Numerical simulation of air ow and heat transfer in

domestic refrigerators
O. Laguerre
a,
*
, S. Ben Amara
a
, J. Moureh
a
, D. Flick
b
a
UMR Ge nie Industriel Alimentaire INRA-INAPG-Cemagref-ENSIA, BP. 44, 92163 Antony Cedex, France
b
UMR Ge nie Industriel Alimentaire INRA-INAPG-Cemagref-ENSIA, 16 rue Claude Bernard, 75231 Paris, France
Received 24 February 2006; received in revised form 17 October 2006; accepted 21 October 2006
Available online 15 December 2006
Abstract
This work was carried out in order to study heat transfer by natural convection in domestic refrigerators without ventilation. Only the
refrigerating compartment was studied for three congurations: empty refrigerator, refrigerator equipped with glass shelves and refrig-
erator loaded by product. Both experimental and numerical approaches were used.
The simulations were carried out using CFD (computational uid dynamic) software by taking into account or by neglecting radi-
ation heat transfer. The following conditions were assumed: constant evaporator temperature, three-directional laminar air ow. Numer-
ical results show temperature stratication in the refrigerating compartment (warm zone at the top and cold zone on the bottom) for all
congurations. A comparison of the calculated air temperature and the experimental values shows good agreement when radiation is
taken into account.
2006 Elsevier Ltd. All rights reserved.
Keywords: CFD; Simulation; Closed cavity; Refrigeration; Domestic refrigerator
1. Introduction
Domestic refrigerators are widely used in industrialized
countries. There are approximately 1 billion domestic
refrigerators worldwide (IIR, 2002). In France, there are
1.7 refrigerators per household (AFF, 2001). In developing
countries, the production is rising steadily: total production
rose 30% in 2000 (Billiard, 2005). Some indications show
that food is often stored in domestic refrigerators at tem-
peratures that are too high. In refrigerators without venti-
lation, strong temperature heterogeneity is often observed,
with warm zones (sanitary risk) and cold zones (freezing
risk) due to very low air circulation. For this type of refrig-
erator, widely used in Europe and in developing countries,
heat transfer occurs principally by natural convection.
Knowledge of air temperature and velocity proles in a
refrigerator is important for food quality control. Indeed,
if the consumer knows the position of warm and cold zones
in the refrigerator, products can be placed correctly.
This work was carried out in order to gain a better
insight into air ow and heat transfer inside a refrigerator.
Three congurations were studied: an empty refrigerator
with and without shelves, and a loaded refrigerator. The
objective was to quantify the air temperature and velocity
distribution in the refrigerating compartment in the pres-
ence of obstacles (shelves and product) and to compare
the results with those obtained using an empty compart-
ment. The inuence of heat exchange through natural con-
vection (between the air and the walls) and by radiation
(between the internal walls) was studied. Both experimental
and numerical (CFD software) approaches were used.
The practical objective of this study is to predict the
warm and cold zones in a domestic refrigerator. This objec-
tive can be reached by the characterisation of air ow and
heat transfer in the appliance.
0260-8774/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfoodeng.2006.10.029
*
Corresponding author. Tel.: +33 140 966 121; fax: +33 140 966 075.
E-mail address: onrawee.laguerre@cemagref.fr (O. Laguerre).
www.elsevier.com/locate/jfoodeng
Journal of Food Engineering 81 (2007) 144156
2. Literature review
To demonstrate the air ow and heat transfer in a refrig-
erator, literature on free convection phenomena in a closed
cavity will be presented, then studies applied to domestic
refrigerators will be mentioned.
2.1. Air ow and heat transfer in an empty cavity
Air ow by natural convection in an empty cavity is
related to the dierence in wall temperatures. Only conven-
tional convection (one vertical cold wall and one vertical
warm wall) is presented in this article. This conguration
is often observed in domestic refrigerators where an evap-
orator is embedded in the vertical back wall and the door
located opposite this wall is warm. The air density varia-
tion due to the temperature gradient (perpendicular to
the gravitational direction) contributes to air circulation,
hot air being lighter than cold air.
The ow regime in natural convection is characterised
by the Rayleigh number (Ra) dened as
Ra =
gbDTL
3
am
(1)
In general, the critical Rayleigh number, which distin-
guishes the transition from laminar to turbulent ows, is
approximately 10
9
(depending on the geometry and bound-
ary conditions, Incropera & Dewitt, 1996).
