Home Search Collections Journals About Contact us My IOPscience

Analytical–numerical analysis of entropy generation for gravity-driven inclined channel flow

with initial transition and entrance effects

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2008 Phys. Scr. 78 045401

(http://iopscience.iop.org/1402-4896/78/4/045401)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 131.104.62.10
This content was downloaded on 11/11/2014 at 19:55

Please note that terms and conditions apply.

IOP PUBLISHING PHYSICA SCRIPTA
Phys. Scr. 78 (2008) 045401 (7pp) doi:10.1088/0031-8949/78/04/045401

Analytical–numerical analysis of entropy

generation for gravity-driven inclined
channel flow with initial transition
and entrance effects
M Havzali1 , A Arikoglu1 , G Komurgoz2 , H I Keser1 and I Ozkol1,3
1
Istanbul Technical University, Faculty of Aeronautics and Astronautics,
Aeronautical Engineering Department, Maslak, TR-34469, Istanbul, Turkey
2
Istanbul Technical University, Faculty of Electrical and Electronic Engineering,
Electrical Engineering Department, Maslak, TR-34469, Istanbul, Turkey
E-mail: ozkol@itu.edu.tr

Received 7 February 2008

Accepted for publication 6 August 2008
Published 30 September 2008
Online at stacks.iop.org/PhysScr/78/045401

Abstract
In this work, the entropy generation due to the flow of a gravity-driven laminar viscous
incompressible fluid through an inclined channel is investigated. Fully developed flow field is
solved for a Newtonian fluid. Then, temperature field is represented in a purely analytical
expression and subject to isothermal boundary conditions on the walls and constant
rectangular temperature profile at the inlet. This analytical solution is not similar to the already
existing ones in the open literature. Also, the temperature field is numerically resolved by
using the method of lines with the same inlet and boundary conditions. These two solutions
overlap, which indicates the correctness of both solutions. In obtaining both analytical and
numerical solutions, no assumption is made on the initial transition and entrance region. It is
shown that the effect of this region, which has been omitted in previous studies, is highly
dominant on the overall entropy generation. Therefore, the detailed thermal analysis of the
entrance section is outlined.

PACS numbers: 44.05.+e, 44.15.+a

(Some figures in this article are in colour only in the electronic version.)

Nomenclature Ns Total entropy generation number

Nh Entropy generation number due to
Be Bejan number heat transfer
Br Brinkmann number (= Ec × Pr ) Pe Peclet number (= Re × Pr = u m h/α)
cp Specific heat J kg−1 K−1 Pr Prandtl number (= µcp /k)
Ec Eckert number (= u 2m cp /1T ) Re Reynolds number (= u m h/ν)
g Acceleration due to gravity m s−2 SG Entropy generation rate for unit volume W m−3 K−1
h Half width of channel m T Temperature K
k Thermal conductivity W m−1 K−1 T̄ Dimensionless temperature
m Summation index (= 1, 2, . . .) T0 Reference temperature K
Nf Entropy generation number due to Tw Wall temperature of channel K
fluid friction u x component of velocity m s−1
ū x component of dimensionless velocity
3 Author to whom any correspondence should be addressed. um Reference velocity m s−1

0031-8949/08/045401+07$30.00 1 © 2008 The Royal Swedish Academy of Sciences Printed in the UK

