See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/343385874

Numerical Study of Natural Convection in an Open- Ended Channel:

Comparison of Characteristic Quantities Between Air and Water

Article · January 2016

CITATIONS READS

0 103

5 authors, including:

Amine Zoubir Imad Kadiri

Ecole Nationale des Travaux Publics de l'Etat Université Moulay Ismail
11 PUBLICATIONS 227 CITATIONS 35 PUBLICATIONS 301 CITATIONS

SEE PROFILE SEE PROFILE

Khalid Sbai Rahmoune Miloud

Université Moulay Ismail Université Moulay Ismail
61 PUBLICATIONS 234 CITATIONS 135 PUBLICATIONS 623 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Khalid Sbai on 02 August 2020.

The user has requested enhancement of the downloaded file.

 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05
Numerical Study of Natural Convection in an Open-
Ended Channel: Comparison of Characteristic
Quantities Between Air and Water
Agounoun Rachid 1, Zoubir Amine 1, Kadiri Imad 2 , Sbai Khalid 1, Rahmoune Miloud 1,
1
Laboratoire d’Etude des Matériaux Avancés et Applications (LEM2A), Université Moulay Ismail, Faculté des sciences, Ecole
Supérieure de Technologie Meknès, Maroc
2
Université Cadi Ayyad, Faculté des sciences et techniques de Marrakech, Morocco
Corresponding author: r.agounoun@est-umi.ac.ma
Abstract-- The present study deals with natural convection
flow inside vertical channel with one wall heated at uniform heat
flux and the other wall isolated. The purpose of this numerical
investigation is to study the differences on flow configurations in
pure natural convection between air and water in vertical channels.
A comparison in terms of Nusselt number and mass flow rate is
then discussed. The numerical code is developed using Finite
Differences scheme to solve the Navier-stokes equations under the
Boussinesq assumption in two dimension. The comparison in terms
of characteristic quantities shows that Nusselt number are slightly
different between air and water, flow patterns are completely
different and adimensionless velocity and flow rates are more
important in water than in air. Physical understanding of these
differences is discussed. Using water to model natural convection in
an air vertical channel gives reasonable Nusselt numbers but only
for small modified Rayleigh numbers. Discrepancies increases with
increasing the modified Rayleigh number.
Index Term-- Natural convection; Open-Ended Channel.
I. INTRODUCTION
Cooling by natural convection in a vertical channel has been
used in several industrial applications. In fact, the channel is
representative of problems such as chimney, solar wall, Trombe
wall, electronic components of computers, or double-skin
façade. Two approaches emerge from the studies carried out in
this context. A macroscopic approach which provides analytical
expressions assuming uniformity of global quantities such as
velocity profiles and temperatures. For exemple, Brinkworth [1]
determined and analytical expression for the mass flow for
different heating power and gave head losses between the inlet
and the outlet of the channel.
A second numerical approach based on detailed experimental
studies was also apprehended. The first experiment reported in
this context was that Elenbaas [2] for an air-filled channel which
walls are heated to uniform temperature, he studied various flow
regimes according to the modified Rayleigh number (10 1 ≤ Ra* ≤
105), he has also provided correlations in terms of dissipation
rate and Nusselt number for the same range of Ra *. Bodoia and
Osterle [3] presented the first numerical results in agreement
with the results of Elenbaas [2] only in the case of small numbers
Ra*. For the case of large Ra*, the results showed more disparity
(11% of difference). This discrepancy can be explained by the
fact that in the study of Bodoia a uniform velocity was used as a
boundary condition at the entrance of the channel, which is not
sufficient to consider the case where the flow regime reaches the
boundary layer regime (large Ra*).
165205-3737-IJMME-IJENS © October 2016 IJENS
1
Other numerical and experimental studies were presented
later, Aung [4], Aung et al. [5], Kihm et al. [6], Sparrow and
Bahrami [7], Sparrow et al. [8], Webb and Hill [9] for both
heating temperature and uniform flow imposed. Some authors
have considered adiabatic extensions and artificial volumes at
the input and output of the channel Kettleborough [10], Naylor et
al. [11], Morrone et al. [12], Campo et al. [13], Bianco and
Nardini [14], Giroux-Julien et al. [15]. They showed interest to
add extensions to a parallel channel whose walls are heated at
uniform temperature or heat flux. They give the results in terms
of maximum temperatures on the walls, mass flow rate, and
pressure profiles. They also obtain correlations of mass flow rate
and maximum wall temperature as a function of the modified
Rayleigh number and the extension spect ratio. Later, other
correlations in terms of Nusselt number and optimum spacing
for different modified Rayleigh numbers were given by Bar-
Cohen and Rohsenow [16]. These correlations were improved
and compared by Olsson [17]. He also refers to an optimum
spacing between the walls of the channel for a better heat
transfer. These previous investigations have focused only on the
study of pure convection and neglected heat exchange due to
surface radiation. They also overlook or minimize conduction in
heated walls. Most of these studies have been done using air as
working fluid and the radiation. They also overlook or minimize
conduction in heated walls. Most of these studies have been
done using air as working fluid and the radiation part of heat
transfer is still not completely known. In order to contribute in
the understanding of the physical mechanisms of pure natural
convection, some authors used water as a working fluid in order
to get rid of radiation heat transfer. Miyatake et al. [18]
performed both analytical study and experiment for natural
convection between vertical parallel plates heated with uniform
heat flux. Experiments were performed with water as the
working fluid, and results were compared to the theoretical
predictions with good agreement. Sparrow et al. [8] investigated
the flow reversal by visualizing water flows in a vertical channel
heated on one side. They also made a numerical study using a
parbolic model to solve the Navier-Stokes equations. They
underlined that Nu (Ra*) correlations are slightly different
between air and water and difference increases with Ra*.
Daverat [19] studied a pure convective flow between two parallel
plates placed in a water box. Their velocity profiles and
temperature averages reveal the presence of a change in flow
regime inside the channel for an identified modified Rayleigh
number. They also study the effect of thermal stratification on
the flow and temperature. Few experimental and numerical
IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05 2