Several experimental studies have been carried out to
measure air temperature and/or velocity in closed cavities
(Ampofo & Karayiannis, 2003; Armaly, Li, & Nie, 2003;
Betts & Bokhari, 2000; Mergui & Penot, 1996; Tian &
Karayiannis, 2000).
Tian and Karayiannis (2000) used a Doppler laser
anemometer to measure the air velocity in a rectangular
cavity (height width depth = 75 75 150 cm, Ra =
1.58 10
9
) (Fig. 1). They observed two types of air cir-
culation. The rst one is the principal air recirculation
loop near to walls where the air temperature and velocity
vary rapidly. The second one consists of small recircula-
tion loops located between the boundary layers (near
walls) and the centre of the cavity.
Eckert and Carlson (1961) carried out an experimental
study and they observed that outside the boundary layers,
the temperature is homogeneous at a given height and this
temperature increases in the vertical direction. They also
proposed a correlation between Nusselt (Nu) and Rayleigh
(Ra) numbers. No velocity measurements were undertaken
in this study.
Ostrach (1988), Catton (1978) and Yang (1987) carried
out a literature review on this subject, which presents the
experimental and modelling results (2-D and 3-D). These
authors emphasise the importance of the aspect ratio of
the cavity and the temperature dierence between walls
on the ow regime.
Heat exchange by radiation between the internal walls
of the cavity is as important as that achieved by natural
convection and this should be taken into account. Sev-
Nomenclature
I intensity of radiation, W m
2
per unit solid an-
gle
L characteristic length, m
~n normal vector
g acceleration due to gravity (9.81 m s
2
)
Nu Nusselt number
Q heating power, W
R radius, m
Ra Rayleigh number
N
RC
radiationconvection interaction parameter
T air temperature, C or K
T
amb
ambient temperature, C or K
T
evap
evaporator temperature, C or K
T
h
warm wall temperature, C or K
T
c
cold-wall temperature, C or K
T
s
surface temperature, C or K
DT temperature dierence between cold and warm
walls, C or K
Greek symbols
a air thermal diusivity, m
2
s
1
b thermal expansion coecient, K
1
e emissivity of the wall
q air density, kg m
3
k air thermal conductivity, W m
1
K
1
t air kinematic viscosity, m
2
s
1
X solid angle
U radiative ux, W m
2
Fig. 1. Air ow in a closed cavity (Tian & Karayiannis, 2000).
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 145
eral authors (Balaji & Venkateshan, 1994; Ramesh &
Venkateshan, 1999; Velusamy, Sundarajan, & Seethar-
amu, 2001; Li & Li, 2002) showed by experimental and
numerical approaches that these two heat transfer modes
occur simultaneously. Ramesh and Venkateshan (1999)
showed experimentally that for a square enclosure (verti-
cal walls maintained at 35 and 65 C, adiabatic horizon-
tal walls, Ra = 5 10
5
), the heat transfer by convection
and radiation between high emissive vertical walls
(e = 0.85) is twice that between polished ones (e = 0.05).
The result of this study is relatively dierent from the
case of a domestic refrigerator since the wall temperature
dierence between the evaporator and the other walls is
on average 15 K. The eect of radiation is, therefore, less
signicant. Balaji and Venkateshan (1994) proposed cor-
relations (established from numerical simulations) to
express the convection and radiation in a square cavity
in function of e, Ra, T
c
/T
h
and a radiationconvection
interaction parameter N
RC
=
rT
4
h
L
k(T
h
Tc)

.
These correlations show that the radiation eect
increases when the wall emissivity and/or wall tempera-
tures increase. Moreover, Li and Li (2002) reported that
the radiation increases in comparison with convection as
the size of the enclosure increases. Colomer, Costa, Con-
sul, and Oliva (2004) reported that in a transparent med-
ium, radiation signicantly increases the heat ux. These
authors also reported that for a given Planck number,
and constant reference temperature ratio, the contribu-
tion of radiation remains almost constant for a range
of Rayleigh number. An estimation of convection and
radiation heat transfer in a refrigerator was carried out
in our previous study (Laguerre & Flick, 2004). The
equivalent radiative heat transfer coecient between
two parallel plates was evaluated to represent the radia-
tive exchange between the evaporator wall and the door.
It was found that, the radiative heat transfer coecient is
the same order of magnitude as the convective heat
transfer coecient. This conrms the importance of
radiation.
