Numerical Modelling of Combined Heat Transfers in a

Double Skin Façade - Full Scale Laboratory Experiment

Validation
Frédéric Kuznik, Tiberiu Catalina, Lucie Gauzere, Monika Woloszyn,
Jean-Jacques Roux

To cite this version:

Frédéric Kuznik, Tiberiu Catalina, Lucie Gauzere, Monika Woloszyn, Jean-Jacques Roux. Nu-
merical Modelling of Combined Heat Transfers in a Double Skin Façade - Full Scale Labora-
tory Experiment Validation. Applied Thermal Engineering, Elsevier, 2011, 31 (14-15), pp.3043.
�10.1016/j.applthermaleng.2011.05.038�. �hal-00781339�

HAL Id: hal-00781339

https://hal.archives-ouvertes.fr/hal-00781339
Submitted on 26 Jan 2013

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Accepted Manuscript

Title: Numerical Modelling of Combined Heat Transfers in a Double Skin Façade - Full
Scale Laboratory Experiment Validation

Authors: Frédéric Kuznik, Ph.D Tiberiu Catalina, Lucie Gauzere, Monika Woloszyn,
Jean-Jacques Roux

PII: S1359-4311(11)00298-5
DOI: 10.1016/j.applthermaleng.2011.05.038
Reference: ATE 3585

To appear in: Applied Thermal Engineering

Received Date: 20 October 2010

Revised Date: 6 May 2011
Accepted Date: 28 May 2011

Please cite this article as: F. Kuznik, T. Catalina, L. Gauzere, M. Woloszyn, J.-J. Roux. Numerical
Modelling of Combined Heat Transfers in a Double Skin Façade - Full Scale Laboratory Experiment
Validation, Applied Thermal Engineering (2011), doi: 10.1016/j.applthermaleng.2011.05.038

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
*Manuscript
Click here to view linked References ACCEPTED MANUSCRIPT

Numerical Modelling of Combined Heat Transfers in a

Double Skin Façade - Full Scale Laboratory

PT
Experiment Validation
Frédéric Kuznika,∗, Tiberiu Catalinab , Lucie Gauzerec , Monika Woloszyna ,

RI
Jean-Jacques Rouxa
a
Université de Lyon, CNRS

SC
INSA-Lyon, CETHIL, UMR5008, F-69621, Villeurbanne, France
Université Lyon 1, F-69622, France
b
Technical University of Civil Engineering, Bucharest, Romania
c
EDF R&D, département EnerBAT

U
Avenue des Renardières - Ecuelles - 77818 MORET SUR LOING

Abstract
AN
M
The present paper presents a numerical modelling of a double skin façade
(DSF). The model developed includes a zonal model approach for the mass
D

transfer based on the pressure difference in the DSF. The radiative and con-
vective heat transfers are also taken into account to obtain a global coupling
TE

between the different phenomena. A full-scale DSF has been experimentally

studied in summer configuration with different airflow rates through the air
EP

channels of the façade and for different angles of the solar shading devices.
First the numerical modelling has been validated using the experimental
C

data. Then, The model has been used to study the influence of airflow rates
AC

and blades angles (Venetian blind) on heat transfer in DSF.

Keywords: Double Skin Façade, Numerical Modelling, Full-scale

∗
Corresponding author. Tel.: +33-472-438-461; Fax: +33-472-438-522
Email address: [email protected] (Frédéric Kuznik)

Preprint submitted to Applied Thermal Engineering May 6, 2011

 ACCEPTED MANUSCRIPT

experiment, Venetian Blinds.

PT
1. Introduction

Double skin façades (DSF) are architectural elements that have increased

RI
in popularity in modern buildings and renovations. They can be found in
many projects, especially in office buildings. A fully glazed façade brings

SC
the transparency that is often desired by architects when designing an office
building. The DSF seams are also interesting features, notably in terms of

U
sound insulation, thermal comfort, visual comfort or energy gain [1].

AN
In the current context of global warming, depletion of fossil fuels and
increasing costs of energy consumption, the target is to obtain energy reduc-
tion while maintaining correct indoor comfort. To achieve these objectives,
M
it is essential to optimize different parts of active façades and, in particular,
for double skin façades. The advantages of using a DSF are located on two
D

levels: during the winter period, the solar energy is used to heat the air in
TE

the façade [2], and, in summer, it reduces the overheating periods [3].
A double skin façade consists of a double glazing separated by a ventilated
air space with two openings at the top and bottom of the façade. Generally,
EP

it can also be equipped with internal or external shading devices. In this

article, Venetian blinds are assumed to be located in the air channel between
C

the two glazed panes. Following this definition, a double skin façade can
AC

have different configurations, and an example of the geometry can be found

in Figure 1.
The first two groups of DSFs are the double skin façades with a large
cavity width (0.5m − 1m), such as the one studied in the research work of

2
 ACCEPTED MANUSCRIPT

Xu et al. [4], and those with a small air channel (0.10m − 0.3m), like the
one studied in this article. Some other DSFs that can also be considered are

PT
those with high rise, with a single channel, as in the case of the Occidental
Chemical Center in New York City cited in the research study of Poirazis [5],

RI
or DSFs of smaller dimensions. These DSFs are less practical from a cleaning
and energy efficiency point of view.

SC
DSFs can also be mechanically or naturally ventilated. Four configu-
rations are possible by varying the inlet and outlet air (indoor air-indoor

U
air, outdoor air-outdoor air, indoor air-outdoor air, outdoor air-indoor air),
depending on the desired indoor climate conditions.
AN
A key issue of double skin façades is to avoid overheating, especially in
summer. Thus, DSFs are equipped with different types of solar protection
M
blinds. Then the research can focus on the influence of the cavity and the
air temperature on the air velocity field based on the position (inside, out-
D

side or between the cavities), color and angle of the blinds. Few studies
TE

have considered this problem because of the lack of multi-parameter studies.

The glass properties are also very important concerning the summer thermal
comfort [6].
EP

Finally the behaviour of double skin façades depends greatly on the ex-
ternal conditions, and thus different orientations [7] or different season con-
C

figurations [8] may be analyzed. One possible research objective on double

AC

skin façades is to use a control system for the various parameters mentioned
above that will reduce the energy consumption while attempting to main-
tain acceptable thermal comfort for the occupants. Park et al. [9] proposed
strategies for controlling the tilt angle of the blinds and the ventilation of

3
 ACCEPTED MANUSCRIPT

the double skin.

