International Journal of Heat and Mass Transfer 43 (2000) 2089±2099

www.elsevier.com/locate/ijhmt

Entropy generation for natural convection in an inclined

porous cavity
A.C. Baytas°
Istanbul Technical University, Institute for Nuclear Energy, 80626 Ð Maslak, Istanbul, Turkey
Received 30 March 1999; received in revised form 16 September 1999

Abstract

The issue of entropy generation in a tilted saturated porous cavity for laminar natural convection heat transfer is
analysed by solving numerically the mass, momentum and energy balance equations, using Darcy's law and
Boussinesq-incompressible approximation. As boundary conditions of cavity, two opposite walls are kept at
constant but di€erent temperatures and the other two are thermally insulated. The parameters considered are the
angle of inclination and the Darcy±Rayleigh number. When available, present solutions are compared with known
results from the previous researches. Excellent agreement was obtained between results that validate the used
computer code. The results show that the calculation of local entropy generation maps are feasible and can supply
useful information for the selection of a suitable angle of inclination. # 2000 Elsevier Science Ltd. All rights
reserved.

Keywords: Entropy generation; Porous media; Natural convection; Cavity

1. Introduction Nield and Bejan [1] and Ingham and Pop [2] contribu-
ted to a wide overview of this important area in heat
In computational simulation of thermal-hydraulic
unsteady investigations, the quantities to be calculated transfer of porous media. There are many published
are usually temperature, velocity ®elds, enthalpy, press- studies related to natural convection in rectangular
ure, etc.; but rarely contains entropy properties using porous enclosures. Moya et al. [3], Bejan [4], Parasad
the second law of thermodynamics. Principally, only and Kulacki [5], Baytas and Pop [6], Beckerman et al.
heat transfer was re¯ected in natural convection [7], Gross et al. [8], Lai and Kulacki [9], Monale and
studies. The necessity for the utilisation of the second Lage [10], and Walker and Homsy [11] have donated
law of thermodynamics in thermal design decisions is many important results for this problem. Caltagirone
vividly demonstrated in this research work. The and Bories [12] studied the stability criteria of free con-
research of natural convection in porous media has vective ¯ow in an inclined porous layer. Moya et al. [3]
been conducted widely in recent years, which involves investigated the natural convection problem for tilted
post-accidental heat removal in nuclear reactors, cool- rectangular porous material and in other contributions
ing of radioactive waste containers, heat exchangers, by Vasseur et al. [13] and Sen et al. [14]. The literature
solar power collectors, grain storage, food processing, on the natural convection of inclined porous enclosure
energy ecient drying processes, to name of a few. is limited. Besides, only a very small number of studies
were published in the past years considering the
entropy generation in convective heat transfer of en-
E-mail address: [email protected] (A.C. Baytas° ). closure problems.

0017-9310/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 1 7 - 9 3 1 0 ( 9 9 ) 0 0 2 9 1 - 4
2090 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099

Nomenclature

Be Bejan number, Eq. (12) u, v velocity components in x, y directions (m/s)

FFI Fluid Friction Irreversibility, Eq. (9) U, V dimensionless velocity components in X, Y
g acceleration due to gravity (m/s2) directions
HTI Heat Transfer Irreversibility, Eq. (9) x, y Cartesian coordinates (m)
k e€ective thermal conductivity of the porous X, Y dimensionless coordinates
medium (W/m K)
K permeability of the porous medium (m2) Greek symbols
L cavity length (m) a e€ective thermal di€usivity of the porous
N local entropy generation, Eq. (9) medium (m2/s)
NS entropy generation number, Eq. (11) b coecient of thermal expansion (Kÿ1)
Nu local Nusselt number, Eq. (6) Y dimensionless temperature, Eq. (4)
Nua average Nusselt number, Eq. (7) m dynamic viscosity (kg/m s)
Ra Rayleigh number, Eq. (4) n kinematic viscosity (m2/s)
Smgen entropy generation rate per unit volume (W/ s ratio of heat capacity of porous medium to
m3 K) that of ¯uid
t time (s) j inclined angle (deg), Fig. 1
T ¯uid temperature (K) t dimensionless time, Eq. (4)
TC temperature of the cold wall (K) f irreversibility distribution ratio, Eq. (10)
TH temperature of the hot wall (K) c stream function (m2/s)
T0 -…TH ‡ TC †=2 (K) C dimensionless stream function, Eq. (4)
DT temperature di€erence, DT ˆ TH ÿ TC (K)

