Desalination 344 (2014) 371–382

Contents lists available at ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Modeling of simultaneous transfers of heat and mass in a trapezoidal

solar distiller
M.S. Maalem ⁎,1, A. Benzaoui 1, A. Bouhenna 1
Faculty of Physics, University of Sciences and Technology Houari Boumediene (USTHB), BP.32 EL-Alia, Bab-Ezzouar 16111, Algiers, Algeria

H I G H L I G H T S

• Modeling of heat and mass transfer at low temperatures in a trapezoidal cavity

• Simulation of simultaneous transfer of heat and mass in a solar trapezoidal distiller
• Numerical heat and mass transfer in a trapezoidal cavity

a r t i c l e i n f o a b s t r a c t

Article history: A model of heat and mass transfer phenomena in a trapezoidal cavity is established. This work is partly based on
Received 12 April 2013 the experimental results obtained in the case of a trapezoidal-shaped solar distiller. The resolution of the related
Received in revised form 3 March 2014 system of equations gives results that are in a good agreement with those obtained experimentally. The modeling
Accepted 29 March 2014
was made on the basis of a stagnant zone within the closed cavity, whose walls are at different temperatures. The
Available online 3 May 2014
numerical simulation could contribute to the study and design of numerous applications such small solar
Keywords:
distillation units, solar driers and solar greenhouses, among others.
Heat and mass transfer © 2014 Elsevier B.V. All rights reserved.
Closed trapezoidal cavity
Natural convection
Solar distillation

1. Introduction
Many countries, such as those of the south bank of the Mediterranean
Sea, having an important solar energy potential and brackish water, suffer
from the scarceness of drinking water due to low rainfalls. These difficul-
ties can be overcome by small solar distillation units, locally usable and in-
expensive. The operating principle of these units is based on trapping by
greenhouse effects, the solar energy entering a closed cavity. This energy
will serve partly for evaporating the brackish water. The distillate is collect-
ed by condensation on one or several surfaces. In this respect, the most im-
portant point is the outputs of these distillation units. The mastery of the
phenomena of simultaneous heat and mass transfer in the distillers
could contribute to improve their efficiency. To this end, we propose an ex-
perimental study on simultaneous heat and mass transfers in a trapezoidal
cavity with three non-adiabatic walls, in view of developing a more gener-
al mathematical model to simulate the distillation unit. With respect to the
theoretical study, the proposed cavity is dimensioned so that the moist air
flow inside it could be two-dimensional.
⁎ Corresponding author.
E-mail addresses: maalemms@yahoo.fr (M.S. Maalem), abenzaoui@mail.com
(A. Benzaoui), bouhenna_mmn@yahoo.fr (A. Bouhenna).
1
Tel./fax: +213 21247344.
Published research material shows that most of the theoretical studies
[1–4, 6–8] and experimental works [5,8] undertaken on the heat and mass
transfer in trapezoidal and triangular closed cavities, concern differentially
heated systems. Two walls contribute to heat and mass (condensation)
exchange, the others are adiabatic. In the theoretical studies, the flow of
fluid inside these cavities is, most often, treated in laminar regime. Very
few theoretical studies are undertaken in turbulent regime [9].
To the author's knowledge no work regarding two different surfaces
of condensation has been done.
The design of the proposed experimental system takes into account
the results obtained by R. Tripathi and GN Tiwari [10] and Anil Kr Tiwari,
and GN Tiwari [11], which state that increasing the thickness of the strip
of water, in a solar distiller, negatively influences the performance and a
lid inclination of 30° leads to better results than inclination values of 15°
and 45°. The use of experimental results led us to propose correlation
expressions for the average of the Sherwood and Nusselt numbers of
the isothermal flat plate, established for the laminar flow regime.
To model these simultaneous heat and mass transfer phenomena, we
relied on our experimental results [12] and on a preliminary study of the
structure of a laminar and stationary saturated moist air flow, in a cavity
similar in shape and having the same geometrical dimensions as those of
the experiment setup. Writing the global balance equations along with

http://dx.doi.org/10.1016/j.desal.2014.03.042
0011-9164/© 2014 Elsevier B.V. All rights reserved.
372 M.S. Maalem et al. / Desalination 344 (2014) 371–382

the appropriate initial conditions identical to those of the experiment, Glass +Aluminum
leads to the establishment of the theoretical model. The validation of the

Amount of condensed water in (g)