Concerning the DSF models, Poirazis [5] described two types of numeri-

PT
cal models, the CFD numerical modelling (Computational Fluid Dynamics)
approach and the "network" numerical modelling. Generally, the numerical

RI
models combine CFD modelling of airflow and nodal modelling of other phe-
nomena. In the case of nodal models, two-dimensional studies are generally

SC
performed.
For the nodal models, the channel inside the façade is cut into strips and

U
describes the radiation exchange, convection and mass transfer, including
identifying coefficients for each transfer. Safer et al. [10] determined the
AN
convective heat transfer coefficients of convective and radiative transfer from
the velocity and air temperature in the cavity through a CFD model including
M
a turbulence model.
The results of the 2D double skin façade turbulent regime were analysed
D

by Safer et al. [11] and showed the influence of the tilt angle of the shad-
TE

ing devices, the position of the blinds in the cavity and the position of the
inlet/outlet.
Other approaches such as the dimensionless model [12], a "common"
EP

model suitable for testing specific DSF [13], an optical model, CFD [14]
and global [15] have also been developed.
C

The objective of this study is to describe a new approach for the simu-
AC

lation of DSFs in terms of combined heat transfers. The main objective of

this model is to develop a model that can be used for the evaluation of DSFs
integrated in a building for at least one year. Thus, the model must be phys-
ical (all of the heat transfer processes must be taken into account), accurate

4
 ACCEPTED MANUSCRIPT

(nodal modelling) and computationally fast (numerical solving process and

hypothesis).

PT
The first part of the paper (part 2) describes the numerical modelling of
a DSF, the assumptions and particularly the zonal model used to predict the

RI
mass transfer in the cavity. The numerical modelling can only be validated
through accurate experimental data. Then part 3 of the paper is dedicated

SC
to the description of the full-scale laboratory experiment used in this study.
The numerical modelling and experimental data are compared in part 4 of

U
this paper. Useful conclusions concerning the use of DSFs are developed in
part 5.
AN
2. DSF numerical modelling
M
To model the double skin façade, the combined heat and mass transfer
are modelled:
D

• mass transfer - section 2.1

TE

• radiative heat transfer - section 2.2

EP

• convective heat transfer - section 2.3

Once the combined heat and mass transfers are modelled, the thermal energy
C

balance (part 2.4) is developed to solve the coupled problem.

AC

2.1. Mass transfer modelling

A zonal model is an intermediate approach between CFD and multi-

zone models. This approach has the convenience of simplicity in terms of
problem definition and calculation procedure, and there is a considerable

5
 ACCEPTED MANUSCRIPT

computational time reduction. Several authors, like Saelens [16], Jiru et

al. [17] or Stec et al. [18], have used the zonal models to understand the

PT
airflow inside the DSF. In our study, a similar approach is used with the
difference that a detailed look at the influence of the Venetian blinds and

RI
ventilation rate is studied, and it is compared to the experimental data.
For the modelling process, the space in the DSF is discretized in n horizon-

SC
tal bands (see Figure 3). For each of the cells i, the variables are the exterior
glazing temperature T1 (i), the left channel temperature Tair (i, 1), the shad-

U
ing blinds temperature Tb (i), the right channel temperature Tair (i, 2) and the
interior glazing temperature T2 (i). The assumptions of two channels coupled
AN
via the Venetian blinds are justified by previous CFD studies (see Figure 2).
The zonal model is developed by assuming that the cavity internal airflows
M
are functions of the internal pressure difference as proposed by Saelens [16]:

Qmass = ρin KA |∆p|n (1)

D

Based on the Safer research work, the airflow inside the façade is turbulent
TE

and thus n = 0.5. The cavity is divided into horizontal stripes with a height
equal to the shading blind length. For each of the cells, the air mass balance
EP

can be written, so 2n equations are obtained with the pressures as unknowns.

The advantage of using this discretization is that the airflow Qa between the
C

cells of the same stripe can be considered as a Poiseuille plane flow, and the
K coefficients can be calculated based on the geometrical characteristics of
AC

the DSF. However, knowing that there is a large amount of shading blinds,
the large number of equations to be solved and the processing time could
slow this process, and, thus, it cannot be considered as a simple and faster
model. An arbitrary discretization is preferred, which can reduce the number

6
 ACCEPTED MANUSCRIPT

of equations but at the same time increase the difficulty in determining the
K coefficients for different flows. Each cell is defined by the couple i, j, where

PT
i indicates the line location and j takes the value 1 or 2, depending on the
cavity side, the left or the right part. The cells are assumed to have an

RI
uniform pressure p(i, j). For each case (see Figure 3), three mass flows are
calculated:

SC
Qa = ρin Ka A |∆pa |n
Qb = ρin Kb A |∆pb |n (2)
Qc = ρin Kc A |∆pc |n

U
with
AN
∆pa (i, j) = p (i, 1) − p (i, 2)
∆pb (i, j) = p (i, 1) − p (i − 1, 1) (3)
∆pc (i, j) = p (i, 1) − p (i + 1, 1)
M
The mass balance for the cell can be written as
D

X
Qmass−υ = Qa (i, j) + Qb (i, j) + Qa (i, j) = 0 (4)
υ
TE

where υ is the interface with an adjacent cell. The mass balance can be
further presented as follows:
EP

f (i, j) = sign (∆pa (i, j)) Qa (i, j)+sign (∆pb (i, j)) Qb (i, j)+sign (∆pc (i, j)) Qc (i, j) = 0
(5)
C

Proceeding with the mass balance for all the cases, a 2n-dimensional
vector of equations is
AC

f = {fk }k=1...2n = {f (i, 1)}k=1...n ∪ {f (i, 2)}k=1...n (6)

The unknown vector is

p = {pk }k=1...2n = {p (i, 1)}k=1...n ∪ {p (i, 2)}k=1...n (7)

7
 ACCEPTED MANUSCRIPT

The first boundary condition is the inlet outdoor air flow Qext , and, based
on Figure 3, it is clear that

PT
Qb (1, 1) = −Qext (8)

RI
Moreover, the reference pressure is imposed to be the external pressure
Pext in the upper part of the DSF, so the final flow is written as

SC
Qc = ρin Kc A |p (n, 1) − pext |n (9)

U
To calculate the mass flows for each cell, the air density is calculated
based on the outdoor temperature in the upper/lower part of the DSF and
a temperature vector written as
AN
M
Tair = {Tair,k }k=1..2n = {Tair,k (i, 1)}k=1..n ∪ {Tair,k (i, 2)}k=1..n (10)

ρ = {ρk }k=1..2n = {ρk (i, 1)}k=1..2 ∪ {ρk (i, 2)}k=1..2 (11)

D

The solving methods used require an initial value to be defined, repre-

TE

sented by an initial vector:

p0 = {p0k }k=1..2n = {p0 (i, 1)}k=1..2 ∪ {p0 (i, 2)}k=1..2 (12)

EP

The model was integrated in Matlabr , and the resolution of the non-
C

linear system uses the existing function from the optimization toolbox. Six
layers were finally considered and found to be representative for the DSF.
AC

The solver converged for this case with a computational time of less than 10
seconds.
The model is very sensitive to the coefficients linking the flows (in the left
and right vertical channels and between the horizontal bands) and pressure

8
 ACCEPTED MANUSCRIPT

differences between cells. To obtain the values of these coefficients, they were
adjusted based on the experimental results obtained by Gavan et al. [19]. Ini-

PT
tially, three coefficients Kb , Ka , Ki were considered. These factors represent
the permeability of the horizontal blinds, air channels and air inlet.