The utilisation of the second law of thermodynamics angle for di€erent Rayleigh numbers are presented in
in convective heat transfer is very well presented in graphical forms. A numerical study about the entropy
Refs. [15±19]. Drost and White [20] studied numerical generation for natural convection in an inclined porous
calculation of local entropy generation map in an cavity by using the second law of thermodynamics has
impinging jet. San et al. [21] studied entropy gener- not yet been encountered.
ation in convective heat and mass transfer within a
smooth channel under some speci®c thermal boundary
conditions. Cheng et al. [22] presented a numerical 2. Mathematical modelling
study of entropy generation for mixed convection in a
vertical channel with transverse ®n array. Baytas [23] Consider the ¯ow of a Newtonian ¯uid within a
studied the optimisation in an inclined enclosure for square porous enclosure as depicted in Fig. 1. The non-
minimum entropy generation in natural convection for dimensional governing equations are obtained with fol-
the ®rst time. lowing assumptions; in porous cavity, Darcy's law is
This presented that the main subject of the investi- assumed to hold, the ¯uid is assumed to be a normal
gation is not only computational heat transfer of por-
ous cavity but also to investigate entropy generation
distribution according to inclination angle for satu-
rated porous cavity by using the second law of thermo-
dynamics. The present paper reports a numerical
entropy generation calculations of the two-dimensional
laminar free convection ¯ow in an inclined cavity ®lled
with a saturated porous medium. The e€ect of the in-
clination angle …j† on the ¯ow and heat transfer
characteristics and the entropy generation was studied
by varying j from 08 to 3608 and dimensionless Ra
from 102 to 104. The isotherms, the patterns of stream-
lines and their corresponding entropy generation maps,
the variation of entropy generation due to heat trans-
fer and ¯uid friction irreversibility versus inclination Fig. 1. Physical model of the 2D inclined porous cavity.
 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099 2091

Boussinesq-incompressible ¯uid and inertial e€ects are

neglected. The saturated porous medium is assumed to
be isotropic in thermal conductivity. Finally, the set of
non-dimensional governing equations in terms of the
stream function C and temperature Y are
 
@ 2C @ 2C @Y @Y
‡ ˆ Ra cos j ÿ sin j …1†
@X 2 @Y 2 @X @Y

and

@Y @Y @Y @ 2Y @ 2Y
‡U ‡V ˆ ‡ …2†
@t @X @Y @X 2 @Y 2
The velocity equations:

@C @C
Uˆ ; Vˆÿ …3†
@Y @X
where the non-dimensional variables are de®ned by

X, Y ˆ x, y=L; C ˆ c=a; U, V ˆ u, v=…a=L†

Fig. 2. Benchmarking of the predicted local entropy gener-
  ation with analytical solution.
a gKbDTL T ÿ T0
tˆt ; Ra ˆ ; Yˆ …4†
sL2 an TH ÿ TC solid boundaries, cause a continuous entropy gener-
ation in the ¯ow ®eld of porous cavity. This entropy
Eqs. (1) and (2) are subjected to following initial and
generation is due to the irreversible nature of heat
boundary conditions:
transfer and viscosity e€ects, within the ¯uid and at
tR0 for whole space: the solid boundaries. From the known temperature
and velocity ®elds, volumetric entropy generation can
be calculated by the equation [17],
YˆCˆ0
k m ÿ 2

