You are on page 1of 22

INTERNATIONAL JOURNAL OF ENERGY RESEARCH

Int. J. Energy Res. 2002; 26:611}632 (DOI: 10.1002/er.807)

Heat losses through building walls with closed, open


and deformable cavities

Sylvie Lorente*
Department of Civil Engineering, National Institute of Applied Sciences (INSA), 135 Avenue de Rangueil,
31077 Toulouse, France

SUMMARY

This article reviews a recent body of work that documents the heat #ow through walls with relatively
complicated internal structure. The "rst part presents the fundamentals of natural convection heat transfer
in two-dimensional enclosures "lled with air. Numerical simulations for Rayleigh numbers in the range
6000}30000 and aspect ratio A"40 show that the air circulation is periodic (pulsating) and the core space is
dominated by almost equidistant rolls. The heat transfer e!ected by such #ows is documented, and the
application to walls with tall cavities (hollow bricks) is discussed. The paper continues with a study of heat
transfer through a ventilated wall heated by solar radiation from the side. The heated side is ventilated by an
air channel opened at both ends. It is shown that the design of the air channel has a signi"cant e!ect on the
share of the radiative heat input intercepted by the air #ow. The concluding part of the paper is a "rst-time
review of new work on natural convection heat transfer across elongated vertical cavities with deformed
side walls (e.g. air gaps in double-pane windows). It is shown that when the side walls are bent inward so that
the cavity is narrower at mid-height than at its top and bottom ends, the total heat transfer rate through the
system is increased signi"cantly. The deformation of the walls also a!ects the intensity and structure of the
buoyancy-driven #ow in the air space. Overall, this article reviews some of the newest work on heat losses
through complicated wall structures and projects it on the background provided by the existing literature.
Copyright  2002 John Wiley & Sons, Ltd.

KEY WORDS: heat #ow through walls; heat loss; building walls

1. OBJECTIVE

In this article, a recent body of work is reviewed that documents the loss of heat through relatively
complicated internal wall geometries: closed, open and deformable cavities. This work is new on
the background o!ered by the very active and important "eld of research dedicated to heat and
#uid #ow at the interface between energy systems and the environment.
The global interest in energy conservation has stimulated a steady stream of fundamental and
applied studies on how energy is lost*how heat &leaks'*from the systems in which energy
currents are essential. Key advances are cited in the next section. This activity has placed the focus

* Correspondence to: Sylvie Lorente, Department of Civil Engineering, National Institute of Applied Sciences (INSA),
135 Avenue de Rangueil, 31077 Toulouse, France.

Received 16 January 2001


Copyright  2002 John Wiley & Sons, Ltd. Accepted 4 June 2001
612 S. LORENTE

on the interaction between energy systems and their immediate #uid environments. The #ow of
energy (heating) drives the #ow of #uids, and, in turn, #owing #uids convect energy. To predict
the rate of heat loss e!ected by #uid #ow is one goal. Another is to interpret this result for the
purpose of design: the development of structural changes in the wall of the system so that in future
designs the loss of heat is minimized. The work reviewed in this article covers both aspects with
application to the loss of heat through walls of buildings. The same mechanism, or the same
&coupling' between #owing energy currents and #owing #uids stands behind the large-scale
environmental #ows reviewed in other articles in this issue.

2. NATURAL CONVECTION IN CAVITIES WITH DIFFERENT SIDE WALL


TEMPERATURES

2.1. Review of previous works


Convection heat transfer is an important mechanism in the total heat exchange measured in
vertical enclosures heated from the side. In most of the enclosures used in building walls, the #ow
is generally laminar and unicellular. Fluid movement in the cavity is due to buoyancy forces
resulting from a temperature di!erence between both vertical surfaces. There are situations,
always in the laminar regime, in which secondary cells appear in the core cavity. This multicellu-
lar #ow tends to increase local and average heat transfer coe$cients.
The fundamentals of convective heat transfer were reviewed by Bejan (1995). A rigorous
analysis of #uid #ows can be found in the book by Padet (1991). Theoretical and numerical
studies by Raithby and Wong (1981), de Vahl Davis (1968), Korpela et al. (1982) re"ned the
resolution of the problem, i.e. the accuracy of heat transfer calculations. These studies showed
that the solution is a function of three dimensionless parameters: the aspect ratio of the cavity A,
the Prandtl number of the #uid, Pr and the Rayleigh number, Ra. These and other parameters are
de"ned in the Nomenclature.
The speci"c e!ect of multicellular #ow was described by Elder (1965), who observed experi-
mentally a secondary #ow pattern attributable to hydrodynamic instability. Several numerical
models have been able to resolve secondary #ow in a vertical slot. Some authors point out that
the onset of secondary cells can be delayed by the false di!usion resulting from numerical
upwinding schemes. Others use a central di!erence discretization scheme and uniform grids
(e.g. Lee and Korpela, 1983; Ramanan and Korpela, 1989; Lauriat and Desrayaud, 1985; Roux
et al., 1980; Wright and Sullivan, 1994; Le QueH reH , 1990). In particular, Lee and Korpela (1983)
found the expression of the critical value of the Grashof number (Gr"Ra/Pr) at which the onset
of secondary cells takes place. Several studies document the number of secondary cells in the core
of the cavity according to the Rayleigh number (Lee and Korpela, 1983; Wright and Sullivan,
1994; Wakitani, 1997). Wright and Sullivan (1989) performed a literature review on natural
convection in Insulated Glazing Units (IGU). Studies by Zhao et al. (1997) shed considerable light
on the multicellular #ows that form in laminar regime. Their work provided a map of the
appearance of these cells for several values of Rayleigh number and aspect ratio (Figure 1).

