to the model
validity and the problem data none of the DSF levels are located
between 5 < Y plus < 30.
Velocity vectors, dynamic pressures and the difference in pressure created around and inside of the model are used to analyze the
model as well as to study the dynamic pressure and velocity of the
air ow, as shown in Fig. 3, the selected lines (a, b, c) are used.
4. Results and discussions
The main categories of the results and analysis are classied as
follows for precision in providing detailed results and analysis.

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 1. Physical properties of main models.

4.1. Models evaluation without divisions of the cavity

Table 1
The dominant wind velocity for Isfahan.
Average wind speed value from 1955 to 2005

Wind speed, m/s

July

Aug

Sep

Annul

3.6

3.4

2.8

3.6

4.1.1. Group A: a study of DSF upper part based on the amount of

cavity air ventilation
In Fig. 4, eight states of the models are analyzed for velocity,
dynamic pressure, and presence of difference in pressure within
the DSF and windows as well as its relationship with the wind ow
direction (0
and 180
) in addition to dynamic pressure and velocity of DSF selected lines (a, b, c) with vertical direction. Due to

Fig. 2. Boundary conditions of CFD model (left), Validation of models based on Y plus (right).

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Table 2
Summary of simulation solver parameters.
Parameters

URF/under relaxation factor

Discretization

Pressure
Density
Body force
Momentum
Turbulence kinetic energy
Turbulence dissipation rate
Turbulent viscosity

0.3
1
1
0.7
0.8
0.8
1

Standard
e
e
Second order upwind
Second order upwind
Second order upwind

Fig. 3. Positions of the selected lines in modeled DSF.

the similarity of a and c, a and b are solely used for comparison of the models. Fig. 5 (left) shows (a) at 0
wind ow for
different models. The (a) velocity is identical in these models.
Model A3 in the upper part of the DSF has the maximum velocity,
while models B and C have the minimum magnitude of velocity. In
Fig. 5 (right), a line is shown at 180
wind. Model A3 with 180
wind
ow shows maximum velocity. However in this model, the moving
force of wind causes a severe confusion inside the cavity upon
entering the DSF, leading to an increase in velocity. Line b for 4
models at 0
wind ow is represented in Fig. 6. Due to the lack of
direct ow into the DSF space, relatively similar results are
observed. Although, model A3, having similar conditions, shows a
greater velocity at the top of DSF.
In general, Fig. 6 presents average of line velocity in vertical
direction for all the models (group A). As previously shown, model
A3 has the maximum velocity at 0
degree wind ow. Fig. 6 also
suggests that models B and C relatively provide a suitable condition
as well. However, due to special design of the DSF upper part in
these models and their extra expenditure, model A3 is recommended for 0
wind ow. The magnitude of velocity and volume of
the outlet ow via the DSF windows is also shown in Fig. 6. Model
A3 performs well in a way that it provides the best result at 0
wind

ow. Hence, according to the obtained results model A3 is the best

alternative at 0
wind ow. In the state of 180
wind ow, no model
presents a desirable result and it considered that opening the DSF
lower part of window provides better results which are discussed
in the next model. Complementary explanations are required for
familiarity with model A3.
In this model, there is a direct relationship between velocity
vectors, the DSF upper part and the northern window. The Wind
force makes the outlet ow from the DSF occur faster and thus
increases the air suction of the DSF which is caused by the wind
force as Fig. 7 (left) shows. Air ow velocity at the entrance of the
DSF upper part is 1.32 m/s and subsequently it is 1.38 m/s at the exit
point. However, the magnitude of velocity entering the upper part
of the DSF in a 3-m height equals 0.87 m/s (main outlet). Fig. 7
(right) represents the plate with the pressure difference inside
and around the model. Since the air ow enters the northern
window directly, the pressure inside DSF, ranging (0e3 m/s), is
positive and it seems that it accounts for a part of ow velocity
increase inside the DSF. The negative pressure of the DSF at the exit
point (0.55 p) has a minor role in air suction of the DSF.
Fig. 8 provides complementary information. According to the
velocity diagram within all lines no turbulence is found inside the

Fig. 4. Group A models.