Phys. Scr. 78 (2008) 045401 M Havzali et al

x Axial distance m parallel plates has been studied by many researchers [4–10].
x̄ Dimensionless axial distance Analysis of such simple geometries is useful to understand
y Normal distance m the mechanism of heat transfer, fluid flow and their complex
ȳ Dimensionless normal distance interactions for thermodynamic irreversibilities.
With regard to irreversibilities and entropy generation
Greek letters mechanisms, the following studies can be found in the
open literature. A short review of them is, as follows:
α Thermal diffusivity m2 s−1 Saouli and Aiboud-Saouli [11] investigated the heat transfer
θ Inclination angle of channel rad of a laminar falling liquid film along an inclined heated
µ Dynamic viscosity Pa s−1 plate for a Newtonian fluid via second law analysis. They
ρ Density of the fluid kg m−3 considered the upper surface of the liquid film to be free
3 Degenerate hypergeometric function and the lower wall to be fixed with constant heat flux.
8 Irreversibility number In another study, Makinde examined the application of the
 Dimensionless temperature difference second law of thermodynamics to the laminar flow of an
incompressible viscous fluid through an inclined channel
1. Introduction with isothermal walls [2]. In their study, based on some
simplifying assumptions and using separation of variables,
For more than three decades, an increasing awareness analytical solutions for the fluid velocity and temperature
has been emerging that the world’s energy resources are were constructed.
limited, which has caused the political climate in almost all For the horizontal channel case, geometry-related
countries to re-examine their energy policies. Governments boundary and symmetry conditions result in having, in most
have felt the need to take drastic measures to eliminate cases, an exact analytical solution. Inclined channel-type flow
waste. The scientific community has started taking a closer field equations also, although a little more difficult than the
look at energy conversion devices and is interested in horizontal channel, have exact solutions. These types of flow
developing new techniques and analysis methods to better fields find wide applications in engineering, particularly in
utilize the existing limited resources. Therefore all the heating and cooling applications. Some analytical studies
energy-producing, -converting and -consuming systems must based on separation of variables with simplifying assumptions
be re-examined carefully and all possible available-work are found in the literature [1]. In these studies, the temperature
destruction mechanisms removed. On the theoretical side, field is solved by simplifying the separation variable approach
this can be done only by utilizing the second law of as a summation of two separate functions. Therefore the
thermodynamics, which is related to entropy generation. solution obtained for the partial differential governing
Efficiency calculation of heat exchange systems has been equation is not a general solution but a special solution.
very much restricted to the first law of thermodynamics. In the present study, flow-field governing equations
Calculations using the second law of thermodynamics, which for a gravity-driven, laminar, viscous, incompressible fluid
is related to entropy generation and efficiency calculation, through an inclined channel are reduced to a simple ordinary
are more reliable than the first-law-based calculations. In differential equation (ODE). Despite many studies of flow
almost all thermal systems, second-law-based efficiency can and temperature field for various types of inclined-channel
be defined in terms of the ratio of the actual thermal efficiency problems [1], none of them have considered the effect of
to reversible thermal efficiency under the same conditions. the initial transition and entrance. These effects are highly
Therefore, the second law of thermodynamics can be applied dominant on the overall entropy generation, which was
to investigate the irreversibilities in terms of the entropy omitted in previous studies. In contrast, in this study, first
generation rate. The determination of entropy generation is the temperature field is represented in a purely analytical
also important for upgrading the system performance, because expression, which is also checked against the governing
the entropy generation is the measure of the destruction of equation with the inlet and the boundary conditions. This
the available work of the system [1]. As entropy generation analytical expression [12] is not similar to the already existing
takes place, the quality of energy decreases. It is important ones, since no assumption on solution function was made.
to study the distribution of the entropy generation within Then, with the same inlet and boundary conditions the
the fluid volume for preserving the quality of energy in temperature field is numerically resolved by using the method
fluid flow processes or reducing the entropy generation. of lines (MOL) without any assumptions. It is observed that
Therefore calculation of the correct entropy generation is these two solutions give the same results, which indicates
of great importance in many engineering fields such as the correctness of both solutions. Therefore the analytical
heat exchangers, cooling of nuclear reactors, energy storage expression obtained can be considered as a general solution.
systems, cooling of electronic devices, etc [2]. The effects of initial transition and entrance on entropy
The entropy generation method as a measure of system generation are presented in terms of entropy generation
performance was first introduced by Bejan in 1980 [3]. number (NS ). Bejan number (Be) is derived and plotted using
Since then, many studies have been published on the second dimensionless velocity and temperature profiles.
law of thermodynamics, entropy generation rates and the For the sake of the reader, the development of heat
irreversibility of basic arrangements. One such essential basic boundary at the entrance is graphically presented. It is found
arrangement is the channel-type geometry flow. The fluid flow that the entrance section has great dominance over the entire
inside a channel with a circular cross section or made of two region.