2.2. Mathematical formulation


Figure 2 presents the geometry of a two-dimensional rectangular enclosure. The laminar two-
dimensional air #ow can be put in a mathematical form by assuming that the #uid is Newtonian

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 613

100

Yin et al. (1978): limit


of transition from laminar
regime to turbulent regime
Zhao et al. (1996): limit
of transition to multicellular
regime
40 cases studied by
A=H/L

Lartigue et al. (2000)

Multicellular regime

Turbulent regime

Laminar regime

10
10 3 10 4 10 5 10 6 10 7

Ra
Figure 1. Flow regimes de"nitions for natural convection in a tall enclosure with side walls at di!erent
temperatures (Lartigue et al., 2000).

Figure 2. The geometry and boundary conditions of the two-dimensional cavity


with di!erent side-wall temperatures.

and satis"es the Boussinesq approximation. It is governed by the laws of conservation of mass,
momentum and energy. These equations can be put into dimensionless form using the following
dimensionless variables:

x z H
X" , Z" , A " ,
¸ ¸ ¸

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
614 S. LORENTE

u w
;" , =" with < "(g¹¸)
< < 
  (1)
p ¹!¹ ¹ #¹
P" " with ¹ "   and ¹"¹ !¹
< ¹ 2  


 
t ¸ ¸ 
tH" with t " "
t  < g¹
 
The governing equations assume the following dimensionless form:

; =
# "0 (2)
X Z

   
; ; ; P Pr  ; ;
#; #= "! # # (3)
tH X Z X Ra X Z

   
= = = P Pr  = =
#; #= "! # # # (4)
tH X Z Z Ra X Z

 
    
#; #= "(Ra ) Pr)\ # (5)
tH X Z X Z

The boundary conditions of velocity and temperature become

;"="0 at X"0, X"1 and Z"0, Z"A (6)

"0.5 at X"0 (7)

"!0.5 at X"1 (8)

/Z"0 at Z"0 and Z"A (9)

The solution to Equations (2)}(5) was developed using the commercially available computer
code ESTET (Ensemble de Simulations Tridimensionnelles d'Ecoulements Turbulents), which
was developed by ElectriciteH de France. ESTET solves the equations of #uid mechanics, which are
discretized in a structured grid that uses "nite volumes and "nite di!erences. Grid independence
tests revealed that approximately 30 000 grid nodes were necessary to obtain a grid-independent
solution when the aspect ratio A is equal to 40. To ensure the convergence of calculation, it is
imperative to place the "rst nodes within the viscous boundary layer of the walls, which is
the region with the greatest temperature gradients. For this reason, the spacing between
the "rst nodes was 0.02. The boundary conditions imposed were: non-slip, impermeability and
adiabatic transfer on the top and bottom of the cavity, and non-slip and impermeability on
the vertical walls.
The heat transfer literature contains many results on the thermal "eld in a high aspect ratio
cavity with natural convection. This work began with Eckert and Carlson (1961), and is

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 615

Figure 3. Schematic diagram of the experimental set-up (Lartigue et al., 2000).

summarized in more recent reviews (Kakac and Yener, 1994). In the present review the focus is on
dynamic "eld results obtained using the experimental set-up described below.

2.3. Experiment
The experimental apparatus is a parallelepipedic closed cavity, "lled with air. The main vertical
plates are made of aluminium, to approach the isothermal boundary condition. The two other
vertical plates are made of glass, to allow the passing of the laser sheet of the particle image
velocimetry (PIV). The bottom and top horizontal plates are made of PVC to limit thermal short-
circuiting. A zero heat #ux boundary condition is very di$cult to reach experimentally. Neverthe-
less, in view of the work of Raithby and Wong (1981), if the aspect ratio is large enough (A*40),
the in#uence of the horizontal boundary conditions (perfectly adiabatic or perfectly conducting)
on the Nusselt number is small. The temperatures of the aluminium plates are regulated by heat
exchangers made of rubber pipes, which are glued on the back of the plates. Water and refrigerant
circulate in the heat exchangers to maintain desired conditions.
The PIV system is based on the relation velocity"distance/time. Thus, the velocity is derived
from the measurement of the travelling distance of a #uid particle in a given time interval. It is
necessary to seed the #ow with small tracer particles of the same density as air, which travel with
the #ow, and which can be visualized. Incense smoke was chosen as seeding material. A two-
dimensional slice of the #ow "eld is illuminated by a light sheet in the middle of the cavity, as
shown in Figure 3. The illuminated seeding scatters the light, which is detected by a camera

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
616 S. LORENTE

placed at right angle to the light sheet. Two camera images are recorded, the "rst showing the
initial position of the seeding particles, and the second showing their "nal position due to the
movement of the #ow "eld. The time between the recording images is known.
The two camera frames are then processed to "nd the velocity vector map of the #ow "eld.
They are divided into smaller regions that are considered individually. Correlation and Fourier
transform methods are used to measure the average displacement of the ensemble of particles in
a "nite-size region. The #uid velocity is then calculated over the time interval between the
successive images. This is repeated for each region, so the entire two-dimensional velocity vector
map is constructed. One hundred velocity vector maps are constructed in this manner during
every second.