RI
The determination of coefficients relied on some assumptions. The coef-
ficient Kb should vary depending on the angle of the blinds. The horizontal

SC
flow must be very low when the blinds are closed α = 0◦ . At α = 45◦ , it can
also be considered that the flow in the left channel and the one in the right

U
channel should be substantially equal. Thus, the ratio Kb /Ka should vary
from 0 to 1 depending on α, with the value 0.5 for α = 45◦ . In addition, the
AN
coefficients Ka and Ki should be approximately equal. The adjusted values
of the pressure coefficients are presented in Table 1.
M
The pressure model determines the pressure field inside the double skin
façade by imposing a mechanical flow and the air temperatures. It can also
D

determine the values of the air flow in the double skin façade from the pressure
TE

field solution.

2.2. Radiative heat transfer

EP

In a DSF, the natural convection and heat transfer play an important

C

role. For building components, the radiative heat transfer can be divided
into long-wave (LW) heat transfer and short-wave (SW) heat transfer. In
AC

the first stage, the SW heat flux absorbed by the various components of
the façade is evaluated, and the net flux exchanged in LW is calculated. In
this section, the SW and LW radiation heat transfers of the DSF model are
studied. In the first part 2.2.1, information regarding the movement and

9
 ACCEPTED MANUSCRIPT

the position of the sun are given. The form factors influencing the short-
wave radiation heat transfer process are evaluated as well. Part 2.2.2 is then

PT
described.

2.2.1. Heat flow absorbed by short-wave radiation

RI
Before calculating the flow, it is necessary to perform preliminary steps.

SC
First, the form factors of different components are calculated, and the direct
solar flux through these components is determined. Finally, different SW

U
irradiation received is established by taking into account the direct and diffuse
solar irradiation. AN
Calculation of the form factors. The form factors represent the fraction of
M
energy exchanged between the different facets of a surface domain. The form
factor Fij represents the fraction of energy leaving the facet i and received
D

by the facet j. These factors depend only on geometry. To calculate these

TE

different factors, the method of Hottel [20] is used. The assumptions are as
follows.
EP

Assumption 1: the radiative exchanges are two-dimensional.

Assumption 2: the Venetian blinds are flat.

C

Assumption 3: the exchange surfaces of windows are represented by

AC

the facets 1 and 2 (see Figure 4).

This method is a 2D method that does not take into account the phe-
nomena that occur on the edges of the double skin façade. In this method,
it is assumed that the windows receive the same radiation as the vertical

10
 ACCEPTED MANUSCRIPT

surfaces between the two strips. The interaction between the bands defined
by 2 strips is neglected. One limitation of this method is the width of the

PT
air channel. In the case of a double skin façade with a wider channel, this
method must be revised.

RI
Calculation of the solar ratio. This ratio, known as Fsol , estimates the amount
of direct solar radiation that passes through the horizontal blinds and enters

SC
the internal glazing surface directly. Therefore, (1 − Fsol ) is the amount of
radiation coming directly on the sunscreen. This ratio depends on the width

U
of the blinds, the height and position of the sun. It is assumed that the
AN
Venetian blinds tilt angle varies from 0◦ to 90◦ . It is also assumed that the
relative azimuth is zero which implies that the double skin façade is facing
south. The expression of the solar ratio is written as a function of α (angle of
M
incidence of sunlight), β (blind’s angle) and e (spacing between the Venetian
blinds):
D

e − (a + b)
Fsol = (13)
e
TE

Estimation of the heat flow absorbed at SW. In the following section, assess-
ments of all four surfaces are carried out: upper blind SL , lower blind SS ,
EP

internal glazing S2 and external glazing S1 (see Figure 5). The direct so-
lar flux and the diffuse solar flux are taken into account, and the irradiance
C

received by each surface is written as:

AC

n
X
0
Ei,SW = Ei,SW + Fij ρj Ej,SW (14)
j=1

0
where Ei,SW represents the solar flux received directly without the multi-
reflections and ρj are the reflection factors, which leads to the following

11
 ACCEPTED MANUSCRIPT

system:

0
E1,SW =0

PT





0
E2,SW = F12 Edif f use + Fsol Edirect


(15)
0
 ES,SW = (S1 /SS ) F1S Edif f use + (S1 /SS ) (1 − Fsol ) Edirect

RI




E0 = (S /S ) F E


L,SW 1 S 1L dif f use

By solving the above system, access to the global irradiance at SW (di-

SC
rectly received and due to multi-reflections) received by each of the 4 surfaces
is possible.

U
2.2.2. Heat flow absorbed at long wavelengths
AN
There is also a heat transfer between the glazing in LW, directly related
M
to the temperature of the surface and the emissivity. The model uses the
radiosity method and the form factors determined by Hottel’s approach to
D

determine the flows for LW radiation. Each radiosity is defined in terms of

TE

emittance of all surfaces. Then the net heat flux radiated by each surface is
expressed, considering that each surface is gray and diffuse for the emission
and the reflection. The net flow in LW radiation of each surface is determined
EP

by the radiation exchange balance between surfaces:

C

 net 
φLW = [MAT ] [hr] T − Ts (16)
AC

with
 
[MAT ] = [S] [ε] [[I] − [F ] [I − ε]]−1 [F ] [ε] (17)

{T } = {Ts }s=1,b,2 (18)

 
hri = σ0 Ti2 + Ts2 Ti + Ts


(19)

12
 ACCEPTED MANUSCRIPT

where Ts is equal to 273.15K and σ0 is the Stefan-Boltzmann constant.