studies on pure natural convection between parallel plates have

been carried out with water. Mitra et al. [20] made a review of
Nu (Ra*) correlations of only five studies. This study shows how
few researches exist on the subject. In addition, most of the
investigations were limited to thermal measurements and heat
transfer analysis, no studies showed the differences in flow
patterns between air and water in vertical channels. In order to
contribute in the understanding of the physical mechanisms of
pure natural convection, we present a comparison study in terms
of thermal and kinematic quantities for pure convective flow
between using water and air. The numerical study performed
here, used an elliptic model to solve of Navier-Stokes equations.
In the following, the model and numerical methods that
address the problem of natural convection of fluid (air or water)
between two vertical walls, are detailed. Finally, a comparison
of characteristic quantities between the two fluids is discussed. Fig. 1. Configuration of study: asymmetrically heated vertical channel by a
uniform heat flux.
II. PHYSICAL MODEL AND NUMERICAL METHODS The governing Eq.(3) are transformed as follows
II.1. Numerical Modeling
 U V
The geometrical configuration subject to investigation is   0
shown in Fig. 1. It concerns the simplified (2D) configuration of  X Y
 U U U P   2U  2U 
the experimental apparatus. The wall heating produces a  U V   Pr . Ra 1/2   
 t X Y X  X 2 Y 2 
temperature gradient in the fluid leading to small density   
 V U V V V  P  Pr . Ra 1/2   V   V   Pr
2 2
differences which gives rise to the buoyancy force. Thus, a
 t X Y Y  X 2 Y 2 
natural convective flow occurs with the fluid entering at the   
bottom and leaving at the top. Throughout the analysis the     
1/2  
2
 2 
 t U X V Y  Ra  2  2 
following assumptions are made: fluid properties, except   X Y 
density, are independent of temperature; density variations are (3)
significant only in the buoyancy force; and the flow is two-
dimensional, laminar and incompressible with negligible viscous Where T used to define the adimensional temperature and the
dissipation and surface radiation. Thus, the well-known unsteady
Boussinesq equations in their elliptic form, governing the flow reference velocity is: q w , with q w is the uniform heat flux fixed
are:
0
 0 
 u v at the wall. The Prandtl number Pr is defined by Pr  
  0
 x y
0
  2 2 
 u u u  v u  1 p  v0   u   u 
Furthermore, the Rayleigh number, expressing the strength of
 t    