Sm
gen ˆ 2
…rT †2 ‡ U ‡V2 …8†
t > 0: T0 KT0

Dimensionless form of Eq. (8) can be obtained by uti-

c ˆ 0, Y ˆ 0:5 on plane X ˆ 0 lising the dimensionless variable listed in Eq. (4) and
then de®ning the local entropy generation number, N,
c ˆ 0, Y ˆ ÿ0:5 on plane X ˆ 1 for 2D square cavity of Fig. 1 as given below

@Y …rY †2 ‡ f…rC †2
N ˆ |‚‚ …9†
‚{z‚‚‚} |‚‚‚‚{z‚‚‚‚}
c ˆ 0, ˆ 0 on Y ˆ 0, 1 …5† HTI FFI
@Y
The local Nusselt number is de®ned as where f is the irreversibility distribution ratio
" #
@Y mT0 a2
Nu ˆ j …6† fˆ …10†
@ Y Xˆ0, 1 k K…DT †2
And the average Nusselt number is
…1 The local entropy generation number would be inte-
grated over the whole domain to obtain the entropy
Nua ˆ Nu dY …7†
0 generation number for whole cavity volume as
…1 …1
3. Entropy generation due to convection heat transfer NS ˆ N dX dY …11†
0 0

The non-equilibrium conditions due to the exchange In this study, the dimensionless Bejan number (Be )
of energy and momentum, within the ¯uid and at the [16] is used as the alternative irreversibility distribution
2092 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099

Fig. 3. (a) Isotherms, (b) streamlines, (c) entropy generation due to heat transfer and (d) the local entropy generation (N ) at di€er-
ent inclined angles for Ra ˆ 102 :
 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099 2093

Fig. 4. (a) Isotherms, (b) streamlines, (c) entropy generation due to heat transfer and (d) the local entropy generation (N ) at di€er-
ent inclined angles for Ra ˆ 103 :
2094 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099

Fig. 5. (a) Isotherms, (b) streamlines, (c) entropy generation due to heat transfer and (d) the local entropy generation (N ) at di€er-
ent inclined angles for for Ra ˆ 104 :
 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099 2095

number for inclined porous cavity from Eq. (9), as fol- Ra ˆ 50 by MATHEMATICA and developed compu-
lows: ter code are 2.48829 and 2.4880078, respectively.

…rY †2
Be ˆ …12†
…rY †2 ‡…rC †2 5. Result and discussion

The governing equations (1)±(3) were solved numeri-

cally using Finite Di€erence Control Volume Method
of Patankar [25] along with boundary conditions given
4. Benchmark solution for entropy generation in Eq. (5). Resulting algebraic equations were solved
by Alternating Direction Implicit (ADI) method. The
For benchmarking purpose, the analytical solution numerical details of solution method are given in Ref.
for temperature and stream function given in Ref. [24], [23,26,27]. The presented numerical study was checked
Eqs. (18) and (20), was used for natural convection in for accuracy against the earlier published numerical
a porous enclosure. From these analytical solutions, results reported by di€erent authors, and the agree-
local entropy generation map and entropy generation ment between the present and previous results was
number were calculated by utilising MATHEMATICA very good in Ref. [6]. For this reason, it is not
version 3.0. These results of MATHEMATICA were repeated here for brevity.
compared with numerical results from the developed This investigation is mainly concerned about entropy
computer program. This comparison (for Eq. (9)) is generation distribution within an inclined porous
shown in Fig. 2 for Y ˆ 0:5 and Ra ˆ 50: It is clear square cavity by using the second law of thermody-
from Fig. 2 that the numerical and analytical results namics. Results are presented for inclination angle, j
are best matched. Benchmark results from Eq. (11) for (in Fig. 1) from 08 to 3608 and Rayleigh number from

Fig. 6. Variation of entropy generation number (NS) and FFI versus inclined angle.
2096 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099