The model also takes into account the exchange in terms of the LW radi-

PT
ation with the outdoor (ground and sky) and the inner walls of the adjacent
room. The double skin façade exchange in LW radiation with the sky and

RI
the ground and the radiative flux can be calculated using the formula:

1 + cos θ 4 1 − cos θ 4
φnet 4 4



LW,ext = σ0 αLW Tsky − Ts,ext + σ0 αLW Tground − Ts,ext

SC
2 2
(20)
where θ is the angle of the wall from the ground. In the case of a vertical

U
double skin façade, θ = 90◦ , and, therefore, the second term disappears. The
coefficient hrc is used: AN
2 2


hrc = σ0 (Tsky + Ts,ext) Tsky + Ts,ext (21)
M
The values of hrc are in a small range for temperatures usually encoun-
D

tered in the building, so a constant value of hrc = 5W m−2 K −1 can be as-

sumed.
TE

φnet
LW,ext = αLW hrc (Tsky − Ts,ext ) (22)
EP

2.3. Convection transfer modelling

In DSFs, a convective heat transfer occurs at each air-wall interface. Six

C

convective flows can be distinguished in the double skin façade, which re-
AC

quire convection heat transfer coefficients. The six convective flows are out-
door air/external glazing, external glazing/left channel, left channel/blinds,
blinds/right channel, right channel/internal glazing and internal glazing/indoor
air (see Figure 6). It is assumed that the convective heat transfer coefficients

13
 ACCEPTED MANUSCRIPT

are constant over the height of the double skin façade and do not vary much
depending on the inclination of the Venetian blinds (further information can

PT
be found in [11]). For the calculation of the hc1 , hc2 , hc3 , hc4 , the calculation
of the Nusselt number is used [20]. Regarding the convective heat transfer

RI
coefficient between the outdoor air and the air inside the room, the model
uses values proposed by the actual French Thermal Regulation [21].

SC
2.4. Thermal energy balance

U
The time constant of the heat transfer process due to heat flow is much
AN
lower than the time constant of the building. Thus, the transient process is
assumed to be a succession of steady states. For each surface element (i, s), a
M
heat energy balance is performed by taking into account the radiative transfer
and the convective transfer:
D

P P
ent (i, 1) = [ρi,1,inlet cair Qi,1,inlet Ti,1,inlet ] − [ρair (i, 1) cair Qi,1,outlet Tair (i, 1)]
inlet f lows outlet f lows
TE

+hc1 S (T1 (i) − Tair (i, 1)) + hc2 S (Tb (i) − Tair (i, 1))
(23)
where Qi,1,inlet is the air flow coming from an adjacent cell and entering the
EP

cell i of the left channel, ρi,1,inlet is the mass density of air entering the cell
i of the left channel, Ti,1,inlet is the air temperature entering the cell i of the
C

left channel, cair is the specific heat of air, Qi,1,outlet is the outlet flow from
AC

cell i, ρair (i, 1) is the air density in cell i, Tair (i, 1) is the air temperature in
the left channel of the cell i, T1 (i) is the temperature of the external glass
the band i, Tb (i) is the Venetian blind temperature of cell i and S is the
amount of area of convective heat transfer S = el, where l is the width of

14
 ACCEPTED MANUSCRIPT

the double skin façade and e is the height of a cell. Similarly, the enthalpy
balance for the cell i of the right channel is:

PT
P P
ent (i, 2) = [ρi,2,inlet cair Qi,2,inlet Ti,2,inlet ] − [ρair (i, 2) cair Qi,2,outlet Tair (i, 2)]
inlet f lows outlet f lows

+hc3 S (Tb (i) − Tair (i, 2)) + hc4 S (T2 (i) − Tair (i, 2))

RI
(24)
For each cell i, the thermal energy balance of radiative and convective

SC
heat transfer for the three surfaces (interior glazing, Venetian blinds, external
glazing) is written. As an example for the external glazing, the following

U
phenomena are observed:

AN
- On the external face (direct and diffuse irradiation at short wavelengths,
radiative heat flux exchanged with the ground and sky, convective heat
M
exchange with the air)

- On the left channel face (heat flux absorbed in short wavelength, net heat
D

flux at long wavelengths, convective flow exchanged with the air of the
TE

left channel)
EP

th (i, int) = α1 S (Edif + Edir ) + hcground S (Tground − T1 (i)) + hcext S (Text − T1 (i)) +
α1 SE1,SW − Φnet,LW,DSF + hc1 S (Tair (i, 1) − T1 (i))
C

(25)
The global model is solved iteratively in 2 phases. The first phase is the
AC

determination of flows within the DSF from the model of mass transfer. In
this phase, the model uses the initial temperatures of the air and surface of
the DSF at the first iteration and then given by the second phase during the
following iterations. The second phase is the calculation of the temperatures

15
 ACCEPTED MANUSCRIPT

of air and surface from the system resolution resulting from energy balances.
For this stage, the model is applied using the flows inside the double skin

PT
façade given in the first step. The resolution stops when the differences
between the temperatures obtained during two consecutive cycles are below

RI
a chosen value. The model algorithm is presented in Figure 7.

SC
3. Full-scale experiment

The main problem with simplified models is to validate them with ac-

U
curate experimental data. A full-scale experiment was developed for this

AN
purpose, and the aim of this section is to describe this unique test cell ex-
periment. More information about these experiments are given in [19].
M
3.1. Description of the test cell

To study and validate the numerical model proposed, detailed experi-

D

mental data must be obtained. The reliability of different measurements can

TE

only be acquired in a controllable experimental environment in terms of solar

irradiation, air temperature or air flow. A double skin façade was installed
in the laboratory and was integrated with a test cell called MINIBAT.
EP

The experimental cell MINIBAT presented in Figure 8 is composed of two

identical enclosures (Cell 1 and Cell 2), whose dimensions are 3.10m, 3.10m
C

and 2.50m according to the coordinate directions (x, y, z). A double glazed
AC

façade isolates Cell 1 from a climatic chamber where the air temperature
is controlled by means of an air-treatment system. The climatic chamber
temperature can vary between −10◦ C and 40◦ C as a function of time. Using a
battery of 12 spotlights, of 1000W each, it is possible to simulate an artificial

16
 ACCEPTED MANUSCRIPT

sunning for Cell 1. The spectrum of the gas discharge lamps with metal
halide is similar to the sun.