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 5. Line (a) velocity in 0 & 180-degree wind ow for different models.

DSF. Flow velocity in the DSF with the range of (0e3 m/s) is about
(0.2e0.6 m/s) which is increased to 1 m/s upon arrival at the DSF
upper part. Moreover, the dynamic pressure within the DSF upper
part undergoes a signicant increase. It can be concluded that this
model functions well up to the suitable parameters and also the DSF
outlet becomes faster.
4.1.2. Group B: DSF lower part analysis based on the amount of the
cavity air ventilation
Eight different types of models are analyzed in Fig. 9. The velocity lines (a, b, and c) are studied based on the vertical direction of
the DSF for these models. Similar to the previous comparison, only
a and b are used for this purpose. Unlike group (A), the crucial
change of this group is adding a window to the lower part of the
DSF. An extra window at the lower part of DSF allows the ow to
enter the cavity directly. The reason behind designing an opening is
to conduct the ow from the room to the DSF space. In Fig. 10 (left),
ow condition of 4 models at 0
is represented by a line. As previously discussed in the determined range, A6 enjoys the maximum
while A4 has the minimum velocity. However, the velocity of lines
hardly differs. Changing the wind direction to 180
in Fig. 10 (right),
there is comparison between the line and the other 4 models.
Model A5 has the maximum velocity at most points, but models C1
and B1 at the top of DSF witness a surge of velocity which is because
of the lines' location inside the DSF. However, analysis of A5 model
presents a much favorable behavior. Fig. 11 (left) demonstrates the
comparison of line b at 0
ow for the relevant models. It is evident
that line b behaves similarly to these models, and the ow

turbulence inside the DSF accounts for the insignicant differences.

In general at 0
wind ow, the DSF lower window's efciency reduces due to the downward suction of the cavity air. Fig. 11 (right)
compares line b at 180
ow for the relevant models. The suitable
suction of the outer window causes model A5's velocity to be
maximum at most points.
Fig. 12 presents vertical average velocity of the lines in group B
models. As previously mentioned, at 180
ow, model A5 shows
maximum velocity at most points. However, models C1 and B1
behave in a suitable manner, but due to some limitations and
production costs, model A5 is recommended. Also Fig. 12 shows the
magnitude of velocity as well as the outlet volume of the DSF in
which model A5 shows the best result in 180
wind ow. In order to
become more familiarized with model A5, complementary explanations are presented: here changes on A2 model are included. The
wind ow enters the DSF directly through the lower window and it
provides a desirable yet suitable condition for the cavity air to be
evacuated. Due to leeward orientation of the northern window
(negative pressure) and predominant positive pressure inside the
cavity, the ow enters the inner space through the DSF and exits the
building from the northern window. This conveys warm air into the
room; and in order to improve the situation, providing positive
pressure in the northern window can be helpful. This model is quite
expedient to dismiss the hot air within the DSF cavity at the summer (Fig. 13) and indeed, the inner skin of the window should be
closed to prevent ow of warm air into the building.
In addition, inner and outer ow volume analysis demonstrates
that a large amount of the ow exits through the DSF top window
while room windows play no role. Concentration of air at the DSF
top portion and the reduction in dynamic pressure along the way,
cause the ow velocity in the vertical direction of the DSF to
gradually decrease. However, the amount of the outlet ow is
acceptable.
In comparison with other models, model A3 at 0
ow and
model A5 at 180
ow, show more appropriate behavior (Fig. 14).
Combination of these two models is also possible, by using
dampers in some windows and making them intelligent in various
wind conditions.

4.2. Dividing DSF into smaller parts, studying the cavity airow
increase

Fig. 6. Line velocity average in vertical direction for all the models (Group A), velocity
and volume of outlet through DSF.