2.2. Air ow and heat transfer in a domestic refrigerator
In an empty refrigerator, cold air near the evaporator
ows downward and warm air near the door and the other
side walls ows upwards (Fig. 2). The heat exchanges inside
the cavity are related to natural convection between inter-
nal walls and air, radiation between evaporator and the
other walls and conduction within the walls (Laguerre &
Flick, 2004). In the case of a refrigerator lled with prod-
ucts, the products are cooled by natural convection, by
radiation between the surface of the products and the inter-
nal walls of the refrigerator, and through conduction and
radiation between products.
Several studies have been carried out on the cold pro-
duction system of domestic refrigerators (Alsaad & Ham-
mad, 1998; Bansal, Wich, & Browne, 2001; Chen, Wu, &
Sun, 1996; Graviss & Zurada, 1998; Grazzini & Rinaldi,
2001; Radermacher & Kim, 1996). The main objective
of these studies is to optimize energy consumption. How-
ever, fewer studies have been carried out on phenomena
inside the refrigerating compartment. Among these stud-
ies, those conducted by Masjuki et al. (2001) and James
and Evans (1992) were experimental studies on empty
and loaded refrigerators. The objective of these studies
was to analyze the eects of several parameters on the
temperature in the refrigerating compartment (thermostat
setting, frequency of door openings, lled volume, tem-
perature and humidity of ambient air). It is dicult to
understand the mechanism of heat transfer by natural
convection from the results obtained by these studies,
due to the complexity of the refrigerator operation (com-
pressor on and o cycles, dierent degrees of insula-
tion in walls, heat loss through gaps, etc.). Measurement
of air ow in a freezer compartment under real operating
conditions was carried out by Lacerda, Melo, Barbosa,
and Duarte (2005) using PIV (particle image velocimetry).
It was observed that the ow eld was strongly inuence
by the temperature variations due to the on and o
operation cycles of compressor. This behavior was attrib-
uted to natural convection and the physical properties
(viscosity) of air, which strongly depend on the tempera-
ture. Another study on air ow in a ventilated domestic
freezing compartment was carried out by Lee, Baek,
Chung, and Rhee (1999). In this study, comparison of
the velocity eld obtained by CFD simulation and by
experiment (PIV measurement) was undertaken. These
authors observed that the ow was very complex: jet-like
ow around entrance ports, impinging and stagnation
ow on the walls and large recirculation ow in the
cavity.
Several numerical studies have been carried out on heat
transfer in empty domestic refrigerators (Deschamps,
Prata, Lopes, & Schmid, 1999; Pereira & Nieckele, 1997;
Silva & Melo, 1998). However, few studies have been car-
ried out on loaded refrigerators. The numerical studies
Fig. 2. Various heat exchange modes and air ow inside a domestic
refrigerator (source: Laguerre and Flick, 2004).
146 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156
mentioned previously provide knowledge on the tempera-
ture and velocity heterogeneity under determined condi-
tions. However, radiation was not taken into
consideration in spite of the fact that this heat transfer
mode is of the same order of magnitude as that of convec-
tion. In our study, both empty and loaded refrigerators
were studied and both natural convection and radiation
were taken into account in the simulation.
3. Materials and methods
3.1. Refrigerator
A static cold refrigerator (without ventilation) was used
in this work. It was a single-door appliance with only a
refrigerating compartment (without a freezer). The general
characteristics are shown in Table 1.
Three cases were studied (Fig. 3): an empty refrigerator
without shelves, empty refrigerator tted with glass shelves
(5 mm thickness, thermal conductivity of glass
0.75 W m
1
K
1
) and a refrigerator equipped with glass
shelves and loaded with a test product. This product is
made of methylcellulose (thermal conductivity
0.5 W m
1
K
1
) and the dimensions of one package are
10 10 5 cm (length width depth). The arrange-
Fig. 3. Domestic refrigerator geometry: (a) empty refrigerator; (b) refrigerator tted with glass shelves; (c) refrigerator with glass shelves and products.
Table 1
Characteristics of the refrigerator
External dimensions (height width depth) 149 cm 60 cm 59 cm
Internal dimensions (height width depth) 136 cm 52 cm 44 cm
Dimensions of the evaporator 90 cm 48 cm
Thermostat setting +5 C
Number of shelves 4
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 147
ment of the packages is shown in Fig. 3c. All experiments
were carried out in a temperature-controlled room
(20 0.2 C). As shown in Fig. 3, the evaporator is located
in the upper part of the cabinet. The indentation observed
in the lower right area of the gures represents the com-
pressor placement. To avoid a too complex geometry, the
containers for butter, eggs and bottles usually attached to
the door were removed during our experiments. This facil-
itates the meshing of the refrigerator and the result
interpretation.