2000
model is performed through comparison with the experimental results.
HS=940 W
The trapezoidal configuration is chosen with three non-adiabatic 1600
walls, and different orientations, nature and temperatures.
The analysis of the experimental results, the study of the structure of Stationary area Aluminum
1200
the flow in the cavity, the formulated model, the resolution method, the
numerical results and the other structure configurations used in solar
distillation are presented below. 800
Glass
Nomenclature 400
Aj Area of the wall j (m2). HS=0 W
D(Ti) Mass diffusion coefficient at temperature Ti, (m2/s). t1
0
Ci Heat capacity of the material medium i, (J/K).
Cp Specific heat at constant pressure, (J/kg K). 0 1 2 3 4 5 6 7
Fij Form factor of surfaces i and j. Time in (hours)
g Gravitational acceleration (m/s2).
T Fig. 1. Experimental results.
Gr Grashoff thermal number.
G⁎r Grashoff mass number.
hij Coefficient of heat exchange between moist air and surface j, i or j = 6 Refers to atmospheric air outside the cavity.
(W/m2 K). r Refers to radiance.
hij⁎ Coefficient of mass exchange between medium i and medium
j (m/s).
hrij Radiance heat exchange coefficient between surfaces i and j 2. Analysis of the experimental results
(W/K).
hb Heat transfer coefficient between the bottom of the tray and The experimental study [12] showed a stationary region defined by a
indoor air, (W/m2 K). constant mass flow (Fig. 1).
he Global exchange coefficient between glass and external envi- In view of defining the simulation model of the distiller operation,
ronment, (W/m2 K). the knowledge of the simultaneous heat and mass transfer modes
Hs Global incident solar flux, (W/m2). within the distiller is required. To have an idea about these modes
l1 Glass width (l1 = 0.70 m). of transfer, the knowledge of the saturated humid air flow structure
l2 Width of strip of water (l2 = 0.63 m). within the distiller experimental setup is needed. To this end, a pre-
l5 Width of aluminum plate (l5 = 0.36 m). liminary study on the humid air flow structure within a theoretical
L Length of cavity (L = 1.60 m). cavity (see Fig. 3), similar and with the same dimensions than
Lj Reference length (m). those of the experimental distiller (see Fig. 2) before a simulation
Lw(Ti) Latent heat of phase change of water at temperature Ti, (J/kg). model is proposed.
mi Mass flow at wall i, (kg/s).
NU L Average number of Nusselt. 3. Study of the flow
Pr Prandtl number.
qij Heat flow received by medium j from medium i, (W). 3.1. Introduction
qij⁎ Heat flow received by medium j from medium i through mass
transfer, (W).qrij: Heat flow received by medium j from medi- The flow of saturated humid air in a cavity (Fig. 3), of the same shape
um i by infrared radiation, (W). and size as that of Fig. 2, where the non-adiabatic wall temperatures are
ρi Density of moist air at temperature Ti, (kg/m3) different and constant, is studied. These temperatures are equal to those
Sc Schmidt number. of their counterparts in the experimental device at time instant t1 corre-
SH L Average Sherwood number. sponding to the constant flow rate, Fig. 1. The theoretical study is then
Ti Average temperature of medium i material, (K). extended to the trapezoidal cavity of Fig. 4.
u Component of velocity along Ox, (m/s).
v Component of velocity along Oy, (m/s). 3.2. Study of the flow
αi Absorption coefficient for medium i.
εi Emission factor of surface i. The equations established for natural convection at steady state, make
φi Concentration of water vapor in air at temperature Ti, (mol/m3). use of simplifying hypotheses that are in agreement with experiment.
λ Thermal conductivity, (W/m K).
μ Dynamic viscosity, (Pa·s). 3.2.1. Basic hypotheses
ρ Density of moist air at temperature T, (kg/m3) – The length (L = 160 cm) of the cavity is about four times its height
σ Boltzmann's constant (l5 = 36 cm); the flow of humid air in the cavity, Fig. 3, is then
τi Total solar radiation absorption coefficient for medium i. assumed to be two-dimensional.
– The air within the cavity is saturated with vapor (condensation on
the walls). Its pressure P is the atmospheric one since the cavity
Subscript communicates with the outside air through recovery pipes. The
e Refers to outside cavity. thermal conductivity λ, the specific heat C p and the dynamic
i or j = 1 Refers to glass. viscosity μ are taken equal to those of saturated moist air at normal
i or j = 2 Refers to water in the tank. atmospheric pressure and at temperature corresponding to time
i or j = 3 Refers to absorbent tank. instant t1, Fig. 1; their values are, respectively, 0.025 (W/m K),
i or j = 4 Refers to moist air inside the cavity. 1006 (J/kg K) and 1.9 · 10−5 (Pa·s).
i or j = 5 Refers to aluminum plate. – Due to the low moist air flow velocities (natural convection), we can
 M.S. Maalem et al. / Desalination 344 (2014) 371–382 373

Fig. 2. Experimental model.