3. THERMAL FIELD

The temperature "eld in the cavity was obtained numerically (Lartigue, 1999). The Nusselt
number was calculated in order to evaluate the rate of heat transfer across the enclosure. Local
Nusselt numbers are de"ned as the ratio of convective heat transfer over conductive heat transfer:

h(z)x
Nu(z)" (10)
k(¹)

(z) 1 ¹ !¹(x, z)
h(z)" " k(¹)  (11)
¹ !¹ ¹ !¹ x
   
where (z) is the wall heat #ux calculated on the hot face in the viscous boundary layer (in this
case, for example, x"0.02¸). The value of k is evaluated locally along the vertical wall according
to the temperature. The aspect ratio used for all the presented simulations is "xed at A"40,
which agrees with con"gurations found in practice. Figures 4(a)}4(e) show the evolution of the
local Nusselt number along the hot wall, as a function of the dimensionless height, and in the
Rayleigh number range 3550}17 750 (this Ra range was chosen in order to compare the results
with those found in the literature). When the Rayleigh number is equal to 3550, the #ow is in the
conductive regime, according to Batchelor (1954). The Nusselt number is then equal to 1 over the
height of the cavity, except close to the horizontal walls. If the Rayleigh number increases to
roughly 6000, the #ow becomes multicellular. The #ow is the seat of instabilities, which are
evident in Figure 4(b). Thus, for Ra"6800, the local Nusselt number tends to oscillate in
a regular manner around the value corresponding to the conductive regime. The same behaviour
is observed for Ra"10 102, which resembles closely the experimental conditions. If the oscilla-
tions remain regular, their amplitude grows and the average Nusselt number exceeds 1. If the
value of Rayleigh number increases further (Figures 4(d) and 4(e)), the regularity of the oscilla-
tions is destroyed and the multicellular laminar #ow becomes fully established before becoming
turbulent.

4. VELOCITY FIELD

In this section, the numerical and experimental results that describe the velocity "eld are
reviewed. Prior to this work, instabilities in laminar #ow were highlighted only with temperature

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 617

Figure 4. The dependence of the local Nusselt number on the dimensionless height Z/A for A"40 and
several values of Ra (Lartigue et al., 2000).

measurements. The study carried out by Lartigue et al. (2000) describes the results obtained using
a non-intrusive method of measure: the PIV.
Figure 5 shows the numerical streamlines in the cavity for Rayleigh numbers ranging from 3550
to 17 750. For a Rayleigh number of 3550 no #uctuations are observed, and this means that the
#ow is unicellular. When the Rayleigh number increases, the stream patterns begin to #uctuate
indicating the onset of the instability, and the #ow becomes multicellular. These numerical results

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
618 S. LORENTE

Figure 5. Numerical streamlines for A"40: (a) Ra"3550, (b) Ra"6800, (c) Ra"10 102, (d) Ra"14 200
and (e) Ra"17 750 (Lartigue et al., 2000).

Table I. Number of cells obtained in di!erent numerical simu-


lations, as a function of Ra (Lartigue et al., 2000).

Ra Lartigue et al. (2000) References

6800 16 15 Wright and Sullivan (1994)


10 102 14 14 Wright and Sullivan (1994)
14 200 13 13 Lee and Korpela (1983)
17 750 11 13 Lee and Korpela (1983)

are compared in Table I with the results available in other papers. The table lists the number of
cells obtained in the simulations of previous authors and the results obtained by our group. These
data show that the number of cells decreases as the Rayleigh number increases.
Figure 6 shows the steady-state horizontal and vertical components of velocity obtained
experimentally and numerically. The cavity core represents a zone ranging from approximately

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 619

Figure 6. Non-dimensionalized horizontal and vertical components of velocity in experimental (exp) and
numerical (num) cases in the core of the cavity (Lartigue et al., 2000).

Z/A"0.15 to Z/A"0.85. Non-dimensionalized horizontal and vertical components of the


velocity vector are presented here for several heights of the measurement window. The experi-
mental and numerical values of velocity are in close agreement. Two zones of the same intensity,

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
620 S. LORENTE

Figure 7. Usual brick, and brick with many vertical cavities (Lorente et al., 1996).

and the opposed direction of the component ; coupled with the component =, indicate the
presence of a cell. Thus, from the results obtained, we note that the velocity of a secondary cell is
practically reduced to its horizontal component in the centreline of the cavity (X"0.5).
Figure 6(a) shows the existence of two half cells in the test area between z/H"0.79 and 0.84. In
the lower part of the cell the horizontal component of velocity is directed from the cold face
toward the hot face. Closer to the hot boundary layer, the vertical component is oriented upward
and dominates. It is the reverse in the upper part of the cell, where the horizontal component is
directed toward the cold face. In the vicinity of the cold boundary layer, the velocity vector is
reduced to its downward vertical component. Consequently, in this case, the rotation of the
secondary cell is oriented in the trigonometric sense. Thus, experimental and numerical results
make it possible to highlight the existence of secondary cells and their rotation.