In the previous models, the DSF cube is modeled rectangular

and by changing its cross-section to a square, its performance
would be more efcient. Therefore, as seen in Fig. 15, the cavity
inner space is divided into three equal parts and for each cavity, one

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 7. Flow velocity vectors in section X 1.25 (left), contours of pressure coefcient in section X 1.25 (right).

Fig. 8. Lines velocity in 0-degree wind ow.

window is designed in the inner skin of the DSF (the area summation of the windows is equal to the previous version). These
changes are applied in two selected models namely; model A3 at
0
ow and model A5 at 180
ow (Fig. 16). The predominant

condition and the way of simulating the models correspond with

the previous samples and this has made the comparison of the
models possible. Similar to the previous samples, three selected
lines are used to study ow velocity in the vertical direction.

Fig. 9. Group B models.

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 10. Line (a) velocity in 0 & 180-degree wind ow for different models.

Fig. 11. Line (b) velocity in 0 & 180-degree wind ow for different models.

4.2.1. Model A33 at 0
wind ow

According to Fig. 17 (left) and the result analysis of the windows,
after the ow enters the inner space (through the northern window), the same amount of wind ow enters the cavity space
through the created windows of the DSF inner skin. Also the ow
passing through the DSF upper part leads to suction in the cavity air.
The volume of outlet is the same (12%) for all cavities. Centric cavity
outlet velocity is lesser than the other two (Table 3) due to the
severe concentration of ow at the top portion of the DSF. Fig. 17
(right) represents the ow in the vertical direction. After entering

Fig. 12. Average of line velocity in vertical direction for all the models (Group B), velocity and volume of outlet through DSF.

the cavity space, the airow moves downward and upward

subsequently.
4.2.2. Model A55 at 180
wind ow
180
wind ow enters the cavity space through the DSF down
windows and exits forcefully through outer windows (Fig.18). Fig.18
displays the presence of negative pressure in the northern window;

Fig. 13. The wind ow enters DSF directly through the down window.

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 14. A3 and A5 models as the chosen optimal samples.

Fig. 15. Plan and positions of the selected lines in modeled DSF.

accounting for entrance to the cavity air ow into the inner space.
Table 3 shows the amount of arrival ow from the DSF cavity into the
space (the solution for this issue is suggested in the next part). The
volume of outlet for all cavities is the same (77% in m3/s). Since the
dynamic pressure in the lower part of the DSF is high, the outlet
velocity for the centric cavity is higher than the others (Table 3).
The position of guide lines in the middle of DSF cavity impacts
the amount of ow velocity that is different from the median velocity in the outer window because the velocity of ow near the
walls is higher. It seems that dividing the cavity volume into smaller
parts would force the behavior of air ow inside the cavities to be
less turbulent.

4.2.3. Comparison of the models: the effect of the DSF cavity

division on increasing the ow
Models A3 and A5 are reevaluated by dividing DSF cavity into
smaller parts. Initially, it seemed that changing the plate into
smaller parts would certainly increase the speed of the ow inside
the cavities. However, no signicant change is observed by analysis of the data (Fig. 19). A decline in the speed of ow in the
vertical direction is obvious in some lines. The only noticeable
point is the increase in DSF outlet volume which is not important.
Finally, it is concluded that attaching and applying the divisions to
the cavity space imposes extra expenses to the structure as a
whole.

Fig. 16. Physical properties of A33 model in 0-degree wind ow (left), Physical properties of A55 model in 180-degree wind ow (right).

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 17. Flow velocity vectors in section X 1.25 (left), lines velocity in 0-degree wind ow (right).