neglect viscous friction and assume that the air is incompressible, Fig. 4. Theoretical model (trapezoidal configuration).
we assume also the Newtonian fluid and obeys the Boussinesq ap-
proximation. 3.2.3. Boundary conditions
– Heat transfer by infrared radiation between walls is neglected due to Interest is paid to the structure of the flow at time instant t1 within
the low temperature difference. the stationary region, Fig. 1. The temperatures of non-adiabatic walls
– The diffusion coefficient of vapor in the air is that of water in air at are 308.6 (K) for glass, 322.6 (K) for water and 294.2 (K) for an alumi-
atmospheric pressure and temperature corresponding to time num wall. These temperatures are taken as boundary conditions for
instant t1, Fig. 1. Its value is 0.22 · 10−4 (m2/s). the non-adiabatic walls.
The concentrations are calculated [2] using the temperatures of
3.2.2. Equations the non-adiabatic walls. Their values are φ1 = 2.24 (mol/m3) for glass,
The governing equations are the following: φ2 = 4.99 (mol/m3) for water and φ5 = 1.02 (mol/m3) for the alumi-
num surface.
– Momentum equations.
The initial conditions are identical to those at the beginning of the
! experiment:
∂u ∂u 1 ∂p Ëc ∂2 u ∂2 u
u þv ¼− þ þ
∂x ∂y ρ ∂x ρ ∂x2 ∂y2 – The temperatures are: 300.2 (K) for glass, 294.2 (K) for water, 294.2
(K) for the aluminum plate and 294.2 (K) for moist air.
– The concentration is determined at ambient temperature. Its value
!
∂v ∂v 1 ∂p Ëc ∂2 v ∂2 v is φ0 = 1.02 (mol/m3).
u þv ¼− þ þ þ gβðT−T c Þ – The velocities are zero at the walls (no slip condition) and at the
∂x ∂y ρ ∂x ρ ∂x2 ∂y2
initial instant.

– Continuity equation.
3.2.4. Numerical resolution
∂u ∂v The system of equations is solved by the finite element method. The
þ ¼0 computer code COMSOL Multiphysics available at CDER (Center for
∂x ∂y
Development of Renewable Energies) is used.
– Heat transport equation.
3.2.5. Results and discussion
!
2 2
∂T ∂T k ∂ T ∂ T
u þv ¼ þ 3.2.5.1. Velocity field. The streamline and surface velocity fields are
∂x ∂y ρ:C p ∂x2 ∂y2 depicted in Figs. 5 and 6, respectively. The y-velocity (for five values
of x) versus y, and x-velocity versus x (for five values of y), are depicted
– Mass transport equation. in Figs. 7 and 8, respectively.
! Far from the walls, we can assume, at a first approximation that the
∂φ ∂φ ∂2 φ ∂2 φ flow is horizontal at a very low velocity and air may be considered as
u þv ¼D þ
∂x ∂y ∂x2 ∂y2

Fig. 3. Theoretical model (triangular configuration). Fig. 5. Streamlines velocity field.

374 M.S. Maalem et al. / Desalination 344 (2014) 371–382
Fig. 6. Surface velocity field.
being quasi still. Indeed, Figs. 5 and 6 show three low strength vortexes
and almost horizontal streamlines. Fig. 7, in turn, shows insignificant y-
velocity values compared with those in the vicinity of the walls. Fig. 8,
on the other hand, shows relatively low x-velocity values which in-
crease as y increases slightly. For a fixed y, horizontal velocity value is
roughly constant. However, near the walls, y-velocity and x-velocity
values become more significant, which means that the main flow occurs
between the walls and a quasi motionless air core.
3.2.5.2. Temperature field. The isotherms and the surface temperature are
depicted in Figs. 9 and 10, respectively. The temperature variations
versus y (for five values of x) and the temperature variations versus x
(for five values of y) are represented in Figs. 11 and 12, respectively.
Fig. 9 shows that from the walls the isotherms are at a first approx-
imation parallel and equidistant. This reflects an upward-conduction
heat transfer in the stagnant zone. Fig. 11, on the other hand, shows
two areas. In the first one (0 m b y b 0.08 m), the temperature gradients
increase as y decreases. In this area, both transfer modes, conduction
and convection, are present. In the second area (0.08 m b y b 036 m),
the temperature gradient is constant. This reflects an upward-transfer
by sole conduction. Figs. 9, 10 and 12 show the thermal stratification
and confirm the ascendant heat transfer. Indeed, the temperature de-
pends only on y and decreases as y increases. It can be seen that, along
the glass, the temperature gradients are relatively more significant at
the top as compared to the gradients at the bottom part. The reverse
is observed along the aluminum plate.
3.2.5.3. Concentration field. The isoconcentration lines and the surface
concentration are depicted in Figs. 13 and 14, respectively. The
Fig. 7. y-Velocity.
Fig. 8. x-Velocity.
concentration variations versus y (for five values of x) and the concen-
tration variations versus x (for five values of y) are depicted in Figs. 15
and 16, respectively.
At the first approximation, horizontal isoconcentration lines can be
assumed parallel in the zone far from the walls. Indeed, Fig. 13 shows
that they are almost parallel to the surface of water. They tighten as y
goes down to zero. This reflects, as for the temperature field, diffusion
and convection mass transfer of water into the air. Beyond y =
0.08 m, they relax and become almost equidistant. Their concentration
decreases with y while the corresponding gradient is constant. This re-
flects an ascendant transfer by the sole diffusion. In the vicinity of the
non-adiabatic walls, the concentration gradients are significant (tight
lines of equal concentration) and larger at the top than at the bottom
part, along the glass. The reverse is observed in the vicinity of the
aluminum plate.
Figs. 15 and 16 confirm the conclusions relative to Fig. 13. Indeed, for
y b 0.08 m, Fig. 15 shows that the concentration gradients increase as y
decreases. This reflects an ascendant mass flow. For values of y N 0.08 m,
the concentration gradient is roughly constant. It reflects an ascendant
mass flow by diffusion (constant gradient). Figs. 14 and 16, in turn,
show that the concentration along x depends only on y, and for a
given y, the concentration is nearly constant. As y decreases to about
0.08 m, the concentration increases. This reflects a mass transfer by
diffusion in the direction of y N 0. Figs. 14–16 also confirm that, near
the walls, a mass exchange by diffusion and convection between the
main flow, the walls (condensation) and the layer of stagnant air
(as for the temperature field) takes place.
This shows an area that can be considered as stagnant at first ap-
proximation. The heat and mass transfer occur between this area and
Fig. 9. Isotherms.
 M.S. Maalem et al. / Desalination 344 (2014) 371–382 375