3.2. Measurement of the thermal resistance of refrigerator
insulation
Measurement of the thermal resistance of refrigerator
insulation was carried out in a temperature-controlled
room (6 C). A heating coil was placed inside the switch
o refrigerator. The heat supplied to the coil is equal to
the heat loss to external air through the walls. The heating
power was adjusted in such a manner as to maintain the
average internal air temperature at 30 C. In this manner,
the average temperature of the insulating walls is almost
the same as under real operating conditions. To ensure a
homogeneous air temperature inside the refrigerator, a
small fan was installed near the heating coil. The internal
air temperature (T
int
controlled at 30 C), external air tem-
perature (T
ext
controlled at 6 C), power supplied to the
heating coil (Q
1
) and fan (Q
2
) were recorded when the
steady state was attained (after 12 h) and the average val-
ues were calculated over 3 h. Thus, the thermal resistance
of the refrigerator insulation can be calculated knowing
Q
1
+ Q
2
and T
int
T
ext
.
The measurement was used afterwards for the bound-
ary conditions in the CFD simulation. In fact, this exper-
imental thermal resistance takes into account the thermal
resistance between external air and internal walls. There-
fore, a correction was undertaken on the measured value
by subtracting the thermal resistance between internal air
Fig. 4. Air (average value on the symmetry plan), side wall (average value of three measurements: top, middle and bottom levels) and evaporator
temperature changes in the empty refrigerator without shelves (thermostat setting at 5 C).
Table 2
Resolution parameters used in simulation
Relaxation factor Type of discretization
Pressure 0.8 Presto
Density 1
Gravity forces 1
Momentum 0.2 Second order upwind
Energy 1 Second order upwind
Radiation 1
Pressurevelocity Simple
Table 3
Number of cells used for the simulations
Mesh number Height
(136 cm)
Half width
(26 cm)
Depth
(44 cm)
Total
Empty refrigerator 138 28 66 255024
Refrigerator with
shelves
222 28 66 410256
Refrigerator with
shelves and products
240 62 74 1101120
148 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156
and walls. The internal convective heat transfer coe-
cient was assumed to be about 10 W m
2
K
1
. This cor-
rection is weak because the thermal resistance between
air and internal wall represents only around 7% of the
overall thermal resistance (between external and internal
air).
3.3. Temperature measurement
Air and product temperatures were measured experi-
mentally using calibrated thermocouples (Type T) placed
in dierent positions of the symmetry plane of the refriger-
ator and on the plane situated at 8 cm from side wall
(Fig. 3). On each plane, the air temperature was measured
at ve height levels (31.0, 61.0, 94.0, 114.5, 134.5 cm) and
for each height, ve air temperature measurements were
recorded (1, 2, 21.5, 42, 43 cm from the evaporator).
Firstly, the refrigerator operated over 24 h to ensure stabil-
ization conditions, then the temperatures were recorded
every 2 min for 24 h and the average value was calculated
at each measurement point. An example of temperature
evolution inside the refrigerator is shown in Fig. 4. It can
be seen that the evaporator temperature varies from
16 C to +7 C, due to the thermal inertia, the air temper-
ature varies less, from +3.5 C to +7 C, and the wall tem-
perature varies from 4 C to 9 C.
4. Modelling
4.1. Main assumptions and boundary conditions
In the present study, the Rayleigh number (Ra) is about
6 10
8
(estimation based on the height of the evaporator
and the temperature dierence between the internal air
and the cold-wall surface). Laminar ow assumption was
made for the ow regime in our simulation since
Ra < 10
9
. Furthermore, several numerical studies showed
that turbulence does not change the predicted air tempera-
ture pattern (Deschamps et al., 1999; Kingston, Woolley, &
Tridimas, 1994). Boussinesq approximation was used since
the air temperature variation is small compared with the
mean absolute value.
The thermal boundary conditions are based on experi-
mental data:
v Uniform global heat transfer coecient between exter-
nal air and internal wall (0.34 W m
2
K
1
).
v Constant external air temperature (20 C).
v Constant evaporator temperature (0.5 C) which is the
average value during on and o running cycles of
compressor. This constant temperature is used in order
to avoid excessive complexity in the calculation and to
reduce calculation time.
Fig. 5. Mesh structure: (a) empty refrigerator; (b) refrigerator tted with glass shelves; (c) refrigerator loaded with the test product.