5. HOLLOW BRICKS

The physical phenomenon studied is important in practice, as vertical cavities of high aspect ratio
are encountered frequently in building components. The #ow in the cavity is laminar, since the
transition from laminar to turbulent #ow is around Ra"40 000 when the aspect ratio is equal to
40 (Figure 1). Below Ra+6000, the energy transfer from the hot face toward the cold face is
carried out directly by conduction. As the Rayleigh number increases, conduction is no longer
able to transfer the imposed heat. The establishment of a secondary #ow (secondary cells) has the
role of mitigating this de"cit, permitting the transfer of heat by convection. The observed cells are
ordered and characterized by a very slow average movement. The experimental and numerical
results also reveal a downward #ow. If the Rayleigh number increases further, these structures
lose their ordered character and the #ow enters on the turbulent regime.
Usually, terra-cotta bricks are made with horizontal cavities. When the product is designed
with vertical cavities, less material is necessary in order to achieve the same mechanical resistance.
In the case of vertical hollows bricks (Figure 7), the heat transfer di!ers from heat transfer through
usual bricks because the aspect ratio is large enough to allow #uid motion: an important part of
heat transfer is due to natural convection.
In the "rst part of this paper, results on the thermal "eld inside a high aspect ratio enclosure
were presented. The development of such results requires the use of a #uid mechanics code, which
demands lots of computational time and high-performance computers. In the case of an assembly
of enclosures (a brick with many cavities), the total heat transferred through the wall cannot be
determined with this kind of numerical code, because of insu$cient time and/or computer
memory. Consequently, the thermal "eld in each cavity is approximated with a good accuracy by
the Karman}Pohlhausen method (Lorente et al., 1996).

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 621

INPUT: higher and lower temperatures,


terra-cotta conductivity, geometric
characteristics (height, width, thickness...),
number of cavities along Ox and Oy axis

inner surface temperature conductive heat transfer


distribution (terra-cotta sides)

convective radiative heat


heat transfer transfer

heat transfer
through the brick

steady state
condition

NO YES

thermal resistance of the product

Figure 8. Algorithm for assembling the heat currents (conduction, convection, radiation) for global heat
transfer through a brick with many cavities (Lorente et al., 1996).

The heat transfer through the brick is obtained by calculating the convective and radiative heat
transfer through each cavity, the conductive heat transfer along the terra-cotta ribs, and assembl-
ing these currents using an electrical network, in series and in parallel (Lorente et al., 1996). An
example of the algorithm is given in Figure 8. Once the thermal "eld is known, the convective heat
transfer is calculated locally using the temperature pro"les of the boundary layers. Local values
are integrated along the vertical faces to obtain the averaged values. The radiosity method (Siegel
and Howell, 1981) is used to determine the radiative heat #ux, and the Fourier law is used for
conductive heat transfer. This analytical procedure was carefully validated by numerous experi-
ments on cavities with several sizes, which were assembled in hollow bricks.
A brick is usually 20 or 25 cm high, while a wall has a height of roughly 250 cm (the height of
a #oor). This means that the aspect ratio of each continuous enclosure is multiplied by 10, and,
consequently, the heat transfer by natural convection will be di!erent. We performed calculations
using the analytical model described above (Lorente et al., 1996) and we compared the thermal

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
622 S. LORENTE

0,9

0,8 H = 25 cm

Rth (m² ˚C/W)


H = 250 cm
0,7

0,6

0,5

0,4

0,3
10 20 30 40 50

(Th-Tc) (K)

Figure 9. The evolution of the global thermal resistance through bricks with
di!erent heights (Lorente et al., 1998).

resistance of 25 or 250 cm high bricks. The total thickness of the bricks remains constant but
the number of enclosures (and thus their thickness) could vary. The cavity width was always
greater than the air-space thickness, so that the natural convection pattern was essentially
two-dimensional.
Figure 9 shows a sample of the results obtained for several values of the temperature di!erence
between both sides of bricks. When ¹ is increased, the thermal resistance of the 25 cm high brick
decreases by 27 per cent while this decrease is less than 10 per cent for the 250 cm high brick. In
both cases, the decrease is essentially due to the convective transfer. Indeed, if we analyse the
natural convective transfer phenomenon inside enclosures (with constant air thickness and
constant temperature di!erence), the average heat #ux density by convection is smaller in a higher
cavity. For our range of Rayleigh numbers the #ow regime is the boundary layer type. When the
height of the cavities increases boundary layers thicken and temperature gradients decrease.
Thus, the global heat rate by convection is weaker. This behaviour was con"rmed in other
con"gurations as well (Lorente et al., 1998).