Fig. 12. Temperature versus x.

Fig. 10. Surface temperature.

temperature is uniform. The heat and mass transfer coefficients are de-
duced from correlated relations of the mean Nusselt and Sherwood
the non-adiabatic walls. Within this stagnant zone, transfer of heat and
numbers proposed in the experimental part [12].
mass is due to the sole diffusion mode.
Under the same conditions, the steady state flow of saturated moist
4.2. Physical problem
air in the cavity is studied, (Fig. 4). In the latter, the dimensions of the
water strip and the temperatures of the non-adiabatic walls are identi-
The physical problem, summarized in Fig. 29, consists of determin-
cal to those discussed above. As shown in Figs. 17–19, for around
ing, according to the meteorological parameters and to the experimen-
the values of the ratios 0 ≤ ba ≤0:38, the results are similar to those of
tal setup, the evolution with time of T1, T2, T3, and T5 of the condensed
the triangular configuration. However, for ratios greater than 0.38
water (PG) on the glass pane and on the aluminum plate (PS).
(Figs. 20–22), the increasing deformation of the velocity field is not
in agreement with the hypothesis of the stagnant zone, particularly
4.3. Simplifying hypotheses
above the evaporation surface.
The variations of the velocity components, the temperature and the
The retained hypotheses for the physical problem are:
concentration versus x (for five values of y) and versus y (for five values
of x) in the limit case of ab ¼ 0:38, are represented in Figs. 23–28. From – The temperatures T1, T2, T3, T4 and T5, are assumed to be uniform.
these figures, it can be seen that the derived results are similar to – The temperature T7 is assumed to be constant along the aluminum
those obtained in the case of the triangular configuration. plate, due to low condensation rate (below 0.11 (g/s) Fig. 1) and,
relatively, high flow rate for the cooling water.
4. Mathematical formulation – Heat and mass are transported from the water strip to the stagnant
zone and then to the condensing walls (glass panel and aluminum
4.1. Theoretical model plate).
– The lateral walls at z = −0.80 m and at z = +0.80 m are assumed
The proposed model is based on the experimental results [12] and adiabatic.
those obtained in the above study. Indeed, the former are used for deter- – In the receptacle, heat is evacuated at the bottom (which constitutes
mining the heat and mass exchange coefficients through the correlated 92% of its external surface).
relationships of the average Nusselt and Sherwood numbers. The latter – The moist air within the stagnant zone is assumed saturated (con-
ones show a stagnant zone in the cavity with which the non-adiabatic densation on the walls). The pressure is at the atmospheric level
walls exchange heat and mass. Also, a model of simulation of the exper- (the experimental cavity communicates with the outside air through
imental device is chosen based on global balances incorporating a the recovery pipes of the condensed water) and does not exchange
stagnant zone. Although the above preliminary study has shown a ther-
mal stratification in the stagnant zone, the model assumes that its

Fig. 11. Temperature versus y. Fig. 13. Isoconcentration lines.