x y  0 x  x

2 y 2  buoyancy, is constructed with the plate spacing as follows:

v v v  1 p v 0  v   2v  g   0   
 2 (1) 0 gqwb4
 v
 u   Rab 
 t x y  0 x 
 x2 y 2 
    000
 
  
 T
u T v T 0   T2   T2 
2 2 (4)

 t x y  x y 
 
This finally leads to, the modified Rayleigh number (or Elenbaas
number):
Where u, v, p are respectively, horizontal velocity, vertical * Rab
velocityand pressure. ρ₀, v₀, β₀, α₀ are respectively, mass Rab 
Ah
density, kinematic viscosity, thermal expansion coefficient and
thermal diffusivity calculated at some reference temperature. By (5)
introducing the following adimensional quantities with reference Where Ah  Lh is the aspect ratio of the heated part of the channel
b
variables (Lr, Ur, tr), Eq. 2.

(see fig. 1).


Lr b; X  x ; Y  y
 Lr Lr
 1



 0 g Tb  2
Ur   ;U u ;V v





Pr Lr


Lr
 

 T T0 p  p0
 ; P ; tr  b


T U r2 Ur
(2)

165205-3737-IJMME-IJENS © October 2016 IJENS

IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05
 
 X 1; V  0; U  0; X  0; 0Y  A
 
  0; V  0; U  0; X 1; 0Y  A
 X
  0; V  0; U  0; P  0 si V  0 or P  1 V  X , A 
 Y 2
 Si V  0 à Y  0, 0 X 1
  V
  0; 0; U  0; P  0 si V  0 à Y  A, 0  X 1
 Y Y
  0; V  0; P  0 si V  0 à Y  A; 0 X 1
 Y
(6)
The selected boundary conditions for the thermally driven open
channel are shown in system of Eq. (6). When the boundary
conditions are expressed in pressure at the inlet, the generalized
Bernoulli theorem is used to take into account the pressure loss
between the upstream and the entry of the channel (upstream
velocity assumed to be zero and the pressure is equal to p(x; y) =
patm). Indeed, Morrone et al. [12], Marcondes and Maliska [21]
found that the driving pressure at the entrance of the channel,
due to the velocity of the incoming fluid, cannot be neglected,
the influence is even more important as the Rayleigh number
increases. Local Bernoulli pressure is applied at the channel
entrance. Moreover, pressure at the outlet is taken equal to
exterior pressure assumed to be zero (atmospheric pressure in
dimensional form). When boundary conditions are written on
velocity, zero Neumann boundary condition is applied on
streamwise velocity V and homogeneous Dirichlet boundary
condition is applied on transverse velocity U. Actually, we
assume that the incoming flow at the inlet is parallel to the
channel. Concerning the thermal boundary conditions, we
neglect the longitudinal conductive heat flux at the channel
outlet, knowing that the transverse thermal gradients are more
important. If in some case, an incoming flow can appear at the
channel outlet, we assume it has the outside temperature. In this
approximation, we neglect the fact that the incoming fluid may
be slightly heated in the external surroundings.
II.2. Numerical Methods
Equations of conservation of mass, momentum, and energy, are
discretized with finite differences schemes in space and time.
The temporal scheme adopted to discretize the system of
equations 2 is based on the three-level second-order Euler
backward scheme with Adams-Bashforth extrapolation as
proposed in Vanel et al. [22]. In this scheme, we treat implicitly
the diffusion terms and explicitly the convective ones. The
pressure velocity decoupling is achieved by adopting an iterative
solution based on a projection algorithm Chorin [23] Chorin
[24]. This algorithm consists of finding first a velocity
prediction based on pressure field approximation.
Then an equation for pressure correction is derived from the
continuity equation, it's then solved to obtain the pressure
correction to update the velocity fields and pressure. Concerning
the spatial discretization, the computational domaine is
discretized on a non-uniform Cartesian grid. Use is made of a
staggred variable arrangement. A second order centered
difference schemes are used for all variables. A second order