102 to 104. Benchmark solution for this analysis is Distribution of local entropy generation due to heat
given in Ref. [6]. It was concluded that the present transfer and ¯uid ¯ow, collectively, is shown in Figs.
results match very well with previously published 3±5(d). As is clear from Fig. 3(d), entropy generation
results. In Figs. 3±5, isotherms, streamlines, entropy covers almost whole domain for Ra ˆ 102 , while this
generation due to heat transfer and local entropy gen- covered part of the domain reduces as the Rayleigh
eration number (Eq. (9)) are shown graphically for number increases (Figs. 4 and 5). In case of Ra ˆ 104 ,
Ra ˆ 102 , 103 and 104, respectively. In Fig. 3(c) for the entropy generation is localised along the walls only
j ˆ 08, it is clear that entropy generation is higher at (Fig. 5(d)). It is evident from the above-mentioned
high temperature gradients (Fig. 3(a)). This is due to ®gures that most of the domain is not involved in
heat transfer irreversibility because large heat transfer entropy generation for higher Ra, excluding the case
is con®ned to these locations. As is clear from Fig. 3(c) for j ˆ 908: This is due to the boundary layer regime
for j ˆ 08, entropy generation is mainly con®ned to at high Rayleigh numbers.
the lower and upper corners for the left and right Variation of ¯uid friction irreversibility (FFI) and
walls, respectively. This entropy generation length entropy generation number (NS), for di€erent angular
along the wall increases for j ˆ 408 and 908 for the positions, are shown in Fig. 6. Flow is converted into
same Ra (Fig. 3(c)). After j ˆ 908, the entropy gener- conduction regime for all the Rayleigh numbers from
ation length along the wall reduces gradually for j ˆ 2408 to 3008. This is due to the reason that buoyancy
1308 and 3308. It is evident that entropy generation is is no longer available between these angles. FFI and
directly proportional to temperature gradients. Above NS show a repetition of behaviour from 08 to 908 and
discussion is equally valid for Figs. 4(c) and 5(c). For from 908 to 2008. As shown in Fig. 6, FFI has a mini-
Ra ˆ 103 in Fig. 4(c), the entropy generation covers mum at 908 for Ra ˆ 102 and 103, whereas FFI has a
the whole heated and cooled walls for j ˆ 908; while minimum at 808 for Ra ˆ 104 : As is expected, the
for Ra ˆ 104 (Fig. 5(c)), this is true for j ˆ 408, 908 values of FFI and NS increase with increasing Rayleigh
and 1308. number. Nua and HTI are plotted in Fig. 7 against the

Fig. 7. Variation of entropy generation due to heat transfer (HTI) and average Nusselt number versus inclined angle.
A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099 2097