376 M.S. Maalem et al. / Desalination 344 (2014) 371–382

Fig. 16. Concentration versus x.

Fig. 14. Surface concentration.


with: q32 ¼ h32  A2  ðT3 −T2 Þ; q24 ¼ h24  A2  ðT2 −T4 Þ; q 24 ¼
heat by radiation with the cavity walls, the factor mair.cp is assumed 
h 24  A2  LW ðT2 Þðρ4 −ρ2 Þ , qr25 = hr25 ⋅ (T2 − T5).
negligible.
– At the absorbing receptacle
– Also, the following will be neglected:
• Heat absorbed by condensed water and by water level readjust- dT 3
ment, C3 ¼ τ3 H s −q32− qb ð3Þ
dt
• the multiple reflections inside the glass pane, the water strip and
the cavity, with: qb = hb ⋅ A3 ⋅ (T3 − T4).
• the water vapor excess, in the stagnant zone, due to its tempera- – At the aluminum plate
ture increase, and
• the losses of steam in the cavity. dT 5 
– The diffused solar flux is homogeneous. C5 ¼ τ5 H S þ q45 þ q45 þ qr15 þ qr25 −Q e5 ð4Þ
dt

4.4. The governing equations with: q45 ¼ h45  A5  ðT4 −T5 Þ; q 45 ¼ h 45  A5  LW ðT5 Þðρ5 −ρ4 Þ;
qr15 ¼ hr15  ðT1 −T5 Þ , qr25 = hr25 ⋅ (T2 − T5) and Qe5 = he5 ⋅ A5 ⋅
Considering the above hypotheses, the heat and mass balance can be (T5 − T7).
written as: – At the stagnant zone
⁎ Heat balance
– At the glass pane
  
0 ¼ q24 þ q 24 −q41 −q 41 −q45 −q 45 : ð5Þ
dT 
C 1 1 ¼ α 1  H s −Q e1 þ q41 þ q 41 þ qr21 −qr15 ð1Þ
dt 
⁎ Mass balance (the vaporized water flow mass)ṁ2 ¼ L q ð24T Þ and those
W 2
of condensed water on the glass pane and on the aluminum plate
with:  
 
Q e1 ¼ he1  A1  ðT1 −T6 Þ; qr41 ¼ hr41  A1 ðT4 −T1 Þ; q 41 ¼ h 41  A1  ṁ1 ¼ L q ð41T Þ andṁ5 ¼ L q ð45
T Þ, respectively. Then the mass balance at the
W 1 W 5

LW ðT1 Þ; ðρ1 −ρ4 Þ, qr21 = hr21. (T2 − T1) and qr15 = hr15 ⋅ (T1 − T5). stagnant zone becomes
– At the water strip
q 24 q 41 q 45
¼ þ : ð6Þ
dT  LW ðT2 Þ LW ðT1 Þ LW ðT5 Þ
C 2 2 ¼ τ2 H s þ q32 −q24 −q 24 −qr21 −qr25 ð2Þ
dt

Fig. 15. Concentration versus y. Fig. 17. ba ¼ 0.

 M.S. Maalem et al. / Desalination 344 (2014) 371–382 377

Fig 20. a
¼ 0:40.
Fig. 18. ba ¼ 0:3. b

So, the simultaneous heat and mass transfer within the cavity are • The surfaces of the walls (glass pane, aluminum surface and water
governed by Eqs. (1)–(6) with the dependent variables T1, T2, T3, T4, strip) as gray; they are diffused and their physical properties do
T5 and ρ4. not depend on the wavelength of the radiation.
• The surface temperatures and heat fluxes are uniform.
4.5. Coefficients in the equations system • The radiation heat exchange is restrained to window pane/water
strip, aluminum surface/window pane and water strip/aluminum
The resolution of the system of the previous equations requires surface. In addition, these three surfaces are assumed to be
knowledge of the different coefficients. For text shortening, only their adjacent.
expressions and how they were determined as well as the used refer- The coefficient of exchange by radiation between surfaces i and j is:
ences are given. All coefficients are expressed in the International Sys-
  
tem (IS).
σ T2j −T2i T j þ Ti
– Outside global heat exchange coefficient, he1, by conduction, convec- hrij ¼ ðW=KÞ:
1−ϵ i 1−ϵ j 1
tion and radiation through the window's pane. þ þ
Ai ϵ i A jϵ j Ai F ij
By conduction and convection [13], he = 5.7 + 3.8 V where V is the
wind speed (m/s). By radiation [14], hre = 5.67 · ε1 where ε1
[15]
≈ 0.85. By conduction, convection and radiation:
Replacing the parameters by their numerical values, expressed in In-
  ternational System, in the case of the employed experimental device,
2
he1 ¼ he þ hre ¼ 10:7 þ 3:8 V W=m K : the following expressions are found:
– Between the glass pane and water strip: hr21 = 3.58 ⋅ 10−9(T22 +
– Radiation heat exchange coefficient, hrij. T21)(T2 + T1) (W/K).
Because of the relatively low surface temperatures, the radiation – Between the glass pane and aluminum plate: hr15 = 2.99 ⋅ 10−9(T21 +
heat exchange is essentially in the infrared range. Therefore, we T25)(T1 + T5) (W/K).
consider the following: – Between the water strip and aluminum plate: hr25 = 3.58 ⋅ 10−9(T22 +
T25)(T2 + T5) (W/K).