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 149
The simulations were performed with the nite volume
method using CFD software Fluent 6.1 with the resolution
parameters indicated in Table 2.
Transient simulation was performed but only the results
obtained after simulation convergence were used in the
comparison with the experimental values.
Fig. 6. Predicted temperature elds (C): (a) on the symmetry plan of empty refrigerator; (b) on the symmetry plan of refrigerator with glass shelves; (c) on
the symmetry plan of refrigerator loaded with products; (d) on the plan situated at 8 cm from the side wall of refrigerator loaded with products.
150 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156
4.2. Mesh
Structured mesh was used to describe the geometry of
the refrigerator. Finer meshes were used near walls, shelves
and products. The number of cells used in each case is
shown in Table 3 and mesh structures are shown in
Fig. 5. To ensure that the results were not inuenced by
the cell numbers, a sensitivity study was carried out before-
hand. Only one half of the refrigerator was meshed because
of the symmetry plane.
4.3. Discrete ordinate method (DO) for radiation
The discrete ordinate method (Chui & Raithby, 1993)
was successfully used to simulate the coupling of convec-
tion and radiation in closed cavity (Colomer et al., 2004;
Sanchez & Smith, 1992).
This model can take into account the participating med-
ium. However, in our case, air is considered as transparent
(with neither absorption nor diusion). The general equa-
tion of heat transfer by radiation (in a given~s direction) is
~
\ (I(~r;~s)~s) = 0 (2)
I(~r;~s) is radiative intensity in ~s direction (at ~r position)
(W m
2
per unit solid angle).
For a gray surface of emissivity e
r
, the net radiative ux
leaving the surface is
U
rad out
= (1 e
r
)
Z
~s~n>0
I
in
~s ~ndX
|{z}
incident flux
e
r
rT
4
s
(3)
The walls are assumed as gray diuse: I
out
= /
rad_out
/p. I
in
is intensity of incident radiation in ~s direction (at ~r posi-
tion); ~n is normal vector; T
s
is surface temperature, K; X
is solid angle.
A sensitivity study of solid angle discretization was car-
ried out beforehand in order to ensure that the simulation
results were not inuenced by the number of solid angle
subdivisions.
5. Results and discussion
5.1. Numerical simulation (taking into account radiation)
The results presented in this paragraph concern simula-
tion, which takes into account heat transfer by convection
between walls and air and by radiation between the internal
walls of the refrigerating compartment.
5.1.1. Temperature elds
The temperature elds obtained from simulations for the
dierent cases studied are shown in Fig. 6. Considering only
the main cavity (excluding the vegetable box), for all cases,
thermal stratication is observed with the cold zone at the
bottom (~2 C) of the refrigerating compartment and the
warmzone at the top(89 C). Inaddition, a coldzone is also
observed along the back wall. This is related to cold air com-
ing from the evaporator. When the refrigerator is loaded
with products, the temperature of the product located near
the evaporator is lower than that located near the door. In
the top half of the compartment, the temperature is relatively
homogeneous at a given height (except in the boundary lay-
ers near the walls). The temperature of the vegetable box is
almost constant for all cases studied (~8 C).
The temperature eld is slightly inuenced by the pres-
ence of obstacles: shelves and products. A slightly lower
temperature is observed at the bottom and a slightly higher
one at the top compared with the empty refrigerator case.
This is due to the fact that the shelves and/or the products
slowed down the air circulation in the central zone of the
refrigerator. The presence of shelves and/or products also
inuenced the main air circulation in the boundary layers
situated along the evaporator and the side walls. However,
this inuence is weak because of the presence of air spaces
between the shelves and the vertical walls (1.2 cm between
the back wall and the shelves), which facilitates the air ow.
In our previous study, it was found that the thickness of the
boundary layer was less than 2 cm (Laguerre, Ben Amara,
& Flick, 2005).
In addition to the overall thermal stratication in the
cavity, stratication is also observed in each gap between
two shelves or between a shelf and a product. It is to be
emphasised that for the refrigerator loaded with the test
product, the symmetry plane is located in the gap between
two piles. This explains why the packages are invisible on
this plane (Fig. 6c). On the plane situated at 8 cm from a
side wall which cuts the product pile (Fig. 6d), a cold prod-
uct zone near the evaporator can be clearly distinguished.
This is related to the blockage of cold air by the product.
The average and maximum air temperatures in all cases
are reported in Table 4. The air temperatures increase with
increasing numbers of obstacles.