6. INFLUENCE OF MOISTURE ON HEAT TRANSFER

The heat transfer results presented in the preceding section are not related in any way to the
notion of moisture level. Building components are exposed to moisture level di!erences as much
as temperature di!erences, between the inside of the brick and the outside.
We chose to work on a basic element containing a single vertical enclosure (25 cm high), where
the boundaries are made of terra-cotta (Figure 10). Calculations were made using a numerical
code (HYGRO, developed by CEBTP, France) where the system of equations linking heat and
mass transfer is solved based on the "nite-di!erence method. A description of this code can be
found in Vasile et al. (1998). We focused on several boundary conditions: temperature di!erence
as well as relative moisture di!erence between the cold and hot faces.

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 623

adiabatic surface

cold side hot side

adiabatic surface

Figure 10. Plan view of the terra-cotta element studied numerically (Vasile et al., 1998).

140
120
He at Flux (W/m ² )

100
∆ = 20 K
DT
80 ∆ = 15 K
DT
60 ∆ =5K
DT
40
20
0
500 700 900 1100 1300 1500
∆Pv (Pa)

Figure 11. The e!ect of the vapour pressure di!erence on the heat #ux (Vasile et al., 1998).

Figure 11 shows the results obtained for the dependence of heat #ux on vapour pressure
di!erence. The heat #ux through a vertical hollow terra-cotta element increases with the vapour
pressure di!erence. The more the temperature rises, the greater is the increase. These results stress
the high sensitivity of the heat #ux to the material moisture concentration level, as the heat #ux
di!erences between a low-water content state and one with a more concentrated level are about
35 per cent.

7. OPEN VERTICAL CAVITIES: VENTILATED WALLS

Passive systems were developed since the 1960s: a good example is the Trombe wall, which has
been studied all over the world, and which is still studied (Detunq and Bilgen, 1984; Du$n and
Knowles, 1985; Ben Yedder and Bilgen, 1991; Buzzoni et al., 1998). Most of these systems were
meant for winter conditions. Indeed, in the case of a Trombe wall, air is taken from the room in
the bottom part of the wall. The ventilated wall is made using transparent material (glass) in order
to collect solar heat. Then, a natural #uid motion is created, and the air goes back to the room,

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
624 S. LORENTE

reference ventilated
wall wall

exterior interior exterior interior

Figure 12. Reference wall and ventilated wall operating during summer
conditions (Lorente and BeH gueH , 2000).

solar radiation: 775 W/m²


65
solar radiation: 550 W/m²
60
solar radiation: 370 W/m²

55 solar radiation: 130 W/m²


Temperature (˚ C)

night
50

45

40

35

30
0 4 8 12 16 20 24
Distance (mm)

Figure 13. Temperature pro"les at mid-height in the open channel (Lorente and BeH gueH , 2000).

from the top part of the wall. An important part of the solar energy is imparted to the room, and
this reduces the heating requirements.
The thermal system presented in this section is based on ventilated walls, and is meant for
operating during summer conditions. The study is mostly experimental, based on an in situ
two-room set-up (Lorente and BeH gueH , 2000). The terra-cotta walls have exactly the same surface,
and face south (Figure 12). Consequently, they are exposed to the same climatic conditions, which
are monitored: wind, temperature and solar radiation. One of the walls is used as reference. The
other is equipped with a vertical air slot in front of the wall. This duct is opened at both ends and
is in direct contact with the outside (see Figure 12).
Figure 13 shows the temperature pro"les at mid-height in the open channel in summer. When
solar radiation is almost absent, no #uid motion is detected, and the temperature pro"les

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 625

0,4 z=3cm

vertical component of velocity (m/s)


z=47cm
z=130cm
0,3

0,2

0,1

0
0 5 10 15 20 25 30 35
x (mm)

Figure 14. Numerical results for the vertical component of the velocity in the open-ended duct of the
ventilated wall (Lorente and Massias, 1998).

60 800

700
50
600
Temperature (∞C)

40 500

Heat rate (W/m2)


ventilated wall
400
30 reference wall
300 solar radiation

20 200

100
10
0

0 -100
00 12 00 12 00 12 00 12 00 12
Time (h)

Figure 15. Inner surface temperatures during 4 days in August 1999. Ventilated wall and
reference wall (Lorente and BeH gueH , 2000).

correspond to conductive heat transfer. As the e!ect of solar radiation increases, the ther-
mosiphon phenomenon sets in and generates a double ascending air motion. This motion was
also studied numerically, as shown in Figure 14. Velocity pro"les are plotted for several values of
height (z) along the slot, and x is the distance measured away from the brick side exposed to solar
radiation (Lorente and Massias, 1998). The #ow is ascending along both faces of the duct.
A comparison between the ventilated wall and the reference wall can be seen in Figure 15. This
"gure shows the inner surface temperatures in both cases during 4 days in August 1999. At night,
these temperatures are almost the same. Then, the more solar radiation increases, the more the