forward and backward difference schemes are used to treat the
Neumann boundary condition on vertical velocity respectively at
the inlet and at the outlet. After disretization, we obtain a
Poisson equation on pressure correction and Helmholtz
equations on velocity and temperature. We solve the Poisson
165205-3737-IJMME-IJENS © October 2016 IJENS
3
and Helmholtz equations respectively using partial
diagonalization method and tridiagonal matrix algorithm. For
more details see Zoubir et al [25].
III. VALIDATION OF NUMERICAL CODE
The thermally driven vertical channel problem has served as a
benchmark problem. The GDR AMETH laboratories G.
Desrayaud and Mojtabi [26] have worked on natural convection
of air between two vertical walls, one heated by a uniform heat
flux and the other insulated. This kind of problems leads to a so-
called flow reversal when channel walls are subjected to
asymmetric heating Sparrow et al. [8] Ingham et al. [27] Kihm
et al. [6]. In this benchmark four dynamic boundary conditions
are tested in order to approach the Webb and Hill
experimentation Webb and Hill [9]. A first validation of
numerical code was done in previous works with those of GDR-
AMETH laboratories Zoubir et al. [25] and Zoubir et al.
[28].The comparison was made for modified Rayleigh number
of Ra* = 105 and an aspect ratio of 10 and different grid
resolutions. A spatial grid analysis was also done and shows that
our numerical code is globally of second-order in space Zoubir
et al. [25]. Furthermore, a good agreement was found with other
works on characteristic quantities (mass flow rate and Nusselt
number).
IV. RESULTS AND DISCUSSION
In the present study, the Rayleigh number based on the channel
width is varied from 105 to 107 and the aspect ratio is taken
equal to 5 and 15. Then, the corresponding modified Rayleigh
numbers are Ra*=104, 2.104, 105, 2.105, 106; 2.106. The Prandtl
numbers used are 0.71 and 7.0, which correspond to air and
water respectively. The details of the fluid flow and the
temperature fields are presented below, with em phasis on
comparison on natural convection results of air and water. Fig. 2
and Fig. 3 present fluid flow configurations of air and water in
the form of isotherms and streamlines. In the fig. 3, a flow
reversal occurs for all configurations. Indeed, the air heated at
the wall rises by natural convection and creates a dynamic
boundary layer along the heated wall. To aliment this boundary
layer, the air enters from the top when the bottom alimentation is
not sufficient. However, the penetration depth of fluid at the
channel top is different for different Rayleigh numbers for air
and water. The fluid reversal penetration depth increases with
Rayleigh number.
In addition, the penetration is more important in the case of
air then that in water. There is also a significant difference in the
growth and thickness of the dynamical boundary layer for water
and air. As expected with the increase on Rayleigh number, the
boundary layer at the heated wall decreases, for both water and
air. Furthermore, the dynamical boundary layer is smaller in air.
In the fig. 2, on the contrary to dynamical boundary layer, the
thermal boundary layer is larger in air then in that of water.
The difference in the boundary layer thickness is due to
Prandtl number effects and to the presence of recirculation of air
at the channel exit near the adiabatic wall. We can also see that
for the first modified Rayleigh number, because of flow reversal,
the adiabatic wall is slightly heated in the air case, this is
subjected by the two isotherms reaching the adiabatic walls.
Fig. 4 and Fig. 5 present the adimensional temperature for A
= 5 and velocity profiles for A = 10 at different channel height
and different modified Rayleigh numbers for both fluids.
IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05 4