PT
The double skin façade, presented in Figure 9, is made of two aluminium
frames: external and internal. The internal is composed of one fixed part

RI
(W × H = 2.8m × 2.3m), and the external one is divided into two parts
that can be separately opened. The glass panels separated by the 20cm

SC
of air volume are two simple clear glazing (without a surface treatment).
The glass area is W × H = (2 × 1.3m) × 1.93m for the external pane and

U
W × H = 2.8m × 1.93m for the internal pane.
Four openings are included in the DSF with dimensions of W × H =
AN
2.6m × 0.04m each. Using these openings in the upper and bottom part
of the façade, different ventilation configurations can be tested by simple
M
obstruction of these inflow spaces. As shading devices used in the experiment,
Venetian blinds were installed for the most commonly used case for such a
D

façade. The angle of the blinds can be oriented automatically during the
TE

tests from the exterior of the façade (mechanical orientation device).

The Venetian blades installed in the DSF are made of aluminium, were
2.5cm in width and 0.21mm in thickness, and the chosen color was yellow
EP

gloss. The DSF is exposed to the controlled climatic conditions, and an

accurate airflow control of air supply system was build to vary different air
C

flow rates.
AC

3.2. Metrology

The DSF airflow measurement system is composed of two differential pres-

sure devices (two blades, thus limiting the differential heads and turbulences)
and a sensor box with the differential pressure elements. These two pressure

17
 ACCEPTED MANUSCRIPT

elements have a distribution of apertures on the airfoil permitting the control

of the average airflow (by averaging the differential pressures). The pressure

PT
transmitters working with a differential probe are configured with a square
root function. Via this function, and from the differential pressure, the trans-

RI
mitter calculates airflow rate in the duct, and thus in the DSF. The accuracy
of the airflow measurement device is ±0.5% of the measurement.

SC
The DSF itself is instrumented using thermocouples, distributed on all
the elements of the DSF. Type-T thermocouples with flattened joint are

U
used here for a better contact with the glass surface. Among these sensors,
15 thermocouples are affected for the glass surface temperatures and 10 for
AN
air temperatures. All the thermocouples installed in the DSF are shielded to
the solar simulator radiation using a special paint. This paint is Labsphere
M
Spectraflect. The measured accuracy, taking into account the solar radiation
effects, is ±1◦ C.
D

3.3. Experimental protocol

TE

Every test consists in setting up parameters of the test cell facility (bound-
ary conditions) and DSF. Simultaneously, the solar simulator is switched on
EP

and synchronized with the acquisition system. The time delay between two
measurements is 300s (i.e. a measurement frequency of 0.003Hz). The test
C

cell reaches the steady state condition after 48h when the DSF is ventilated,
respectively, 72h with no airflow in the DSF. All the values presented in this
AC

paper were obtained averaging 100 values recorded at steady state. During
the tests, the air temperature inside the weather generator was fixed at 32◦ C.

18
 ACCEPTED MANUSCRIPT

4. Numerical validation

PT
This section is dedicated to a comparison between experimental data and
numerical results. For the completely closed blinds (90◦ ), because the theo-

RI
retical transmittance is equal to 0, the first consequence is that no radiation
will enter the cell. However, with multiple reflections and solar radiation

SC
leakage regions of the façade, the average radiation passing through the test
cell was found to be around 20W m−2 . Table 2 resumes the thermo-physical

U
proprieties of the external pane, the Venetian blades and internal pane (trans-

AN
missivity, absorptivity and reflectivity).
Different cases were studied with different airflows combined with differ-
ent tilt angles of the shading devices. For these cases, the angle of the Vene-
M
tian blinds can vary between 0◦ (completely opened blinds) and 90◦ (com-
pletely closed blinds). The airflow range varies between 0 and 600m3 h−1 ,
D

corresponding to 0 − 193m3 h−1 for one meter of façade.

TE

The external conditions of temperature and irradiance are set to the ones
during the experimental campaign. Thus, the comparison is significant and
EP

allows the validation of the global model of the DSF. The uncertainty of the
shielded temperature probes is ±1◦ C.
Two experimental measurements were studied, one taken in the bottom
C

part of the double skin façade and the other one taken in the upper part.
AC

The numerical model gives six temperature measurements on the DSF height.
There is a general increase in temperature with the height. In all cases, the
upper part of the double skin façade is warmer than the bottom. The results
were analyzed for different air flows, 0, 200, 400 and 600 m3 h−1 , and, for

19
 ACCEPTED MANUSCRIPT

each flow, five values of blades tilt angle α, 0◦ , 30◦ , 45◦ , 60◦ , 90◦ , were used.
The results of the numerical model are compared with the experimental

PT
data as shown in Figures 10, 11 and 12 in terms of the air temperature in the
left/right channel and the temperature of the Venetian blades. For an air flow

RI
of 400m3 h−1 in the DSF, the theoretical results are very close to the values
obtained by experimentation, excepting for α = 90◦ . The difference between

SC
the numerical modeling and the experimental data, taking into account the
uncertainty of measurement, never exceeds 0.5◦ C.

U
In Figure 10, the theoretical temperature of the air in the left channel
is close to the experimental value in the upper part of the DSF. For the
AN
bottom, a difference can be observed. The air injected into the channel is at
a temperature of 31.5◦ C, and it seems surprising that the air flow through the
M
channel is at 34.6◦ C only a few centimeters from the inlet, so there is doubt
as to the reliability or the position of this measurement. In Figure 13, the
D

temperature of the shading devices increases until α = 45◦ , then decreases

TE

for higher tilt angles.

For α = 0◦ , the angle of incidence is low and the blades receive a small
amount of irradiation radiation at short wavelengths. For α = 45◦ , the
EP

blades receive a higher amount of irradiation and also have the benefit of
multi-reflections. For α = 90◦ , the blades are closed, and all direct radiation
C

is received, but some part of the radiation is reflected directly to the external
AC

face of the DSF, which explains the temperature decrease. The temperature
inside the glass decreases fairly regularly when the tilt angles increases.

20
 ACCEPTED MANUSCRIPT

5. First observations

PT
There is less direct short wavelength illumination received by the inner
pane, which implies the significant role played by irradiation on the tempera-

RI
ture of this element. The temperature of the exterior glazing increases when
the tilt angles increases. The influence of the inclination of the blades is less

SC
important for the exterior glazing than for the internal glazing. The increase
in the temperature of the external pane is linked to the multi-reflections.

U
Among the three temperatures analyzed in Figure 13, the temperature of

AN
the inner pane shows the most important variations. This element of the
double skin façade is therefore particularly sensitive to the management of
the angle of the inclination of the blades.
M
The model results were compared to the experimental data on the width
of the double skin façade for different positions of the x-axis (Table 3). The
D

positions considered were x1 = 0m (external glazing), x2 = 0.05m (left chan-

TE

nel), x3 = 0.1m (Venetian blinds), x4 = 0.15m (right channel) and x5 = 0.2m

(interior glazing).
EP

In the case of the model results, as in the experiment (see Table 3), the
Venetian blades have the highest temperature of all elements, regardless of tilt
angle and ventilation rate, except in the case of α = 0◦ . In this configuration,
C

the shading devices capture less direct solar irradiance. These surfaces have
AC

the highest temperature, especially if the ventilation rate is not important.