Table 3
Velocity and volume of outlet through DSF in selected models, amount of arrival ow from DSF cavity into the space, velocity and volume of outlet through DSF.
Models

Cavity 1

Cavity 2

Cavity 3

Total

Model A33
0 degree

Volume of outlet, m3/s

Velocity of outlet, m/s

0.12
0.84

0.119
0.78

0.12
0.84

0.36
0.82 (average)

Model A55
180 degree

Volume of outlet, m3/s

Velocity of outlet, m/s

0.77
2.25
Window 1

0.77
2.31
Window 2

0.77
2.25
Window 3

2.31
2.27 (average)

Volumetric ow rate, m3/s

0.014

0.0030

0.014

4.3. Prevention of the DSF cavity air ow, entering the inner space
The DSF air ow entrance into the inner space is discussed in the
previous sections. That is due to the negative pressure in the
northern window at 180
wind ow (windward DSF), which is
undesirable for the users. However, this problem does not exist at
0
wind ow, since the wind ow enters directly to the inner space
that causes uncontrollable conditions. Moreover, there is no possibility of air refreshing while airow enters the inner space.
In order to navigate the wind of various directions into the inner
space, in addition to producing positive pressure at the north of the
structure, it is necessary to devise a channel in front of the northern
window and placing windows at its top portion. It is also possible to

Fig. 18. Flow velocity vectors in section X 1.25 in 180-degree wind ow.

control the air ow and refresh it prior to entering the inner space.
For this procedure, a channel is used, as seen in Fig. 20, with a crosssection 50  50 cm and a height of 4.70 m (1.70 m from the roof).
The channel is designed based on the previous models to justify the
comparison between possibility of increasing or decreasing the DSF
functionality (Fig. 21). The controlled conditions and simulation of
the model, are also based on the previous models for the same
purpose of comparison.
4.3.1. Model A33 after installing air channel at 0
wind ow
Wind ow leads to ow movement inside the cavity, as it enters
the inner space through air channel and windows of the DSF inner
skin, which can be witnessed by Fig. 22 (left). According to Fig. 22
(right), vertical velocity inside the cavity, is mostly, upward (positive velocity) indicating the appropriate state of ow within the
DSF. In Table 4, outlet volume in all cavities is (1%), although the

Fig. 19. Comparison of the lines velocity average in DSF vertical direction, velocity and
volume of outlet through DSF.

10

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

Fig. 20. Physical details of A33 model after installing air channel in northern side in 0-degree wind ow (left), Physical properties of A55 model in northern side in 180-degree wind
ow (right).

Fig. 21. Graphical expressions of wind ow conditions on models with additional channel.

outlet velocity in the second cavity is lesser. This can be justied by

reduction of entering velocity in the middle of the DSF.
4.3.2. Model A55 after installation of air channel at 180
wind ow
At 180
wind ow, the negative pressure of the northern front of
the structure induces the return ow from the DSF cavity into the
inner space. Adding a channel and locating its top window in front
of the wind ow, creates positive pressure on the northern window,
which prevents the ow return. Entering ow arrives at the inner
space through the air channel with an average speed of 0.88 m/s
and ows through the windows of the inner skin of DSF subsequently. Thereupon, it joins the incoming ow from the DSF down
window and leaves from the exit window, Fig. 23 (left).
Fig. 23 (right) shows the upward ow inside the cavity and the
downward ow within the channel, which both meet the determined objectives. Velocity and volume of the DSF outlet are provided in Table 4, in which the centric cavity has the highest amount.

The attached channel, in addition to increasing the velocity of ow

inside the cavity, prevents entrance of the hot air into the inner
space. Table 4 also presents the amount of inner air space in the
cavity space.
4.3.3. Comparison of models to study the function of the attached
channel and improving system quality
By creating a vertical channel on the northern window, models
A33 and A55 are re-evaluated. At 0
wind ow, while the attached
channel is in use, the volume of entering ow into the inner space is
decreased due to the concentration of ow and connement of the
channel space. In addition, it affects the output of the DSF and
decreases the volume and velocity of the cavity outlet. Placing the
air channel windward modies the velocity within the input of DSF
upper part and decreases the suction of the cavity air.
The average velocity of the model lines, compared to model A33
according to Fig. 24, shows a reduction of 0.04 in each cavity. The

Fig. 22. Flow velocity vectors in section X 1.25 in 0-degree wind ow (left), lines velocity in vertical direction and additional channel (right).