Temperatures are greater in air fluid then that in water. The wall
temperature differences between the two fluids vary as a
function of modified Rayleigh number from 17 to 19 % for A =
5 and from 9 to 10% for A = 10. This is due to an important heat
exchange between the heated wall and fluid in water case. It is
observed that an increase in the Prandtl number to Pr = 7 results
in a decrease of the thermal boundary layer thickness and in
lower average temperature within the boundary layer. The
reason is that smaller value of Prandtl number is equivalent to
increase in the thermal conductivity of the fluid and therefore
heat is able to diffuse away from the heated surface more rapidly
for higher value of Prandtl number. Hence there is a reduction in
temperature with increase in the Prandtl number. We can also
see that discrepancies are more important for the lowest aspect
ratio and increase with increasing the modified Rayleigh
number.

Let us now analyze the adimensional vertical velocity (see fig. 5,

Table. I and Table. II. Because of viscosity and flow reversals,
velocity profiles are more homogenized for water then for air
and the adimensional mean velocity and mass flow rate is larger
in water (see also Table III). We can conclude that the flow
characteristics in terms of velocity are different between air and
water.
Table III shows the results of average Nusselt number < Nu >.
Results show that Nusselt number is greater in water then in air.
The differences for Nusselt number varied as a function of
modified Rayleigh number from 9,3% to 22,75% for A = 5 and
from 8% to 23,87 for A = 10. Discrepancies on Nusselt number
are seen to increase with increasing the modified Rayleigh Fig. 3. Streamlines in water (Top) and air (bottom) for A = 5, Rad = 105
(a), Rad = 106 (b) et Rad = 107 (c).
number.

(a)

Fig. 2. Isotherms in water (Top) and air (bottom) for A = 5, Rad = 105 (a), Rad =
106 (b) et Rad = 107 (c).

165205-3737-IJMME-IJENS © October 2016 IJENS

IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05 5

(b)
(a)

(b)
(c)

(c)
(d) Fig. 5. Results in terms of velocity profiles for different heights for water and air
(a,b,c), with Rad = 105, 106, 107 and A = 10.
Fig. 4. Results in terms of temperature profiles at different heights (a), (b),
(c); and wall temperatures for water and air (d), with Ra d = 105 , 106, 107and A =
5.

165205-3737-IJMME-IJENS © October 2016 IJENS

IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05 6

TABLE I p adimensional pressure

THE ABSCISSAS OF THE MAXIMUM VELOCITY IN WATER AND AIR
Pr Prandtl number
FOR DIFFERENT HEIGHTS AND FOR Rad = 105; 106; 107 AND A = 5.
qw wall heat flux (W/ mš)
Rad Rayleigh number based on width
Ra* modified Rayleigh number
T Temperature (K)
t time(s)
tr reference time (s)
u horizontal velocity (m/s)
U adimensional horizontal velocity
Ur reference velocity
v vertical velocity (m/s)
V adimensional vertical velocity
TABLE II x abscissa (m)
THE ABSCISSAS OF THE MAXIMUM VELOCITY IN WATER AND AIR X adimensional abscissa
FOR DIFFERENT HEIGHTS AND FOR Rad = 105; 106; 107 AND A = 10.
y ordinate (m)
Y adimensional ordinate