For the closed position (α = 90◦ ) of the blades, a very low temperature
value of the internal glazing is observed. Indeed, in this configuration, the
inner pane, in the case of the numerical model, does not receive any radiation

21
 ACCEPTED MANUSCRIPT

from outdoor; it exchanges only radiation with the interior of the room.
The difference between the model and the experiment may signify that the

PT
interior is receiving a certain amount of irradiance. Furthermore, the model
estimates the radiation exchange with the interior of the piece based on strong

RI
assumptions on our part.
The measure of the temperatures of the walls inside the test cell could give

SC
more certainty in these exchanges. However, the slight differences observed
the temperature curves are close, and the model results seemed to be within

U
the area of uncertainty of the experimental measurements.
One of the problems of double skin façades is to satisfy the need for sum-
AN
mer comfort, so it is interesting to look at the temperature drop achievable
through effective management of the shading and ventilation of the double
M
skin façade. The management of these two design parameters has an impor-
tant impact on the indoor conditions and the level of daylight.
D

When taking into account energy gains of the double skin façade, it is
TE

also interesting to know the cost of the mechanical ventilation of the DSF
compared to several sides in terms of electricity bill for artificial lighting or
the use of an air conditioning system.
EP

The study of the indoor glazing temperature can give interesting insights
into the control of the two parameters (air flow and tilt angle of the blades).
C

There seems to be a great interest in using mechanical ventilation relative

AC

to natural ventilation. The gain in temperature for the inner pane between
natural ventilation and mechanical ventilation with an air flow of 200m3 h−1
varies between 1.8◦ C for α = 0◦ and 3.8◦ C for α = 90◦ . The mean air drop
temperature may be considered to be around 2◦ C.

22
 ACCEPTED MANUSCRIPT

It also seems interesting to use a flow of 400m3 h−1 rather than using
200m3 h−1 because the decrease in temperature on the inner pane is about

PT
1.1◦ C to 1.2◦ C, values which could have an impact on the thermal comfort
amelioration. However, it seems less interesting to increase the ventilation

RI
rate from 400m3 h−1 to higher values because only a 0.7◦ C difference was
noticed. The inner pane goes from a temperature of 40.6◦C, for α = 0◦

SC
in natural ventilation, to 32.4◦ C for α = 90◦ and D = 600m3 h−1 , which
represents a gain of 8.2◦ C on the surface of the glazing with high impact

U
on the indoor climate. This study therefore shows that it is important to
optimize the use of ventilation and shading in a double skin façade.
AN
6. Conclusions
M
A numerical modelling of a DSF has been developed in this paper, based
on a zonal model approach. This model includes convective, radiative and
D

aeraulic heat transfers. A full scale DSF has been monitored to test different
TE

configurations of airflow rates and blades angles. The first result of our study
is the validation of the numerical modelling using the experimental data.
EP

The results were analysed to determine management regulations for both

parameters studied (Venetian blind angle and ventilation rate) to obtain a
temperature of the interior glazing within the lowest possible value. Even for
C

the complex behaviour of the DSF, the numerical model could optimize the
AC

two parameters, and optimal conditions of functioning could be attained. A

more comprehensive approach should also consider the energy cost of venti-
lation and the cost of artificial lighting if necessary.
The next step of our work will be to analyze the coupling between the DSF

23
 ACCEPTED MANUSCRIPT

numerical modelling developed in the present paper and an office building

numerical modelling. The final objective is to optimize the control of the

PT
DSF.

RI
References

References

SC
[1] M. Shameri, M. Alghoul, K. Sopian, M. F. M. Zain, O. Elayeb, Per-
spectives of double skin façade systems in buildings and energy saving,

U
Renewable and Sustainable Energy Reviews 15 (3) (2011) 1468 – 1475.
AN
[2] J. S. Carlos, H. Corvacho, P. D. Silva, J. Castro-Gomes, Modelling and
simulation of a ventilated double window, Applied Thermal Engineering
M
31 (1) (2011) 93 – 102.
D

[3] N. Hashemi, R. Fayaz, M. Sarshar, Thermal behaviour of a ventilated

double skin façade in hot arid climate, Energy and Buildings 42 (10)
TE

(2010) 1823 – 1832.

[4] L. Xu, T. Ojima, Field experiments on natural energy utilization in a

EP

residential house with a double skin façade system, Building and Envi-
ronment 42 (5) (2007) 2014 – 2023.
C

[5] H. Poizaris, Double skin façade a litterature review. Report of IEA SHC
AC

TASK 34 ECBCS ANNEX 43, Tech. rep., Department of Architecture

and Built Environment, Division of Energy and Building Design. Lund
University, Lund Institute of Technology, Sweden. (2006).

24
 ACCEPTED MANUSCRIPT

[6] I. Perez-Grande, J. Meseguer, G. Alonso, Influence of glass properties on

the performance of double-glazed façades, Applied Thermal Engineering

PT
25 (17-18) (2005) 3163 – 3175.

[7] N. Hamza, C. Underwood, CFD supported modelling of double skin

RI
façades in hot arid climates, in: Ninth International IBPSA Conference,
Montreal, Canada, August 15-18, 2005, pp. 365–372.

SC
[8] E. Gratia, A. D. Herde, Optimal operation of a south double-skin façade,
Energy and Buildings 36 (1) (2004) 41 – 60.

U
AN
[9] C. S. Park, G. Augenbroe, M. Thitisiwat, Integrated control strategied
for double skin systems, in: Ninth International IBPSA Conference,
Montreal, Canada, August 15-18, 2005, pp. 929–936.
M

[10] N. Safer, M. Woloszyn, J. J. Roux, G. Rusaouen, F. Kuznik, Modelling

D

the double skin façades for building energy simulations: radiative and
convective heat transfer, in: Ninth International IBPSA Conference,
TE

Montreal, Canada, August 15-18, 2005, pp. 1067–1074.

EP

[11] N. Safer, M. Woloszyn, J. J. Roux, Three-dimensional simulation with

a CFD tool of the airflow phenomena in single floor double-skin façade
equipped with a venetian blind, Solar Energy 79 (2) (2005) 193 – 203.
C
AC

[12] C. Balocco, M. Colombari, Thermal behaviour of interactive mechani-

cally ventilated double glazed façade: Non-dimensional analysis, Energy
and Buildings 38 (1) (2006) 1 – 7.