Table 4
Average and maximum air temperatures for the three simulations
Average temperature in the main
cavity (C)
Maximum temperature in the main
cavity (C)
Average temperature in the
vegetable box (C)
Empty refrigerator 3.8 8.2 7.4
Refrigerator with glass shelves 4.0 9.0 8.2
Refrigerator with glass shelves and
products
5.1 9.1 8.0
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 151
5.1.2. Air velocity eld
Fig. 7 presents the air velocity elds on the symmetry
plane (Fig. 7ac) and on the plane situated at 8 cm from
the side wall (Fig. 7d) for the dierent cases studied. Con-
sidering only the main cavity (excluding the vegetable box),
for all cases, the main air circulation is observed near the
Fig. 7. Path lines: (a) on the symmetry plan of the empty refrigerator; (b) on the symmetry plan of the refrigerator tted with glass shelves; (c) on the
symmetry plan of the refrigerator loaded with products; (d) on the plan situated at 8 cm from the side wall of refrigerator loaded with products.
152 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156
walls, and constitutes a recirculation loop. Air ows down-
wards along the evaporator while its velocity increases
along the course to attain a maximum value at the bottom
of the refrigerator (u
max
~ 0.2 m s
1
). Air then ows
upwards along the door and the side walls of the refriger-
ator while its velocity decreases progressively and becomes
Fig. 8. Temperature eld (radiation not taken into account): (a) on the symmetry plan of the empty refrigerator; (b) on the symmetry plan of the
refrigerator tted with glass shelves; (c) on the symmetry plan of the refrigerator loaded with the test product; (d) on the plan situated at 8 cm from the
side wall of the refrigerator loaded with products.
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 153
stagnant at the top of the refrigerator. This observation is
in agreement with the air temperature eld shown in Fig. 6,
with cold air located at the bottom of the cavity and warm
air at the top. It can also be observed that there is a weak
horizontal air ow from the door to the evaporator. How-
ever, the air velocity at the centre of the cavity is very low
(<0.04 m s
1
). In the case of the refrigerator tted with
glass shelves, in addition to the main air ow along the
walls as mentioned previously, there are also small air
loops between the shelves. For the refrigerator loaded with
products, air ows in the gaps between the shelves and the
products (Fig. 7d).
It should be remembered that the containers attached to
the door were not represented in our study. In practice
Fig. 9. Comparison between experimental air temperatures and predicted values obtained by simulation with and without radiation: (a) empty
refrigerator; (b) refrigerator tted with glass shelves; (c) refrigerator loaded with products.
154 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156
these containers are an obstacle to airow along the door
and reduce the air velocity in this area.
Considering the vegetable box, one or two air recircula-
tion loops were observed (Fig. 7). This is due to the pres-
ence of the glass shelf (cold wall), which separates the
vegetable box from the main cavity, and the ve other walls
which are warmer (heat loss through these walls).
5.2. Comparison with numerical simulation without radiation
Fig. 8 presents the air temperature eld on the symmetry
plane obtained by simulation without taking into consider-
ation radiation (between internal walls of the refrigerating
compartment, shelves and product surface). It was
observed that overall the temperature eld is similar to that
present when radiation is taken into account (a cold zone
at the bottom and a warm zone at the top). However, strat-
ication is more pronounced without radiation, and this
leads to a higher temperature at the top of the cavity. In
fact, for an empty refrigerator, the maximum temperature
rises from 8 C (with radiation) to 15 C (without radia-
tion). This temperature increase can be explained by the
fact that, without radiation, there is no heat exchange
between the warm top wall and the other colder walls, par-
ticularly the evaporator wall. This contributes to a high air
temperature at the top position. When radiation is taken
into account, the heat exchange between the top wall and
the other walls tends to reduce the top wall temperature
and consequently reduces air temperature near this wall.
From a microbiological point of view, the growth rate is
much higher at 15 C than at 8 C. It is therefore necessary
to take into consideration radiation in the simulation in
order to better describe the phenomena occurring in
domestic refrigerators.
5.3. Comparison between the predicted air temperature and
experimental values
Fig. 9 presents a comparison between the experimental
and predicted air temperature results (with and without
taking into account radiation). It can be seen that the sim-
ulation results with radiation agreed with the experimental
values to a greater extent, while simulation without radia-
tion over-estimated the air temperature, particularly at
the top of the refrigerator. The peaks observed on the tem-
perature prole in the presence of shelves and/or products
can be explained by the higher conductivity of glass com-
pared with air and by the cold air ow along the upper
sides of the shelves.