TABLE III
AVERAGED NUSSELT NUMBERS AND AVERAGED VELOCITIES IN
WATER AND AIR FOR Rad = 105; 106; 107 AND
A = 5; 10.
V. CONCLUSION
We have presented a numerical code for studying natural
convection in an open-ended channel. Finite difference scheme
of second order in time and space was adopted to discretize the
Navier-Stokes equations under the Boussinesq assumption. A
comparison of chracteristics quantities (Nusselt number, flow
rate) between water and air has been done and discrepancies are
discussed. Traditionally used water models to predict the natural
convection of air give reasonable heat transfer results only for
small modified Rayleigh numbers. The fluid flow characteristics
for air are different from those of water. The maximum velocity
and exchange flow rates reflect this point. Further studies need
to be carried out in order to improve both experimental and
numerical approaches and understand the physical phenomena.
NOMENCLATURE
A Aspect ratio (J/b)
d channel width (m)
g gravity acceleration (9.81m/s)
L channel height (m)
Lr reference length (m)
Nu Averaged Nusselt number (eq)
p pressure ((Pa)
165205-3737-IJMME-IJENS © October 2016 IJENS
Greek Symbols
α₀ thermal diffusivity (m²/s)
ΔT temperature difference (K)
λ₀ thermal conductivity (W/mK)
μ₀ kinematic viscosity (m²/s )
ρ density (kg/m3)
θ adimensional temperature
θw adimensional wall temperature
Subscripts
0 at ambient temperature
r reference
w wall
REFERENCES
[1] Brinkworth, B, 2000 “Estimation of flow and heat transfer for the
design of PV cooling ducts,” Solar Energy, 69(5), 413–420,
[2] Elenbaas, W, 1942, “Heat dissipation of parallel plates by free
convection,”Physica, 9(1), 1–28.
[3] Bodoia, J., and Osterle, J., 1962, “The development of free
convection between heated vertical plates,” Journal of Heat Transfer,
84(1), 40–43.
[4] Aung, W., 1972, “Fully developed laminar free convection between
vertical plates heated asymmetrically,” International Journal of Heat
and Mass Transfer, 15(8), 1577–1580.
[5] Aung, W., Fletcher, L., and Sernas, V., 1972, “Developing laminar
free convection between vertical flat plates with asymmetric heating,”
International Journal of Heat and Mass Transfer, 15(11), 2293–2304.
[6] Kihm, K., Kim, J., and Fletcher, L., 1995, “Onset of flow reversal
and penetration length of natural convective flow between isothermal
vertical walls,” TRANSACTIONS-AMERICAN SOCIETY OF
MECHANICAL ENGINEERS JOURNAL OF HEAT TRANSFER,
117, 776–776.
[7] Sparrow, E., and Bahrami, P., 1980, “Experiments on natural
convection from vertical parallel plates with either open or closed
edges,” Journal of Heat Transfer, 102, 221.
[8] Sparrow, E., Chrysler, G., and Azedevo, L., 1984, “Observed flow
reversal and measured-predicted Nusselt number convection in a one-
sided heated vertical channel,” J Heat Trans-T ASME, 106, 325–332.
[9] Webb, B., and Hill, D., 1989, “High Rayleigh number laminar natural
convection in an asymmetrically heated vertical channel,” Journal of
Heat Transfer, 111, 649.
[10] Kettleborough, C., 1972, “Transient laminar free convection between
heated vertical plates including entrance effects,” International
Journal of Heat and Mass Transfer, 15(5), 883–896.
[11] Naylor, D., Floryan, J., and Tarasuk, J., 1991, “A numerical study of
developing free convection between isothermal vertical plates,”
Journal of Heat Transfer (Transactions of the ASME (American
Society of Mechanical Engineers), Series C);(United States), 113(3).
IJENS
 International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:05 7