2
Phys. Scr. 78 (2008) 045401 M Havzali et al

and the heat conducted within the fluid and u m is twice

the maximum value of the axial velocity. By using the
dimensionless parameters in equation (5), the analytical
solution in equation (4) is expressed as follows.
ū( ȳ) = 12 1 − ȳ 2 .


(7)
The dimensionless governing equation for the temperature
becomes
∂ T̄ ∂ 2 T̄
ū = , (8)
∂ x̄ ∂ ȳ 2
and its related boundary conditions with the dimensionless
variables are expressed as follows:
dū
ū(±1) = 0, (0) = 0,
d ȳ (9)
Figure 1. The coordinate system and the geometry of the channel. T̄ (0, ȳ) = 0, T̄ (x̄, ±1) = 1.

2. Equation of motion
3. The solution
The gravity-driven, laminar, steady and incompressible flow
of a Newtonian fluid inside an inclined channel is considered. 3.1. Analytical solution
The flow field is assumed to be one-dimensional and isobaric.
The geometry of the problem is presented in figure 1. The The solution for equation (8), together with the boundary
parallel plates located at y = ±h are isothermal and there is conditions in equation (9), is introduced by Polyanin [12]
‘no slip condition’ between the plates and the fluid. as follows:
The governing equations, which are the momentum and
T¯ (x̄, ȳ) = 1 − 1.2 exp −2(1.68)2 x̄ exp − 12 (1.68) ȳ 2



the energy equations, are given in equations (1) and (2):
×3 41 − 14 (1.68), 12 ; (1.68) ȳ 2


∂ u
 2  "∞
µ + ρg sin(θ) = 0, where 0 < θ < π/2, (1) X
∂ y2 − (−1)m 2.27(4m + 1.68)−7/6
m=1
∂T k ∂ T 2
× exp −2(4m + 1.68)2 x̄ exp − 21 (4m + 1.68) ȳ 2



u = . (2)
∂x ρcp ∂ y 2 #
× 3 4 − 4 (4m + 1.68), 2 ; (4m + 1.68) ȳ ,
1 1 1 2


The boundary conditions for the axial velocity and
temperature are as follows:
(10)
du
u(±h) = 0, (0) = 0, where
dy (3)
T (0, y) = 0, T (x, ±h) = Tw . X∞
β(β + 1) · · · (β + m − 1) z m
3(β, χ; z) = 1 +
The solution of equation (1) with the boundary conditions in m=1
χ (χ + 1) · · · (χ + m − 1) m!
equation (3) leads to the following analytical expression for
is the degenerate hypergeometric function. This solution is
the velocity:
also encountered in mass transfer applications [12].
1
u(y) = (ρg sin θ)(h 2 − y 2 ). (4)
2µ 3.2. Numerical solution with MOL

The following dimensionless parameters for the coordinate The MOL is a commonly used technique for solving
system and the system variables are introduced: partial differential equations (PDEs). In this technique, finite
difference relations are used for the spatial derivatives, and the
u y PDE considered transforms to a system of ODEs. By applying
ū = , ȳ = ,
um h the central difference to the ȳ coordinate of equation (8),
(5)
x T − T0 we have:
x̄ = , T̄ = ,
h Pe Tw − T0
dT̄i (x̄) 2 T̄i+1 (x̄) − 2T̄i (x̄) + T̄i−1 (x̄)
= ,
where dx̄ 1 − [(i − 1)1 − 1]2 12
ρgh 2 sin θ umh
um = , Pe = . (6) for 2 6 i 6 n, (11)
µ α
In equation (6), Pe = ρgh 3 sin θ/µα is the Peclet number, where n is the number of sections the ȳ coordinate is divided
which is the ratio of the heat transferred to the fluid into and 1 = 2/n is the length of a section. Therefore there

3
Phys. Scr. 78 (2008) 045401 M Havzali et al

Table 1. Numerical results for T¯ (x̄, ȳ) (the analytical solution).