The agreement between the experimental and simulation
results is relatively poor in the case of a loaded refrigerator,
even though the radiative heat exchange between the prod-
uct and the walls was taken into account. This may be
explained by the geometry complexity. Further renement
could lead to a better agreement, but the computing time is
already very high (about 8 days using a cluster of four pro-
cessors of 2Go of RAM).
6. Conclusions
Numerical simulation of air ow and heat transfer was
carried out within the refrigerating compartment of a
domestic refrigerator without a fan. Three congurations
were studied: an empty refrigerator, an empty refrigerator
tted with glass shelves and a refrigerator loaded with
products. When radiation was taken into consideration in
simulation, the predicted air temperatures were in good
agreement with the experimental values. However, when
radiation was not taken into account, the temperature
was over-estimated, particularly at the top of the refrigera-
tor. Radiation allows heat exchange, particularly between
the top wall and the cold wall (evaporator); consequently,
it limits the stratication phenomena.
The obstacles (shelves and/or products) slow down the
air circulation in the central zone of the refrigerator and
mildly inuence the main air circulation along the walls.
This is conrmed by the maximum values of air tempera-
ture: 8.2 C for an empty refrigerator without shelves and
9.1 C for an empty refrigerator with shelves and refrigera-
tor loaded with products.
Whatever the conguration studied (empty with/with-
out shelves, loaded with products) for this type of refriger-
ator, the air temperature at the top of the refrigerator is
about 5 C higher than the average air temperature, and
therefore it is important to avoid placing sensitive products
in this position.
The CFD simulation developed by our work can be
further used as a tool to study the inuence of operating
conditions on the temperature and velocity elds: the
evaporator temperature (parameter related to the ther-
mostat setting by the consumer), the dimensions of the
evaporator (parameter related to design) and the percent-
age of product-occupied volume in the refrigerating
compartment.
Acknowledgement
The authors would like to thank to the French Ministry
of Agriculture and the Ile de France Regional Council
for their nancial support.
References
AFF, Association Franc aise du Froid (2001). Conseil National du Froid,
Livre blanc sur les uides frigorige` nes, Paris (51pp).
Alsaad, M. A., & Hammad, M. A. (1998). The application of propane/
butane mixture for domestic refrigerators. Applied Thermal Engineer-
ing, 18, 911918.
Ampofo, F., & Karayiannis, T. G. (2003). Experimental benchmark data
for turbulent natural convection in an air lled square cavity.
International Journal of Heat and Mass Transfer, 46(19), 35513572.
Armaly, B. F., Li, A., & Nie, J. H. (2003). Measurements in three-
dimensional laminar separated ow. International Journal of Heat and
Mass Transfer, 46, 35733582.
Balaji, C., & Venkateshan, S. P. (1994). Correlations for free convection
and surface radiation in a square cavity. International Journal of Heat
and Fluid Flow, 15(3), 249251.
O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156 155
Bansal, P. K., Wich, T., & Browne, M. W. (2001). Optimization of egg-
crate type evaporators in domestic refrigerators. Applied Thermal
Engineering, 21(7), 751770.
Betts, P. L., & Bokhari, I. H. (2000). Experiments on turbulent natural
convection in a closed tall cavity. International Journal of Heat and
Mass Transfer, 21, 675683.
Billiard, F. (2005). Refrigerating equipment, energy eciency and refrig-
erants. Bulletin of the International Institute of Refrigeration, 85(1),
1225.
Catton, I. (1978). Natural convection in enclosures. In Proceedings of 6th
international heat transfer conference, Toronto, Canada (Vol. 6, pp.
1331).
Chen, L., Wu, C., & Sun, F. (1996). Heat transfer eect on the specic
cooling load of refrigerators. Applied Thermal Engineering, 16(12),
989997.
Chui, E. H., & Raithby, G. D. (1993). Computation of radiant heat
transfer on a non-orthogonal mesh using the nite-volume method.
Numerical Heat Transfer, Part B, 23, 269288.
Colomer, G., Costa, M., Consul, R., & Oliva, A. (2004). Three-
dimensional numerical simulation of convection and radiation in a
dierentially heated cavity using the discrete ordinates method.
International Journal of Heat and Mass Transfer, 47, 257269.
Deschamps, C. J., Prata, A. T., Lopes, L. A. D., & Schmid, A. (1999).
Heat and uid ow inside a household refrigerator cabinet. In
Proceedings 20th IIR international congress of refrigeration, Sydney.