[12] Morrone, B., Campo, A., and Manca, O., 1997, “Optimum plate
separation in vertical parallelplate channels for natural convective
flows: incorporation of large spaces at the channel extremes,”
International Journal of Heat and Mass Transfer, 40(5), 993–1000.
[13] Campo, A., Manca, O., and Morrone, B., 1999, “Numerical analysis
of partially heated vertical parallel plates in natural convective
cooling,” Numerical Heat Transfer, Part A: Applications, 36(2), 129–
151.
[14] Bianco, N., and Nardini, S., 2005, “Numerical analysis of natural
convection in air in a vertical convergent channel with uniformly
heated conductive walls,” International Communications in Heat and
Mass Transfer, 32(6), 758–769.
[15] Giroux-Julien, S., Menezo, C., Vareilles, J., Pabiou, H., Fossa, M.,
and Leonardi, E., 2009, “Natural convection in a nonuniformly
heated channel with application to photovoltaic facades,”
Computational thermal sciences, 1(3).
[16] Bar-Cohen, A., and Rohsenow, W., 1984, “Thermally optimum
spacing of vertical, natural convection cooled, parallel plates,”
Journal of Heat Transfer, 106, 116.
[17] Olsson, C., 2004, “Prediction of Nusselt number and flow rate of
buoyancy driven flow between vertical parallel plates,” Journal of
heat transfer, 126, 97.
[18] Miyatake, O., Tanaka, H., Fujii, T., and Fujii, M., 1973, “Natural
convective heat transfer between vertical parallel plates- One plate
with a uniform heat flux and the other thermally insulated,” Heat
Transfer-Japanese Research, 2, 25–33.
[19] Daverat, C., 2012, Etude experimentale de la convection naturelle en
canal vertical à flux de chaleur impose: application au
rafraîchissement passif de composants actifs de l’enveloppe des
bâtiments, Ph.D. thesis, INSA de Lyon.
[20] Mitra, A., Dutta, T., and Ghosh, D., 2008, “Natural convective heat
transfer in water enclosed between pairs of differentially heated
vertical plates,” Heat and Mass Transfer, 45(2), 187–192.
[21] Marcondes, F., and Maliska, C., 1999, “Treatment of the inlet
boundary conditions in natural-convection flows in open-ended
channels,” Numerical Heat Transfer, Part B: Fundamentals, 35(3),
317–345.
[22] Vanel, J., Peyret, R., and Bontoux, P., 1986, “A pseudo-spectral
solution of vorticity-stream function equations using the influence
matrix technique,”In Von Karman Inst. for Fluid Dynamics
Computational Fluid Dynamics, Volume 1 11 p (SEE N86-32666 24-
34), vol. 1.
[23] Chorin, A., 1967, “A numerical method for solving incompressible
viscousflow problems,” Journal of computational physics, 2(1), 12–
26.
[24] Chorin, A., 1968, “Numerical simulation of the Navier-Stokes
equations,” Math Comp, 22, 745–762.
[25] Zoubir, A., Daverat, C., Xin, S., Giroux-Julien, S., Pabiou, H., and
Menezo, C., 2012, “Natural Convection in a Vertical Open-Ended
Channel: Comparison Between Experimental and Numerical
Results,” Proceedings of CHT-12, ICHMT, International Symposium
on Advances in Computational Heat Transfer, ICHMT International
Symposium on advances in Computational Heat Transfer.
[26] G. Desrayaud, R. Bennacer, J.C.E.C.A.J.N.L., and Mojtabi, K., May
2007, “Etude numerique comparative des ecoulements
thermoconvectifs dans un canal vertical chauffe asymmetriquement,”
Tech. rep., In 8 eme Colloque Inter Universitaire Franco-Quebecois
sur la Thermique des Systemes, CIFQ2007/ART-06-14, 28-30.
[27] Ingham, D., Keen, D., and Heggs, P., 1988, “Two dimensional
combined convection in vertical parallel plate ducts, including
situations of flow reversal,” International Journal for Numerical
Methods in Engineering, 26(7), 1645–1664.
[28] Zoubir, A., Xin, S., Giroux-Julien, S., and Menezo, C., 2011, “Etude
numerique des transferts thermo-convectifs dans un canal d air
vertical a flux impose,” Congres Francais de Thermique. SFT,
Perpignan, 163–168.

165205-3737-IJMME-IJENS © October 2016 IJENS

IJENS
View publication stats