angular location …j). Again their overall behaviour is

almost similar to that of FFI and NS in Fig. 6, and
also HTI has a minimum at 808 for Ra ˆ 104 , while it
has a minimum at 908 for Ra ˆ 102 and 103.
In Fig. 8, the local Nusselt number variation is
shown graphically for hot wall. In case of angular pos-
itions of 08, 208, 408 and 3308, the local Nusselt num-
ber is highest at the bottom of the cavity while it is
lowest at the top. However, the trend is reversed when
j ˆ 1308 and 2208. This is due to reversal of buoyancy
e€ect. There is a remarkable di€erence for j ˆ 908
from all other values of j for Ra ˆ 103 and 104.
In Fig. 9, the variation of Bejan number (Be )
versus inclination angle …j† is shown as an alterna-
tive irreversibility distribution parameter as described
in Eq. (12). As de®ned in Ref. [16], Be ˆ 1:0 is the
limit at which all the irreversibility is due to heat
transfer, Be ˆ 0 is the opposite limit at which all
the irreversibility is due to ¯uid friction, and Be ˆ
1=2 is the case in which the heat transfer and ¯uid
friction entropy generation rates are equal. Be 
1=2 is the case where the irreversibility due to heat
transfer dominates, while Be  1=2 is the case
where the irreversibility due to ¯uid friction domi-
nates. As seen in Fig. 9, Be ˆ 1:0 at j ˆ 2708 is
the limit at which the heat transfer irreversibility
dominates. As Rayleigh number decreases, heat
transfer irreversibility is dominant around j ˆ 2708:
For high Rayleigh number, ¯uid friction irreversibil-
ity dominates for porous cavity except around j ˆ
2708 as is clear from Fig. 9. As shown in Fig. 9,
Bejan number changes more rapidly when Ra and
j increase after inclined angle j ˆ 1808:
The Bejan number is clearly a measure of the rela-
tive magnitude of the heat transfer and ¯uid friction
irreversibilities. The Bejan number has small values for
the inclination angles between 308 to 608 and 1208 to
1708 in Fig. 9. It shows that the heat transfer and ¯uid
friction contribution to the irreversible losses are not
comparable in these ¯ow cases. It shows evidently that
in these inclination angles, convective heat transfer is
dominated as seen Nua in Fig. 7 and also ¯uid friction
irreversibility is dominated.

6. Conclusions

The distribution of entropy generation in 2D lami-

nar natural convective ¯ows for saturated tilted porous
cavity has been studied numerically by using ADI,

Fig. 8. Local Nusselt number along the hot wall for various
inclined angle; (a) Ra ˆ 102 , (b) Ra ˆ 103 and (c) Ra ˆ 104 :
2098 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099

Fig. 9. Variation of Bejan number (Be ) with inclined angle for Ra ˆ 102 , 103 and 104.

Finite Di€erence Method and the Second Law of Acknowledgements

Thermodynamics. The present solutions for a porous
cavity are compared with the known results from open This paper has been supported by the ITU Research
literature. It was found that these results are in very Fund, through grant 843.
good agreement. A new computer program was devel-
oped to compute the distribution of entropy generation
for present problem. The comparison of numerical and References
analytical results for a simple benchmarking problem
of entropy generation was successful. The in¯uences of [1] D.A. Nield, A. Bejan, Convection in Porous Media,
the physical parameters, Ra, Be, and j are evaluated. Springer, New York, 1998.
Results show that when Ra decreases, heat transfer [2] D.B. Ingham, I. Pop (Eds.), Transport Phenomena in
irreversibility begins to dominate the ¯uid friction irre- Porous Media, Elsevier, Amsterdam, 1998.
versibility. The Bejan number is rapidly changed [3] S.L. Moya, E. Ramos, M. Sen, Numerical study of
between 1508 and 2708. natural convection in a tilted rectangular porous ma-
The entropy generation contains two physical levels: terial., Int. J. Heat Mass Transfer 30 (1987) 741±756.
[4] A. Bejan, On the boundary layer regime in a vertical
at a local level in Figs. 3±5, it shows not only where
enclosure ®lled with a porous medium, Lett. Heat Mass
irreversibilities are present, but also to what extent Transfer 6 (1979) 93±102.
they are sensitive to design changes according to di€er- [5] V. Prasad, F.A. Kulacki, Convective heat transfer in a
ent inclination angle; at a integral level in Figs. 6, 7 rectangular porous cavity e€ect of aspect ratio on ¯ow
and 9, it gives a measure of the ``degree of irreversibil- structure and heat transfer, J. Heat Transfer 106 (1984)
ity'' of the convective ¯ow in the enclosure. 158±165.
 A.C. Baytas° / Int. J. Heat Mass Transfer 43 (2000) 2089±2099 2099