Eckert, E. R. G., & Carlson, W. O. (1961). Natural convection in an air
layer enclosed between two vertical plates with dierent temperatures.
International Journal of Heat and Mass Transfer, 2, 106120.
Graviss, K., & Zurada, J. (1998). A neural network controller for optimal
temperature control of household refrigerators. Intelligent Automation
and Soft Computing, 4(4), 357372.
Grazzini, G., & Rinaldi, R. (2001). Thermodynamic optimal design of
heat exchangers for an irreversible refrigerator. International Journal of
Thermal Sciences, 40(2), 173180.
IIR, International Institute of Refrigeration. (2002). Report on refriger-
ation sector achievements and challenges (77pp).
Incropera, F. P., & Dewitt, D. P. (1996). Fundamentals of heat and mass
transfer (4th ed.). John Wiley & Sons.
James, S. J., & Evans, J. (1992). The temperature performance of domestic
refrigerators. International Journal of Refrigeration, 15(5), 313319.
Kingston, P., Woolley, N., & Tridimas, Y. (1994). Fluid ow and heat
transfer calculations in a domestic refrigerator. FIDAP UK user
meeting, Fluent France S A (pp. 111).
Lacerda, V. T., Melo, C., Barbosa, J. R., & Duarte, P. O. O. (2005).
Measurements of the air ow eld in the freezer compartment of a top-
mount no-frost refrigerator: the eect of temperature. International
Journal of Refrigeration, 28, 774783.
Laguerre, O., Ben Amara, S., & Flick, D. (2005). Experimental study
of heat transfer by natural convection in a closed cavity:
application in a domestic refrigerator. Journal of Food Engineering,
70, 523537.
Laguerre, O., & Flick, D. (2004). Heat transfer by natural convection in
domestic refrigerators. Journal of Food Engineering, 62, 7988.
Lee, I. S., Baek, S. J., Chung, M. K., & Rhee, D. (1999). A study of air
ow characteristics in the refrigerator using PIV and computational
simulation. Journal of Flow Visualisation and Image Processing, 6,
333342.
Li, N., & Li, Z. X. (2002). Relative importance of natural convection and
surface radiation in a square enclosure. International Journal of
Nonlinear Science and Numerical Simulation, 3, 613616.
Masjuki, H. H., Saidur, R., Choudhury, I. A., Mahlia, T. M. I., Ghani, A.
K., & Maleque, M. A. (2001). The applicability of ISO household
refrigerator-freezer energy test specications in Malaysia. Energy,
26(7), 723737.
Mergui, S., & Penot, F. (1996). Convection naturelle en cavite carree
dierentiellement chauee: investigation experimentale a`
Ra = 1.69 10
9
. International Journal of Heat and Mass Transfer,
39(3), 563574.
Ostrach, S. (1988). Natural convection in enclosures. Journal of Heat
Transfer, 110, 11751190.
Pereira, R. H., & Nieckele, A. O. (1997). Natural convection in the
evaporator region of household refrigerators. In Proceeding Brazilian
congress mechanical engineering, Bauru, CD rom, paper COB1236.
Radermacher, R., & Kim, K. (1996). Domestic refrigerators: recent
developments. International Journal of Refrigeration, 19(1), 6169.
Ramesh, N., & Venkateshan, S. P. (1999). Eect of surface radiation on
natural convection in a square enclosure. Journal of Thermophysics
Heat Transfer, 13(3), 299301.
Sanchez, A., & Smith, T. F. (1992). Surface radiation exchange for two-
dimensional rectangular enclosures using the discrete-ordinates
method. Journal of Heat Transfer, 114, 465472.
Silva, L. W., & Melo, C. (1998). Heat transfer characterization in roll-
bond evaporators. MSc dissertation, Federal University of Santa
Catarina, Brazil.
Tian, Y. S., & Karayiannis, T. G. (2000). Low turbulence natural
convection in an air lled square cavity: part I: the thermal and uid
ow elds. International Journal of Heat and Mass Transfer, 43,
849866.
Velusamy, K., Sundarajan, T., & Seetharamu, K. N. (2001). Interaction
eects between surface radiation and turbulent natural convection in
square and rectangular enclosures. Transaction ASME, 123,
10621070.
Yang, K. T. (1987). Natural convection in enclosures. Handbook of single-
phase heat transfer. New York: Wiley.
156 O. Laguerre et al. / Journal of Food Engineering 81 (2007) 144156