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
626 S. LORENTE

600
solar radiation
removed heat
500

400

Heat rate (W/m2) 300

200

100

0
00:00 02:24 04:48 07:12 09:36 12:00 14:24 16:48 19:12 21:36 00:00
-100
Time (h)

Figure 16. Heat removed in the open slots*one day in September 1999 (Lorente and BeH gueH , 2000).

di!erence between the two cases becomes important. Note that the values given by the probes in
the ventilated case are always lower compared to the reference case. Furthermore, if the inner
surface temperatures have a cyclic evolution using both systems, the amplitude of the temper-
atures is lower with the ventilated wall, due to the solar ducts. This is an important result, as fewer
temperature #uctuations on the inner surfaces provide better comfort conditions.
Thermal probes inside the walls allowed us to record the temperature pro"les each hour. These
data were used to calculate the heat transfer removed by natural convection in the solar ducts.
The "rst thermocouples were positioned at only 1 mm from the terra-cotta faces, so that the
convective heat transfer can be calculated by using the de"nition of wall heat #ux on the vertical
faces of the cavities.
Calculations were performed based on the results collected during fall 1999. The heat removed
by natural convection in the opened slots becomes less important, as solar radiation is not as
strong as during summer. Figure 16 shows the evolution of the natural convection heat transfer in
the ventilated wall during 24 h in September 1999. The heat transfer rate follows the time
dependence of solar radiation. The thermal inertia of the outer face induces a time lag of 1 h: on
that particular day solar radiation was maximum at 12.30, while the heat removed reached its
maximum at 13.30. Plotting the evolution of the room temperature versus time, and comparing
this to the solar radiation leads to a phase displacement of 2 h, which accounts for the global
inertia of that system. In this case, the ventilated wall allows us to capture a maximum of 30 per
cent of the total solar radiation. In summer, when solar radiation is at its maximum, 38 per cent of
the heat transfer is removed by natural convection in the solar ducts.

8. DEFORMABLE BOUNDARIES

There are many situations where surfaces of building components undergo deformation due to
weather changes. Consider the case for double glazing windows: if the air pressure inside the

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 627

TH TC TH TC

d
H

x
L 0

Figure 17. Reference cavity with plane walls and cavity with side walls bent inward (Lorente et al., 2000).

1 1 1 1

Z/A Z/A Z/A Z/A

0,5 0,5 0,5 0,5

0 θ 0 θ 0 θ 0 θ
-0,5 0 0,5 -0,5 0 0,5 -0,5 0 0,5 -0,5 0 0,5

(a) (b) (c) (d)

Figure 18. The non-dimensional temperature  versus non-dimensional height Z/A for the reference
Ra"14 200 and several curvatures of deformed cavities: (a) d/¸"100 per cent, (b) d/¸"66 per cent,
(c) d/¸"50 per cent and (d) d/¸"33 per cent (Lartigue, 1999).

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
628 S. LORENTE

cavity decreases (for example, this happens during very cold days when the temperature is below
the initial "lling temperature), the pressure gradient existing between inside and outside the cavity
makes the glass panes bend inward (Figure 17) (Patenaude, 1991). Bernier and Bourret (1997)
presented some interesting results on this phenomenon, based on the assumption that there is no
#uid motion inside the cavity. In this section some recent results (Lorente et al., 2000) on the e!ect
of convection are reviewed. The results are numerical. The de#ected faces of the cavity have
parabolic shape.
Figure 18 illustrates the evolution of the non-dimensional temperature  versus the non-
dimensional height Z/A, at mid-width (X"0.5), for a reference plane cavity of Ra"14 200, and
for several deformed cases with di!erent curvatures. Oscillations indicate the presence of second-
ary cells. As the curvature increases (Figure 18(b)), the cells increase in number but their size
diminishes. The amplitude of the wavy temperature pro"le decreases. When the curvature is
signi"cant (d/¸*0.5), the vertical temperature gradient in the centre of the cavity is almost zero,
which is characteristic of conduction heat transfer. Secondary cells occur only at the ends of the
enclosure, where the dominant mode of heat transfer is convection.
The numerical streamlines of the #ow are presented in Figure 19. In the cavity with plane walls
(Figure 19(a)), the formation of nine secondary cells is very clear. A de#ection of d/¸"0.66
(Figure 19(b)) increases the number of cells to 18. The cell size decreases. The cells are weaker, and
they carry less energy by convection. When the curvature d/¸ is '50 per cent (Figures 19(c) and
19(d)), the #ow presents two distinct patterns in the cavity. In the core the streamlines are vertical,
as the horizontal component of the velocity vector is close to zero. There are no cells in the core
section. The multicellular #ow is only found in top and bottom regions of the cavity. This
behaviour is accentuated when d/¸"0.33. The number of cells diminishes from 14 in Figure 19(c)
to 11 in Figure 19(d).
In the reporting of overall heat transfer rate the Nusselt number is not used, because the
conductive #ux varies with altitude (because of the air thickness) hand in hand with the
convective heat transfer. The alternative is to use a local and overall transfer coe$cient de"ned as


(z)"K(z)¹ (12)


1 &
K " K(z) dz (13)
H

The same de"nition holds for the perfectly rectangular case. The in#uence of the side-wall
curvature on heat transfer is described by the non-dimensional parameter:

K (deformed cavity)
" (14)
K (reference cavity)

In the numerical simulations compared in Table II, the initial (not deformed) aspect ratio is
A"40. A low Rayleigh number, Ra +10, was initially used and then increased until the value
*
of 30 000. A conclusion can be made on the e!ects of the bent faces on heat transfer when thermal
boundary conditions remain constant. The curvature tends to increase heat transfer. It is
noteworthy the  coe$cient becomes less important as the Rayleigh number increases. The
reason is that at low Ra the heat transfer is enhanced when the vertical faces bend inward; while at

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 629

Figure 19. Streamlines for the reference cavity (a) and three cavities with deformed side walls: (a) d/¸"100
per cent, (b) d/¸"66 per cent, (c) d/¸"50 per cent and (d) d/¸"33 per cent. The Rayleigh number is
Ra"14 200 (Lartigue, 1999).

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
630 S. LORENTE

Table II. Values of  showing the e!ect of wall deformation on overall


heat transfer (Lorente et al., 2000).

d (%)* Ra "10 188R Ra R"16 644 Ra "23 394R Ra "30 000R


* * * *
66 1.173 1.156 1.173 1.125
50 1.366 1.308 1.256 1.200
33 1.719 1.591 1.526 1.415

* d (%) ratio between the narrowest part (at mid-height) of the cavity
and the initial thickness.
R Ra Rayleigh number based on reference cases (perfectly rectangular
*
cavities).

higher Rayleigh numbers near the transition from laminar #ow to turbulent #ow, the curvature
tends to stabilize the #ow, and consequently the heat transfer does not increase as much as before.
In sum, the increase of heat transfer through such cavities leads to lower a thermal performance of
such products, i.e. to a higher rate of heat loss.

9. CONCLUDING REMARKS

It is demonstrated in this paper that heat losses in buildings are largely due to convective heat
transfer inside the vertical enclosures of components. Air is often considered as a no-motion #uid
in conductive regime because of the small dimensions of cavities. The reality is quite di!erent.
Convective transfer has to be taken into account in the evaluation of heat losses and in the design
of thermal systems.
It is also shown that since the walls are porous media the level of moisture surrounding
the walls has an important e!ect on heat transfer, enhancing it in a signi"cant way (Vasile
et al., 1998). It is a phenomenon of importance, which could be taken into account in a revision
of the European legislation (European Norm, 1996). Indeed, until now the calculation methods
of building components thermal resistance proposed by the European norm are based only
on heat transfer under the hypothesis of totally dry components. This study is a "rst step
toward a more general project aiming to combine both heat and mass transfer through building
envelopes in order to predict and improve their performance.

NOMENCLATURE

A "aspect ratio, A"H/¸


g "acceleration due to gravity (m s\)
Gr "Grashof number, Gr"Ra/Pr
h "convective heat transfer coe$cient (W (mK)\)
H "height (m)
k "thermal conductivity (W (mK)\)
K "heat transfer coe$cient (W (mK)\)
¸ "thickness (m)

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
HEAT LOSSES THROUGH BUILDING WALLS 631

Nu "Nusselt number
p "pressure (Pa)
P "dimensionless pressure
Pr "Prandtl number, Pr"/
Ra "Rayleigh number, Ra"g¹¸/
t "time (s)
t* "dimensionless time
¹ "temperature (K)
u, w "x-component and z-component of velocity (m s\)
;, = "dimensionless x-component and z-component of velocity
< "reference velocity, < "(g¹¸) (m s\)
 
x, z "Cartesian co-ordinates
X, Z "dimensionless Cartesian co-ordinates
 "thermal volumetric expansion coe$cient (K\)
 "non-dimensionalized heat transfer coe$cient
¹ "temperature di!erence (K)
(z) "local heat #ow (W m\)
"thermal di!usivity (m s\)
 "kinematic viscosity (m s\)
 "density (kg m\)
 "dimensionless temperature

Subscripts
c "cold
h "hot
m "mean
0 "reference

REFERENCES

Batchelor GK. 1954. Heat transfer by free convection across a closed cavity between vertical boundaries at di!erent
temperatures. Quartely of Applied Mathematics 12:209}233.
Bejan A. 1995. Convection Heat ¹ransfer (2nd edn). Wiley: New York.
Ben Yedder R, Bilgen E. 1991. Natural convection and conduction in Trombe wall systems. International Journal of Heat
and Mass ¹ransfer 34:1237}1248.
Bernier MA, Bourret B. 1997. E!ects of glass plate curvature on the U-factor of sealed insulated glazing units. ASHRAE
¹ransactions 103(1):270}277.
Buzzoni L, Dall'Olio R, Spiga M. 1998. Energy analysis of a passive solar system. Revue Ge& ne& rale de ¹hermique
37:411}416.
Detunq B, Bilgen E. 1984. Etude expeH rimentale d'un capteur solaire du type mur Trombe et validation des formules
theH oriques. ¹ransactions of the CSME 18:35}39.
de Vahl Davis G. 1968. Laminar natural convection in an enclosed rectangular cavity. International Journal of Heat and
Mass ¹ransfer 11:1675}1693.
Du$n RJ, Knowles G. 1985. A simple design method for the Trombe wall. Solar Energy 34:69}72.
Eckert ERG, Carlson WO. 1961. Natural convection in an air layer enclosed between two vertical plates with di!erent
temperatures. Intertnational Journal of Heat and Mass ¹ransfer 2:106}120.
Elder JW. 1965. Laminar free convection in a vertical slot. Journal of Fluid Mechanics 23:77}98.
ElSherbiny SM, Raithby GD, Hollands KGT. 1982. Heat transfer by natural convection across vertical and inclined air
layers. Journal of Heat ¹ransfer 104:96}102.