[6] A.C. Baytas, I. Pop, Free convection in oblique en- [18] A. Bejan, A study of entropy generation in fundamental
closures ®lled with a porous medium, Int. J. Heat Mass convective heat transfer, J. of Heat Transfer 110 (1979)
transfer 42 (1999) 1047±1057. 718±725.
[7] C. Beckermann, R. Viskanta, S. Ramadhyani, A nu- [19] D. Poulikakos, A. Bejan, Fin geometry for minimum
merical study of non-Darcian natural convection in a entropy generation in forced convection, J. of Heat
vertical enclosure ®lled with a porous medium, Num. Transfer 104 (1982) 616±623.
Heat Transfer 10 (1986) 557±570. [20] M.K. Drost, M.D. White, Numerical predictions of
[8] R.J. Gross, M.R. Bear, C.E. Hickox, The application of local entropy generation in an impinging jet, J. of Heat
¯ux-corrected transport (FCT) to high Rayleigh number Transfer 113 (1991) 823±829.
natural convection in a porous medium, in: Proc. 8th [21] J.Y. San, W.M. Worek, Z. Lavan, Entropy generation
Int. Heat Transfer Conf., San Francisco, CA, 1986. in convective heat transfer and isothermal convective
[9] F.C. Lai, F.A. Kulacki, Natural convection across a mass transfer, J. of Heat Transfer 109 (1987) 647±652.
vertical layered porous cavity, Int. J. Heat Mass [22] C.H. Cheng, W.P. Ma, W.H. Huang, Numerical predic-
Transfer 31 (1988) 1247±1260.
tions of entropy generation for mixed convective ¯ows
[10] D.M. Manole, J.L. Lage, Numerical Benchmark results
in a vertical channel with transverse ®n arrays, Int. J.
for natural convection in a porous medium cavity, in:
Heat Mass Transfer 21 (1994) 519±530.
ASME Conference, Heat and Mass Transfer in Porous
[23] A.C. Baytas, Optimisation in an inclined enclosure for
Media, HTD-Vol. 216, 1992, 55±60.
minimum entropy generation in natural convection, J.
[11] K.L. Walker, G.M. Homsy, Convection in a porous
Non-Equil. Thermodyn 22 (1997) 145±155.
cavity, J. Fluid Mech 87 (1978) 449±474.
[24] M. Haajizadeh, A.F. Ozguc, C.L. Tien, Natural convec-
[12] J.-P. Caltagirone, S. Bories, Solutions and stability cri-
teria of natural convective ¯ow in an inclined porous tion in a vertical porous enclosure with internal heat
layer, J. Fluid Mech. 155 (1985) 267±287. generation, Int. J. Heat Mass Transfer 27 (1984) 1893±
[13] P. Vasseur, M.G. Satish, L. Robillard, Natural convec- 1902.
tion in a thin inclined porous layer exposed to a con- [25] V.S. Patankar, Numerical Heat Transfer and Fluid
stant heat ¯ux, Int. J. Heat Mass Transfer 30 (1987) Flow, Hemisphere, New York, 1980.
537±549. [26] A.C. Baytas, Buoyancy-driven ¯ow in an enclosure con-
[14] M. Sen, P. Vasseur, L. Robillard, Multiple steady states taining time periodic internal sources, Heat and Mass
for unicellular natural convection in an inclined porous Transfer 31 (1996) 113±119.
layer, Int. J. Heat Mass Transfer 30 (1987) 2097±2113. [27] A.C. Baytas, Transient natural convection and orien-
[15] A. Bejan, Entropy Generation Through Heat and Fluid tation optimisation in a di€erentially heated inclined
Flow, 2nd ed., Wiley, New York, 1994. square enclosure with internal heat sources, in: J. Padet,
[16] A. Bejan, Entropy Generation Minimisation, CRC F. Arinc (Eds.), ICHMT, International Symposium on
Press, Boca Raton, FL, 1996. Transient Convective Heat Transfer, Aug. 19±23 (1996),
[17] A. Bejan, Convection Heat Transfer, 2nd ed., Wiley, Cesme, Turkey, Begel House, New York, 1997, pp. 117±
New York, 1995. 125.