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

11

Table 4
Velocity and volume of outlet through DSF and amount of arrival ow from DSF cavity into the space in selected models.
Cavity 1

Cavity 2

Cavity 3

Total

Model A33, windward

Models
Volume of outlet, m3/s
Velocity of outlet, m/s

0.10
0.71

0.10
0.69

0.10
0.71

0.30
0.70 (average)

Model A55, windward

Volume of outlet, m3/s

Velocity of outlet, m/s

0.79
2.32
Window 1

0.81
2.40
Window 2

0.79
2.32
Window 3

2.39
2.35 (average)

Volumetric ow rate, m3/s

0.05

0.06

0.05

Fig. 23. Flow velocity vectors in section X 1.25 in 180-degree wind ow (left), lines velocity in vertical direction and additional channel (right).

reduction in velocity and volume of the DSF outlet is represented in

Fig. 24. In general, it is evident that attaching the channel on the
northern front, at 0
wind ow, insignicantly decreases DSF
functionality. However, attaching this channel makes it possible to
control the ow entering the structure and receive wind ow from
all directions.
At 180
wind ow, in which the ow from the DSF enters the
inner space, the attached channel in addition to providing ow for
the whole structure, it also solves entrance of warm air issue by
creating positive pressure on the northern window. Furthermore,
due to the northern front ow, the velocity and volume of the DSF
outlet are increased (Fig. 24).
5. Conclusion
In addition to controlling of the sun's heat in summer, the DSF is
a suitable thermal insulator in winter and an effective alternative.

Fig. 24. Comparison of the lines velocity average in DSF vertical direction, velocity and
volume of outlet through DSF.

The DSF is widely used particularly in the ofce buildings since not
only provides a completely glass facade, but also saves thermal
energy.
The aim of this article is to increase the ventilation velocity of
overheating outlet from the DSF cavity. Vertical divisions inside the
DSF cavity provide no positive results and only impose extra expenses. The entrance of airow into the inner space through the
DSF is also interpreted. As the temperature increases in the airlled cavity in summer, returning ow can create an unpleasant
condition for the occupants. Therefore, it is suggested to place the
channel at the northern front to increase the velocity and volume of
the DSF outlet. The simulation results also suggest that at 180
wind
ow, the ow from the DSF enters the inner space, the attached
channel in addition to providing ow for the whole structure, it also
prevents entrance of warm air by creating positive pressure on the
northern window.
As a result of the structure's concentration in cities and the
reduction of wind ow in lower levels, the channel can conduct air
ow from upper levels into the inner space, installation of a water
sprayer within the airway channel of the cavity; can provide a
pleasant air stream during summer.
Although the DSF has many advantages, conversely it has its
particular drawbacks. Scientically obtained results and the current
investigations agree upon the disadvantage of DSF which is
retaining the hot air and creating greenhouse effects. There are
multiple solutions for diminution of the preserved hot air. However
in this paper, the wind ow has been labeled as an important factor
regarding the heat reduction in the cavity holes. The estimated
location of inlet and outlet valves for proper directing of the wind
ow has been suggested which requires more precise and detailed
researches on openings.
Proper design as well as considering climate parameters make
the double skin facades effective in every climate and greatly solve
the challenges involved. In order to compare the answer to a specic climatic condition, some of important factors of the DSF