ȳ = 0.0 0.1 0.2 0.4 0.6 0.8 1.0
x̄ = 0.0 0.000 511 0.000 756 0.000 794 −0.001 06 −0.002 50 −0.003 05 1.031 820
0.1 0.318 085 0.327 602 0.355 721 0.461 987 0.619 132 0.804 491 0.998 928
0.2 0.611 957 0.617 411 0.633 514 0.694 226 0.783 712 0.889 016 0.999 387
0.4 0.874 519 0.876 282 0.881 489 0.901 122 0.930 059 0.964 111 0.999 802
0.6 0.959 423 0.959 994 0.961 677 0.968 026 0.977 383 0.988 395 0.999 936
0.8 0.986 879 0.987 063 0.987 608 0.989 661 0.992 686 0.996 247 0.999 979
1.0 0.995 757 0.995 817 0.995 993 0.996 657 0.997 635 0.998 786 0.999 993

Table 2. Numerical results for T¯ (x̄, ȳ) (the numerical solution).

ȳ = 0.0 0.1 0.2 0.4 0.6 0.8 1.0
x̄ = 0.0 0 0 0 0 0 0 1.000 00
0.1 0.317 970 0.327 513 0.355 694 0.462 206 0.619 690 0.805 367 1.000 00
0.2 0.612 078 0.617 548 0.633 686 0.694 529 0.784 170 0.889 591 1.000 00
0.4 0.874 692 0.876 459 0.881 672 0.901 326 0.930 282 0.964 336 1.000 00
0.6 0.959 523 0.960 094 0.961 778 0.968 126 0.977 480 0.988 480 1.000 00
0.8 0.986 925 0.987 109 0.987 653 0.989 704 0.992 725 0.996 279 1.000 00
1.0 0.995 776 0.995 836 0.996 012 0.996 674 0.997 650 0.998 798 1.000 00

are n + 1 node points. By using the initial boundary conditions from the walls;  = (Tw − T0 )/T0 is the dimensionless
in equation (9) for the temperature, the following relations are temperature difference and Ns = SG h 2 T02 /k(Tw − T0 )2 is the
obtained. dimensionless local entropy generation rate. The ratio Br / 
is called the group parameter, which shows the relative
T̄i (0) = 0, for 2 6 i 6 n,
(12) importance of viscous effects on heat transfer irreversibility.
T̄1 (x̄) = T̄n+1 (x̄) = 1. The entropy generation given in equation (13) can be written
as follows:
The initial value problem for the system of ODEs presented in Ns = Nh + Nf , (16)
equations (11) together with the conditions in (12) is solved by
using the Runge–Kutta method. In the calculations performed where Nh is the entropy generation due to heat transfer
the number of sections used (n) is 30. irreversibility and Nf is the entropy generation due to
fluid friction irreversibility. Because the entropy generation
4. Entropy generation originates from both heat transfer and viscous friction, these
effects should be separately investigated to see the dominance
The equation of entropy generation for the two-dimensional of one over the other. The Bejan number is the ratio of heat
flow of a Newtonian, incompressible fluid, which obeys transfer irreversibility to the total irreversibility generated by
Fourier law, is as follows [13, 14]: the system and is defined as follows [13]:
" #
∂T 2 ∂T 2 Nh
 