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632
632 S. LORENTE

European Norm. 1996. Building components and building elements. Thermal resistance and thermal transmittance.
Calculation method. NF EN iso 6946.
Kakac S, Yener Y. 1994. Convective Heat ¹ransfer. CRC Press: Boca Raton.
Korpela SA, Lee Y, Drummond JE. 1982. Heat transfer through a double pane window. Journal of Heat ¹ransfer
104:539}544.
Lartigue B. 1999. Contribution à l'eH tude thermique et dynamique de doubles vitrages courbeH s. Approche numeH rique et
expeH rimentale. Ph.D. ¹hesis, INSA, Toulouse.
Lartigue B, Lorente S, Bourret B. 2000. Multicellular natural convection in a high aspect ratio cavity: experimental and
numerical results. International Journal of Heat and Mass ¹ransfer 43(17):3157}3170.
Lauriat G, Desrayaud G. 1985. Natural convection in air-"lled cavities of high aspect ratios: discrepancies between
experimental and theoretical results. ASME Paper no 85-HT-37.
Lee Y, Korpela SA. 1983. Multicellular natural convection in a vertical slot. Journal of Fluid Mechanics 126:91}121.
Le QueH reH P. 1990. A note on multiple and unsteady solutions in two-dimensional convection in a tall cavity. Journal of
Heat ¹ransfer 112:965}974.
Lorente S, Petit M, Javelas R. 1996. Simpli"ed analytical model for thermal transfer in vertical hollow brick. Energy and
Buildings 24:95}103.
Lorente S, Petit M, Javelas R. 1998. The e!ects of temperature conditions on the thermal resistance of walls made with
di!erent shapes vertical hollow bricks. Energy and Buildings 28:237}240.
Lorente S, Massias E. 1998. Protection against solar overheatings using high aspect ratio open vertical cavities. =orld
Renewable Energy Congress, Florence, September 20}25.
Lorente S, BeH gueH M. 2000. Experimental study of a passive system: a ventilated wall. Euro-Conference on New and
Renewable ¹echnologies for Sustainable Development, Madeira, June 26}29.
Lorente S, Lacarrière B, Lartigue B. 2000. In#uence of the geometry on heat transfer by natural convection in
a rectangular cavity. 12th International Symposium on ¹ransport Phenomena, Istanbul, July 16}21.
Padet J. 1991. Fluides en Ecoulement, Me& thodes et Mode% les, Masson: Paris.
Patenaude A. 1991. Sealed glazing units distortion. Construction Canada, 40}43.
Raithby GD, Wong HH. 1981. Heat transfer by natural convection across vertical air layers. Numerical Heat ¹ransfer
4:447}457.
Ramanan N, Korpela SA. 1989. Multigrid solution of natural convection in a vertical slot. Numerical Heat ¹ransfer
15:323}339.
Roux B, Grondin J, Bontoux P, de Vahl Davis G. 1980. Reverse transition from multicellular to monocellular motion in
vertical #uid layer. Physics and Chemistry of Hydrodynamics 3F:292}297.
Siegel R, Howell JR. 1981. ¹hermal Radiation Heat ¹ransfer. MacGraw Hill: New York.
Vasile C, Lorente S, Perrin B. 1998. Study of convective phenomena inside cavities coupled with heat and mass transfers
through porous media; application to vertical hollow bricks; a "rst approach. Energy and Buildings 28:229}235.
Wakitani S. 1997. Development of multicellular solutions in natural convection in an air-"lled vertical cavity. Journal of
Heat ¹ransfer 119:97}101.
Wright JL, Sullivan HF. 1989. Natural convection in sealed glazing units: a review. ASHRAE ¹ransactions 95(1):
592}602.
Wright JL, Sullivan HF. 1994. A two-dimensional numerical model for natural convection in a vertical rectangular
window cavity. ASHRAE ¹ransactions 100(2):1193}1206.
Yin SH, Wung TY, Chen K. 1978. Natural convection in an air layer enclosed within rectangulaur cavities. International
Journal of Heat and Mass ¹ransfer 21:307}315.
Zhao Y, Curcija D, Gross WP. 1997. Prediction of the multicellular #ow regime of natural convection in fenestration
glazing cavities. ASHRAE ¹ransactions 103(1):1}12.

Copyright  2002 John Wiley & Sons, Ltd. Int. J. Energy Res. 2002; 26:611}632

You might also like