12

N. Nasrollahi, M. Salehi / Renewable Energy 83 (2015) 1e12

including orientations, cavity depth, cavity performance height,

size and height of openings in external facade require further investigations. In addition, it seems that there is a need for more
research on simultaneous effects of wind ow on natural ventilation within the cavity with respect to temperature and energy
consumption. This is for optimization and proper model selection,
in order to develop design guidelines for the DSF in different
environmental conditions.
We suggest a comparison between experimental results and
theoretical/simulation results of this scientic research for further
validation and accuracy of our purposed hypothesis.
References
 E, Cabeza LF. Numerical modelling of
[1] De Gracia A, Castell A, Navarro L, Oro
ventilated facades: a review. Renew Sustain Energy Rev 2013;22:539e49.
[2] Iyi D, Hasan R, Penlington R, Underwood C. Double skin faade: modelling
technique and inuence of venetian blinds on the airow and heat transfer.
Appl Therm Eng 2014;71:219e29.
[3] Fuliotto R, Cambuli F, Mandas N, Bacchin N, Manara G, Chen. Experimental
and numerical analysis of heat transfer and airow on an interactive building
faade. Energy Build 2010;42:23e8.
[4] Shameri MA, Alghoul MA, Sopian K, Fauzi M, Zain M, Elayeb O. Perspectives of
double skin faade systems in buildings and energy saving. Renew Sustain
Energy Rev 2011;15:1468e75.
[5] Hashemi N, Fayaz R, Sarshar M. Thermal behavior of a ventilated double skin
facade in hot arid climate. Energy Build 2010;42:1823e32.
[6] Safer N, Woloszyn M, Jacques Roux J. Three dimensional simulation with a
CFD tool of the airow phenomena in single oor double skin facade equipped
with a venetian blind. Sol Energy 2005;79:193e203.
[7] Ding W, Hasemi Y, Yamada T. Natural ventilation performance of a double
skin facade with a solar chimney. Energy Build 2005;37:411e8.
[8] Roth K, Lawrence T, Brodrick J. Double skin facades. ASHRAE J
2007;15(October):70e3.

[9] Baldinelli G. Double skin facades for warm climate regions: analysis of a solution with an integrated movable shading system. Build Environ 2009;44:
1107e18.
[10] Faggembau D, Costa M, Soria M, Oliva A. Numerical analysis of the
thermal behaviour of glazed ventilated facades in Mediterranean climates, part II: applications and analysis of results. Sol Energy 2003;75:
229e39.
[11] Gratia E, De Herde A. Optimal operation of a south double skin faade. Energy
Build 2004;36:41e60.
[12] Hossegen R, Wachenfeldt BJ, Hanssen SO. Building simulation as an assisting
tool in decision making, case study: with or without a double skin facade?
Energy Build 2008;40:821e7.
[13] Gratia E, De Herde A. Natural cooling strategies efciency in an ofce building
with a double skin faade. Energy Build 2004;36:1139e52.
[14] Zhou J, Chen Y. A review on applying ventilated double-skin facade to
buildings in hot-summer and cold-winter zone in China. Renew Sustain Energy Rev 2010;14:1321e8.
[15] Oesterle E, Lieb R, Lutz M, Heusler W. Double-skin facades, integrated planning. Munich e London e New York: Prestel Verlag; 2001. p. 207.
[16] Wigginton M, Harris J. Intelligent skins. Tunbridge Wells, Kent: Gray Publishing; 2002. p. 176.
[17] Schittich C. Building skins: concepts, layers, materials. Birkhauser; 2001.
p. 196.
[18] Li Junmei, Xing Xuefei, Hu Cheng, Li Yanfeng, Yin Chenchen, Liu Shanshan.
Numerical studies on effects of cavity width on smoke spread in double skin
facade. Proc Eng 2012;45:695e9.
[19] Janicki M, Heim D. Double criterion optimization of transparent facades based
on solar thermal processes. Front Archit Res 2013;2:23e9.
[20] Gratia E, Deherde A. Is day natural ventilation still possible in ofce buildings
with a double-skin facade? Build Environ 2004;39:399e409.
[21] Poirazis H. Double skin faades for ofce buildings, literature review.
Department of Construction and Architecture, Division of Energy and Building
Design, Lund Institute of Technology, Lund University; 2004.
[22] Launder BE, Spalding DB. The numerical computation of turbulent ows.
Comput Methods Appl Mech Eng 1974;3(2):269e89.
[23] Chen Q. Comparison of different ke3 models for indoor airow computations.
Numer Heat Transfer B 1995;28:353e69.