k
SG = 2 + Be = . (17)
T0 ∂x ∂y Ns
" (    )  #
µ ∂u 2 ∂v 2 ∂u ∂v 2 Heat transfer irreversibility dominates for Be → 1 and fluid
+ 2 + + + , (13) friction dominates for Be → 0. Heat transfer and fluid friction
T0 ∂x ∂y ∂y ∂x
contributions to the irreversibility generated in the system are
where, due to the assumptions made for the characteristics of equal when Be = 1/2.
the flow, ∂u/∂ x = 0, ∂v/∂ y = 0 and ∂v/∂ x = 0. Therefore,
equation (13) reduces to the following form [15]:
5. Results and discussion
" #
∂T 2 ∂T 2 µ ∂u 2
   
k
SG = 2 + + . (14) As mentioned earlier, equation (8) is solved numerically with
T0 ∂x ∂y T0 ∂ y MOL to validate the analytical solution given in equation (10).
The numerical results for the analytical and numerical
The equation of local entropy generation can be
solutions are presented in tables 1 and 2, respectively.
nondimensionalized by using equation (5) as follows:
Comparision of the results presented in tables 1 and 2
2  2 reveals that both solutions overlap at least up to three digits.
∂ T̄ ∂ T̄ Br ∂ ū 2
  
1
Ns = + + , (15) The results show good agreement between the numerical and
Pe2 ∂ x̄ ∂ ȳ  ∂ ȳ
the analytical solutions, which proves the correctness of both.
where Br = µu 2m /k(Tw − T0 ) is the Brinkmann number, the For better agreement between these two solutions, the order
ratio of the heat generation due to velocity differences of the numerical scheme or the number of sections (n) can be
in the viscous fluid to the heat conducted to the fluid increased.

4
Phys. Scr. 78 (2008) 045401 M Havzali et al

Figure 4. Entropy generation profiles at various x stations for

Figure 2. Temperature profiles at various x̄ stations.
Pe = 7.1 and Br −1 = 0.4.

Figure 3. The contour plot of temperature inside the channel.

5.1. Fully developed region for the thermal field

Temperature profiles are obtained by using equation (10) and
these distributions are plotted in figure 2 at different stations
of x̄ for Pe = 7.1, Br −1 = 0.4. It can be seen from this Figure 5. Entropy generation for different values of the group
that temperature profiles are symmetrical along the ȳ = 0 axis, parameter Br −1 for Pe = 7.1 and x̄ = 0.1.
and the magnitude of temperature increases with increasing x̄
as expected. For large values of x̄, the temperature From the figures, it can be easily deduced that the wall
difference between the channel and the fluid disappears, region acts as a powerful generator of irreversibility. To clarify
namely T̄ → 1. and visualize which mechanism is dominant in the entropy
The contour plot of temperature field is given in figure 3 generation, the Bejan number is obtained from equation (17)
for visualizing the development through the channel. and plotted in figure 6 at different x̄ stations. Figure 6 shows
Using the analytical expressions for the thermal and that the irreversibility caused by viscous friction between
velocity fields in equations (7) and (10), the rate of local the walls and the fluid dominates at the walls, especially
entropy generation can be obtained from equation (13). for high values of dimensionless axial distance; on the other
Local entropy generation number NS is graphically presented hand, the heat transfer irreversibility dominates towards the
in figure 4 for Br −1 = 0.4 and Pe = 7.1 at different x̄ centre of the channel. In the vicinity of the centerline of
locations. This figure shows that entropy generation decreases the channel, namely ȳ = 0, the Bejan number is equal to 1,
sharply towards ȳ = 0 since both thermal and velocity since the only contribution at this point comes from the
gradients are zero at this point due to the symmetry of the gradient of temperature in the x̄ direction. The gradients of
problem. Entropy generation becomes significant in the region temperature and velocity in the ȳ direction are zero at this
close to ȳ = ±1 since the wall has a strong effect on entropy point due to the symmetry. Bejan number increases with
generation because of the sudden change of temperature. decreasing x̄, which shows that heat transfer irreversibility is
Entropy generation rate rapidly decreases with x̄, which dominant in the entrance region.
shows the dominance of the entrance region on the overall The effect of the group parameter on Bejan number
entropy generation. For increasing values of x̄, the entropy is presented in figure 7 for Pe = 7.1 at x̄ = 0.1 location.
generation profiles converge to a single curve. Increasing the group parameter causes a decrease in Bejan
The entropy generation rate is presented in figure 5 for number since the portion of entropy generation caused by the
different values of the group parameter Br −1 for Pe = 7.1 fluid friction irreversibility also increases as expected.
and x̄ = 0.1. Figure 5 shows that the group parameter has a Though there is no slip on the walls, stronger velocity
strong effect on entropy generation. The increase in Br −1 gradients occur close to the walls. According to the boundary
increases the entropy generation rate as a consequence of layer theory, these strong gradients exist in the boundary
the fact that the magnitude of viscous dissipation is directly layer, which has a small thickness compared with the channel
related to this parameter. height. Also, in this region, the temperature field creates