25
 ACCEPTED MANUSCRIPT

[13] C.-S. Park, G. Augenbroe, T. Messadi, M. Thitisawat, N. Sadegh, Cal-

ibration of a lumped simulation model for double-skin façade systems,

PT
Energy and Buildings 36 (11) (2004) 1117 – 1130.

[14] G. Gan, Thermal transmittance of multiple glazing: computational fluid

RI
dynamics prediction, Applied Thermal Engineering 21 (15) (2001) 1583
– 1592.

SC
[15] H. Manz, T. Frank, Thermal simulation of buildings with double-skin
façades, Energy and Buildings 37 (11) (2005) 1114 – 1121, research That

U
Inspires 125 Years of EMPA.
AN
[16] D. Saelens, Energy performance assessment of single storey multiple-skin
façades, Ph.D. thesis, Katholieke Universiteit Leuven (2002).
M
[17] T. E. Jiru, F. Haghighat, Modeling ventilated double skin façade–a zonal
approach, Energy and Buildings 40 (8) (2008) 1567 – 1576.
D
TE

[18] W. J. Stec, A. H. C. V. Paassen, Defining the performance of the double

skin façade with the use of the simulation model, in: Eight International
IBPSA Conference, Eindhoven, Netherlands, August 11-14, 2003, pp.
EP

1243–1250.

[19] V. Gavan, M. Woloszyn, F. Kuznik, J.-J. Roux, Experimental study of

C

a mechanically ventilated double-skin façade with venetian sun-shading

AC

device: A full-scale investigation in controlled environment, Solar En-

ergy 84 (2) (2010) 183 – 195.

[20] N. Safer, Numerical modelling of double skin façades equipped with

solar screens - Multiscale approach (in french), Ph.D. thesis, Centre de

26
 ACCEPTED MANUSCRIPT

Thermique de Lyon, département de génie civil et urbanisme INSA de

Lyon. France (2006).

PT
[21] The French Thermal Regulation 2005 - French Ministry of Sustainable
Development, France (2005).

RI
[22] A. P. Faist, The double skin façade (in french), Tech. rep., Report of the

SC
Ecole Polytechnique Federal (EPF) de Lausanne, Lausanne, Switzerland
(1998).

U
AN
M
D
TE
C EP
AC

27
 ACCEPTED MANUSCRIPT

List of Tables
1 Model coefficients in pressure depending on the angle of incli-

PT
nation of the blinds. . . . . . . . . . . . . . . . . . . . . . . . 29
2 Double skin façade properties. . . . . . . . . . . . . . . . . . . 30
3 Air temperature in the DSF for different positions on the x-

RI
axis - 400m3 h−1 . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4 Inner glazing temperature drop for different air flows in the
DSF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

U SC
AN
M
D
TE
C EP
AC

28
 ACCEPTED MANUSCRIPT

PT
RI
U SC
0
AN
α[◦ ] Kb
18
Ka
20
Ki
20
30 12 20 20
M
45 9 20 20
60 6 20 20
90 1 40 40
D

Table 1: Model coefficients in pressure depending on the angle of inclination of the blinds.
TE
C EP
AC

29
 ACCEPTED MANUSCRIPT

PT
RI
U SC
layer
AN
dimensions transmissivity absorptivity
width [m] height [m]
external DSF glass pane 2.6 2.3 0.82 0.11
M
Venetian blade 2.6 - 0 0.1
internal DSF glass pane 2.6 2.3 0.82 0.11
D

Table 2: Double skin façade properties.

TE
C EP
AC

30
 ACCEPTED MANUSCRIPT

PT
RI
SC
DSF width [m] 0 0.05 0.1 0.15 0.2

U
ext. left ch. blades right ch. int
angle [◦ ]

Model
0
30
AN36.2
37.0
36.3
37.1
38.2
40.1
37.2
37.7
39.1
38.1
45 37.3 37.1 40.4 37.5 37.3
results
60 37.5 36.9 40.1 37.2 36.3
M
90 37.7 35.8 38.6 36.3 34.3
0 37.5 36.2 37.9 36.8 39.3
30 38.2 37.2 40.5 36.6 37.8
D

Experimental 45 38.4 373 40.7 36.6 37.6

60 38.8 37.4 41.3 36.1 36.8
TE

data
90 39.1 37.4 41.2 36.1 36.4
Table 3: Air temperature in the DSF for different positions on the x-axis - 400m3 h−1 .
C EP
AC

31
 ACCEPTED MANUSCRIPT

PT
RI
U SC
angle [◦ ]
0
0m3 h
40.5 38.7
AN
200m3 h difference
1.8
400m3 h
37.6
difference
1.1
600m3 h
36.9
difference
0.7
30 39.8 37.7 2.1 36.5 1.2 35.7 0.8
M
45 39.0 36.9 2.1 35.7 1.2 34.9 0.8
60 38.0 36.0 2.0 34.8 1.1 34.1 0.7
90 37.7 33.8 3.8 33.2 0.6 32.6 0.6
D

Table 4: Inner glazing temperature drop for different air flows in the DSF.
TE
C EP
AC

32
 ACCEPTED MANUSCRIPT

List of Figures
1 Schematic diagram of a double skin façade according to Faist [22]. 34

PT
2 Isotherm zone in a DSF predicted by CFD - Extracted from [20]. 35
3 Double skin façade layers discretization and variables. . . . . . 36
4 Hottel method applied to the double skin façade: e = 0.025m

RI
spacing between the blinds, l = 0.025m blind’s width and β
blind’s angle. . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5 Schematic of different surfaces considered for the calculation

SC
of the irradiation in short wavelength. . . . . . . . . . . . . . . 38
6 Convection coefficients of various air-surface interfaces of the
DSF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
7 Model resolution algorithm. . . . . . . . . . . . . . . . . . . . 40

U
8 Test cell facility scheme: 1, cooling unit; 2, weather generator;
3, double skin façade; 4, protection glass of the solar simula-
AN
tor; 5, test cell; 6, concrete; 7, air blowing plenum; 8, solar
simulator’s heat removal ventilators; 9, air extraction plenum;
10, HVAC unit of the buffer zone; 11, solar simulator; 12,
M
controlled buffer zone. (Metric units) [16]. . . . . . . . . . . . 41
9 DSF section view: 1, 6 mm glass layer; 2, solar protections;
3, metallic support; 4, solar protection action box; 5, concrete
D

beam; 6, aluminum frame; 7, pane opening articulation; 8,

ventilation openings. (Metric units) [16]. . . . . . . . . . . . . 42
TE

10 Air temperature profile in the left channel of the DSF for an

airflow equal to 400m3 h−1 - circle: experimental data and line:
numerical modelling. . . . . . . . . . . . . . . . . . . . . . . . 43
11 Air temperature profile in the right channel of the DSF for
EP

an airflow equal to 400m3 h−1 - circle: experimental data and

line: numerical modelling. . . . . . . . . . . . . . . . . . . . . 44
12 Venetian blades temperature for an airflow equal to 400m3 h−1
C

- circle: experimental data and line: numerical modelling. . . . 45

13 Temperature variation in the DSF for different air flows and
AC

blades tilt angles - Numerical modelling. . . . . . . . . . . . . 46

33
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE

Figure 1: Schematic diagram of a double skin façade according to Faist [22].