Fig. 19. ba ¼ 0:38. Fig. 21. ba ¼ 0:50.

378 M.S. Maalem et al. / Desalination 344 (2014) 371–382

Fig. 24. x-Velocity versus x.

Fig. 22. ab ¼ 2:0.

he5.
– Exchange coefficient between the absorbing receptacle and the water It is experimentally evaluated in the range corresponding to the
strip, h32. stationary region, Fig. 1. The main heat input to the aluminum plate
This coefficient is determined using the expressions recommended by is due to condensation. It takes its maximum value in the stationary
McAdams [16]. region, Fig. 1. This value is:

  
q45 ¼ m 5 Lw ðT5 Þ≈0:11  10
−;3 3
ðkg=sÞ:2458  10 ð J=kgÞ
T3 þ T2
λ2   ¼ 270; 38 ðWÞ:
2 −2 −1
h32 ¼ NU L W m K
L2
Experience shows that the maximum deviation between the temper-
    ature of the cooling water and that of the aluminum wall is less than
T3 þT2 T 3 þT 2
where λ2 2 is the thermal conductivity of pure water at 2 , 0.5 K (the temperature of the wall is about the same as that of the
L2 ¼ Lþl2
is the characteristic length and NU L is the mean Nusselt cooling water). The coefficient is then evaluated, as shown below,
2
number. and its value is taken as being constant throughout the simulation.
– Exchange coefficient between the absorbing receptacle and the air within  
q45 −2 −1
the underneath zone, hb. he5 ¼ ≈930 W m K
A5 ðT 5 −T 7 Þ
This coefficient is determined experimentally in the temperature
range corresponding to the stationary region, as shown in Fig. 30; it
– The natural convection thermal and mass exchange coefficients hij, and
is expressed as: 
h ij .
dT 3 They are determined from the mean correlated Nusselt and Sherwood
τ3 HS −h32  A2  ðT3 −T2 Þ−C 3   numbers, established in part I [12]:
dt −2 −1
hb ¼ W m K – For the glass pane (bottom face)
A3  ðT3 −T4 Þ
 qffiffiffiffi 1=4:
where, A3 is the external receptacle surface, and T2, T3 and T4 are the S1=4
c Gr þ PSc  GTr
¼ 1:667  :
r
mean temperatures at instant t2. SHL1c  1=4
1 þ 0:952
– Exchange coefficient between the aluminum surface and cooling water, S c

Fig. 23. y-Velocity versus y. Fig. 25. Temperature versus y.

 M.S. Maalem et al. / Desalination 344 (2014) 371–382 379

Fig. 26. Temperature versus x. Fig. 29.

Heat transfer:
By convection and conduction
By mass transfer

. By infrared radiation

Temperatures
T1 Glass
T2 Water strip
T3 Absorbing receptacle
T4 Stagnant zone
T5 Aluminum plate
T6 Ambient environment

. T7 Cooling water

– For the water strip

  1=3
SHL2c ¼ 0:098  Gr  Sc :
Fig. 27. Concentration versus y.

The mean Nusselt number is deduced from the Sherwood one by

qffiffiffiffi
– For the aluminum plate the following relation NU L ¼ PS r SH L :
c

The heat and mass transfer coefficients, between the stagnant

qffiffiffiffi zone and non adiabatic walls, are then expressed as:
 
1=4  T 1=4
Sc Gr þ PSc  Gr – For the glass pane: h41 ¼ λL11 NU L1c (W/m2 K) and h 41 ¼ DðL1
T1Þ
SHL1c

SHL5c ¼ 0:333  :
r
 1=4
(m/s)
1 þ 0:952 
S c
– For the aluminum plate: h45 ¼ λl55 NU L5c (W/m2 K) and h 45 ¼ Dlð5T Þ
SH L5c (m/s).