C EP
AC

34
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP

Figure 2: Isotherm zone in a DSF predicted by CFD - Extracted from [20].

AC

35
 ACCEPTED MANUSCRIPT

PT
RI
SC
blinds
Ext. Int.

U
Tair(i+2,1) Tair(i+2,2) e
AN
p(i+1,1) Qc p(i+1,2)
M
Tb(i) Qa
p(i,1) p(i,2) T2(i)
D

p(i-1,1) Qb p(i-1,2)
TE
EP

Figure 3: Double skin façade layers discretization and variables.

C
AC

36
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
C EP

Figure 4: Hottel method applied to the double skin façade: e = 0.025m spacing between
the blinds, l = 0.025m blind’s width and β blind’s angle.
AC

37
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP

Figure 5: Schematic of different surfaces considered for the calculation of the irradiation
in short wavelength.
C
AC

38
 ACCEPTED MANUSCRIPT

PT
RI
SC
Ext. Int.

U
hc1
=
AN hc3
=
hcint
=
11Wm-2K-1 11Wm-2K-1 5Wm-2K-1
M
D

hcext hc2 hc4

TE

= = =
20Wm-2K-1 11Wm-2K-1 11Wm-2K-1
EP

Figure 6: Convection coefficients of various air-surface interfaces of the DSF.

C
AC

39
 ACCEPTED MANUSCRIPT

PT
RI
SC
Initialization of iterations

U
PRESSURE MODEL CONVECTION MODEL RADIATIVE MODEL
mass balance

Pressure field
Airflow distribution
AN
Convective heat flux SW/LW radiative heat flux
M
Enthalpy balance Thermal balance New iteration

Air temperature field

Surface temperature No
D

Convergence

Yes
TE

Results
EP

Figure 7: Model resolution algorithm.

C
AC

40
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP

Figure 8: Test cell facility scheme: 1, cooling unit; 2, weather generator; 3, double skin
façade; 4, protection glass of the solar simulator; 5, test cell; 6, concrete; 7, air blowing
plenum; 8, solar simulator’s heat removal ventilators; 9, air extraction plenum; 10, HVAC
C

unit of the buffer zone; 11, solar simulator; 12, controlled buffer zone. (Metric units) [16].
AC

41
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP

Figure 9: DSF section view: 1, 6 mm glass layer; 2, solar protections; 3, metallic support;
4, solar protection action box; 5, concrete beam; 6, aluminum frame; 7, pane opening
articulation; 8, ventilation openings. (Metric units) [16].
C
AC

42
 ACCEPTED MANUSCRIPT

PT
blades angle 0° blades angle 30°
2.5 2.5

2 2

RI
DSF height [m]

DSF height [m]

1.5 1.5

SC
1 1

U
0.5 0.5

0
32 34 36
temperature [°C]
AN
38
0
32 34 36
temperature [°C]
38
M
blades angle 45° blades angle 60°
2.5 2.5
D

2 2
TE
DSF height [m]

DSF height [m]

1.5 1.5
EP

1 1

0.5 0.5
C

0 0
AC

32 34 36 38 32 34 36 38
temperature [°C] temperature [°C]

Figure 10: Air temperature profile in the left channel of the DSF for an airflow equal to
400m3h−1 - circle: experimental data and line: numerical modelling.

43
 ACCEPTED MANUSCRIPT

PT
blades angle 0° blades angle 30°
2.5 2.5

2 2

RI
DSF height [m]

DSF height [m]

1.5 1.5

SC
1 1

U
0.5 0.5

0
32 34 36
temperature [°C]
AN
38
0
32 34 36
temperature [°C]
38
M
blades angle 45° blades angle 60°
2.5 2.5
D

2 2
TE
DSF height [m]

DSF height [m]

1.5 1.5
EP

1 1

0.5 0.5
C

0 0
AC

32 34 36 38 32 34 36 38
temperature [°C] temperature [°C]

Figure 11: Air temperature profile in the right channel of the DSF for an airflow equal to
400m3h−1 - circle: experimental data and line: numerical modelling.

44
 ACCEPTED MANUSCRIPT

PT
blades angle 0° blades angle 30°
2.5 2.5

2 2

RI
DSF height [m]

DSF height [m]

1.5 1.5

SC
1 1

U
0.5 0.5

0
34 36 38
temperature [°C]
40
AN
42
0
34 36 38 40
temperature [°C]
42
M
blades angle 45° blades angle 60°
2.5 2.5
D

2 2
TE
DSF height [m]

DSF height [m]

1.5 1.5
EP

1 1

0.5 0.5
C

0 0
AC

34 36 38 40 42 34 36 38 40 42
temperature [°C] temperature [°C]

Figure 12: Venetian blades temperature for an airflow equal to 400m3 h−1 - circle: exper-
imental data and line: numerical modelling.

45
 ACCEPTED MANUSCRIPT

PT
RI
flow rate 200m3/h flow rate 400m3/h
43 43

42 42

SC
41 41

40 40
temperature [°C]

temperature [°C]
39 39

38 38

37 37

U
36 36

35 35

34

33

32
0 20
external glazing
venitian blades
internal glazing

40
blades tilt angle [°]
60 80
AN 34

33

32
0 20
external glazing
venitian blades
internal glazing

40 60
blades tilt angle [°]
80
M
flow rate 600m3/h
43

42

41
D

40
temperature [°C]

39

38
TE

37

36

35

34 external glazing
EP

venitian blades
33 internal glazing

32
0 20 40 60 80
blades tilt angle [°]
C

Figure 13: Temperature variation in the DSF for different air flows and blades tilt angles
AC

- Numerical modelling.

46