– For the water strip: h24 ¼ λL22 NU L2c (W/m2 K) and h 24 ¼ DðLT22 Þ

Stationary region
325
Temperature of the receptacle (°K)

320

315

310

305

300

295 T3 Experimental
t2
290
0 1 2 3 4 5 6 7
Time (in hours)

Fig. 28. Concentration versus x. Fig. 30. Temperature of the receptacle versus time.
380 M.S. Maalem et al. / Desalination 344 (2014) 371–382

312 330

Tempertures of the aluminum (°K)

Temperture of the glass (°K)

308
320 T5: Theoritical
304 T5: Experimental
310
300
T1 Theoritical
296 T1 Experimental 300

292
290
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Time (in hours) Time (in hours)

Fig. 31. Average temperature of glass versus time. Fig. 34. Average temperature of the aluminum plate versus time.

328 1600

Amount of condensed water in (gr)

Temperature of the water (°k)

324 1400
320 1200
316 1000
312
800
308
600
304 PS Experimental
T2 Theoritical 400
300 PS Theoritical
T2 Experimental
200
296
292 0

0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8

Time (in hours) Time (in hours)

Fig. 35. Amount of condensed water on the aluminum wall versus time.
Fig. 32. Average temperature of water versus time.

SH L2c (m/s). To keep the solar radiation incidence angle on the glass constant
L1 and L2 are respectively the half perimeter of the glass pane during exposure to sun, the experimental setup is reoriented about
and the water strip. every 15 min. Considering the geographical coordinates of the site, the
– Solar absorption coefficient of the glass, α1, total solar absorption inclination of the glass pane, the diffused solar flux, the reflected solar
coefficients of the water strip, τ 2 , of the aluminum plate, τ 5 , and flux by the aluminum plate, the physical properties of the glass pane,
absorbing receptacle, τ3. the aluminum plate, the water strip and the absorbing receptacle, the
Each of these coefficients is defined as the ratio of the solar calculations give:
flux, absorbed by the medium under consideration, over the
incident solar flux, HS (Fig. 29): α 1 ¼ Absorbed solar Hflux S
by the glass
; τ2 ¼ α 1 ¼ 0:08; τ2 ¼ 0:00313; τ3 ¼ 0:646 and τ 5 ¼ 0:015 :
Absorbed solar flux by the water strip Absorbed solar flux by the aluminum plate
HS and τ 5 ¼ HS
and τ3 ¼ Absorbed solar fluxHSby the receptacle :

600
Amount of condensed water in (gr)

328
Temperture of the receptacle (°K)

324 500

320
400
316
312 300

308 200
304 PG Experimental
T3 Theoritical
100 PG Theoritical
300
T3 Experimental
296 0
292
-100
0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8
Time (in hours) Time (in hours)

Fig. 33. Average temperature of the receptacle versus time. Fig. 36. Amount of condensed water on the glass versus time.
 M.S. Maalem et al. / Desalination 344 (2014) 371–382 381

6. Numerical resolution

The governing equations consist of four first order differential

Eqs. (1)–(4) and two algebraic Eqs. (5) and (6). The fourth order
Runge Kutta method is used.

7. Results and comments

Figs. 31–36 depict the theoretical results along with the correspond-
ing experimental ones. The numerical results are in very good agree-
ment with the experimental ones. Indeed, a maximum deviation of
less than 1.5 (K) for the temperature of the window pane, 3 (K) for
the water temperature, 2 (K) for the receptacle temperature, 0.5 (K)
for the temperature of the aluminum plate, 170 (gr) for the production
of the glass pane (which constitutes about the quarter of the total

Fig. 37. Rectangular cavity: case where L/e = 6.

5. Important remark

From the starting instant (about t = 1 h) of the production of glass to

its end (approximately t = 4 h), (Fig. 1), the heat and mass transfer co-
efficients show weak variations in high and low running temperatures.
These coefficients have then been assigned their mean values and
injected into the calculation program. These values, expressed in the In-
ternational System units, are given below:

   
2 2
h41 ¼ 10 W=m K ; h41 ¼ 0:0045
 ðm=s
 Þ; h45 ¼ 5 W=m K ; h45
2
¼ 0:00235
 ðm=sÞ;h24 ¼ 5 W=m K ; h24 ¼ 0:0028 ðm=sÞ; h32
2 2
¼ 239 W=m K and hb ¼ 8:87 W=m K : Fig. 38. Inclined rectangular cavity: case where L/e = 5 and α = 30°.
382 M.S. Maalem et al. / Desalination 344 (2014) 371–382
Fig. 39. Shaped cavity chapel: case where e/b = 0.5.
production) and 150 (gr) for the aluminum plate is observed. However,
concerning the glass pane, Fig. 36, although the form of the curve is
physically acceptable, relatively higher gaps are observed, as compared
with the other curves. These differences are due, in the authors' opinion,
to the hypothesis of uniform temperature in the stagnant zone and the
glass pane. Indeed, the experimental results (temperature records over
three levels of the glass pane) show that the temperature is not uni-
form; it decreases from bottom to top. For the overall output of con-
densed water (glass + aluminum plate), the difference is less than 7%.
8. Other structure configurations used in solar distillation
The elaborated model could be applied to other cavity shapes such as
rectangular, titled rectangular, chapel and differentially heated cavity
configurations. The results obtained so far, in the same conditions as
for the studied case, related to the first three configurations are present-
ed in the following (see Figs. 37–39). These results take account of the
ratios e/b and L/e as well as to the tilting angle α.
The velocity fields show the existence of a stagnant air core in the
cavity in all the different configuration cases, as presented above.
9. Conclusion
The proposed model is reliable since the numerical temperature
curves and condensed water production are close enough to those ob-
tained experimentally. On the other hand, the gaps between numerical
and experimental results are not significant: for the total production of
condensed water, the deviation is less than 7%.
This model can be useful and interesting for many applications such
as the evaluation of calorific losses by heat transfer (solar panels and
solar water heater) or by simultaneous heat and mass transfer (solar
distillation, solar dryer and solar greenhouse). However, in the authors'
opinion, this model can be refined, particularly, concerning the glass
pane and the stagnant zone, where it is assumed that the temperature
is uniform, although the preliminary study of the flow, in the theoretical
cavity and stationary regime, shows a thermal stratification in this zone.
The case where the temperature of the stagnant zone is not uniform
is under study and the related outcome will be given in a later paper.
Furthermore, the theoretical study carried out for the case of differ-
ent cavity structure configurations employed in solar distillers, namely
rectangular, titled rectangular and chapel showed evidence of the exis-
tence of the stagnant air core. Hence the simplifying hypothesis and the
elaborated model could be extended to other configurations.
References
[1] D.A. Saville, S.W. Churchill, Simultaneous heat and mass transfer in free convection
boundary layers, Fluid Mech. 29 (1967) 391.
[2] B. Gebhart, L. Pera, The nature of natural convection flows resulting from the com-
bined buoyancy effect of thermal and mass diffusion, Int. J. Heat Mass Transfer 14
(1971) 2025–2050.
[3] G.N. Tiwari, A. Minocha, P.B. Sharma, M. Emran Khan, Simulation of convective mass
transfer in a solar distillation process, Energy Convers. Manag. 38 (1997) 761–770.
[4] M. Boussaid, A. Mezenner, M. Bouhadef, Convection naturelle de chaleur et de masse
dans une cavité trapézoïdale, Int. J. Therm. Sci. 38 l (1999).
[5] D.J. Reynolds, M.J. Jance, M. Behnia, G.L. Morrison, An experimental and computa-
tional study of the heat loss characteristics of a trapezoidal cavity absorber, Sol.
Energy 76 (2004) 229–234.
[6] S. Kumar, Natural convective heat transfer in trapezoidal enclosure of boxtype, Sol.
Cooker Renew. Energy 29 (2) (2004) 211–222.
[7] Natarajan, Basak Tanmay, S. Roy, Natural convection flows in a trapezoidal enclo-
sure with uniform and non-uniform heating of bottom wall, Int. J. Heat Mass Trans-
fer 51 (3–4) (2008) 747–756.
[8] Prabal Talukdar, Conrad R. Iskra, Carey J. Simonson, E: Combined heat and mass
transfer for laminar flow of moist air in a 3D rectangular duct: CFD simulation
and validation with experimental data, Int J Heat Mass Transf 51 (2008) 3091–3102.
[9] E. Papanicolaou, V. Belessiotis, Double-diffusive natural convection in an asymmet-
ric trapezoidal enclosure: unsteady behavior in the laminar and the turbulent-flow
regime, Int. J. Heat Mass Transf. 48 (2005) 191–209.
[10] R. Tripathi, G.N. Tiwari, Effect of water depth on internal heat and mass transfer for
active solar distillation, Desalination 173 (2005) 187–200.
[11] Anil Kr Tiwari, G.N. Tiwari, Effect of the condensing cover's slope on internal heat
and mass transfer in distillation: an indoor simulation, Desalination 180 (2005)
73–88.
[12] M.S. Maalem, A. Benzaoui, Heat and mass transfer at low temperatures, in a trape-
zoidal closed cavity, experimental study, Adv. Mater. Res. 550–553 (2012)
2996–3003.
[13] M.S. Sodha, Ashvini Kumar, A. Srivastava, Tiwari, Thermal performance of still on
roof system, Sol. Energy Convers. 20 (1980) 181–190.
[14] E.M. Sparrow, R.D. Cess, Radiation heat transfer, augmented edition, McGraw-Hill
book, Company, 1988.
[15] H.C. Hotel, A.F. Sarofim, Radiative Transfer, McGraw-Hill, New York, 1967.
[16] W.H. McAdams, Heat Transmission, McGraw Hill, New York, 1954.