5
Phys. Scr. 78 (2008) 045401 M Havzali et al

Figure 9. Development of heat boundary at the entrance.

Figure 6. Bejan number as a function of ȳ at different x̄ stations,

Pe = 7.1 and Br −1 = 0.4.

Figure 10. Entropy generation at the entrance section for Pe = 7.1

and Br −1 = 0.4.

Figure 7. Bejan number at different values of the group parameter

for Pe = 7.1 and x̄ = 0.1.

Figure 11. The contour plot of entropy generation.

The rectangular temperature profile at the entrance

develops through the channel with increasing x̄ and finally
takes a parabolic shape at a value of about x̄ = 0.04. From
figure 8 it can be deduced that the developed thermal boundary
layers on both upper and lower walls coincide at a value
of 0.04 > x̄ > 0.02. This trend in the development of the
Figure 8. Dimensionless temperature profiles in various x̄ stations. temperature profile is also graphically presented in figure 9.
The entropy generation is derived and plotted in figure 10
for the entrance region. The common trend is an increase
a thermal boundary layer, till the layer’s fully developed
from the centerline to a maximum value, then a continuous
temperature field has nonzero gradients in both the parallel
decrease towards the wall surface. The maximum entropy is
and perpendicular directions to the walls. Therefore this
not generated on the walls; instead, it is generated at a location
region is a powerful entropy generator and is dominant on
of ȳ close to the edge of the thermal boundary layers. This is
total entropy generation.
quite natural since the temperature gradients are very high in
this region.
5.2. The region of initial transition and entrance
The rate of entropy production of this section is
Dimensionless temperature profiles are obtained in figure 8 graphically presented as a contour plot in figure 11. As one
for investigating the effects of initial transition and entrance can realize, the entrance section has great dominance over the
on entropy generation. entire region.

6
Phys. Scr. 78 (2008) 045401 M Havzali et al

6. Conclusion References

In this work, analytical and numerical studies are performed [1] Makinde O D and Gbolagade A W 2005 Rom. J. Phys.
to investigate the effect of entropy generation on a laminar, 50 923–30
[2] Makinde O D 2006 Phys. Scr. 74 642–5
viscous, incompressible flow between two inclined, parallel, [3] Bejan A 1980 Energy 5 721–32
isothermal plates. The equation of entropy generation is [4] Mahmud S and Fraser R A 2002 Exergy 2 140–6
derived and entropy generation number and Bejan number are [5] Nag P K and Kumar N 1989 Int. J. Energy Res.
plotted for different values of the group parameter as well as 13 537–43
for different dimensionless axial distances. It is observed that [6] Narusawa U 2001 Heat Mass Transfer 37 197–203
[7] Sahin A Z 1998 Energy 23 465–73
the dominance of the fluid friction is higher in the vicinity [8] Sahin A Z 2000 Int. J. Heat Mass Transfer 43 1469–78
of the walls for x > 0.2, namely the fully developed region [9] Syeda H T and Shohel M 2002 Int. Commun. Heat Mass
for thermal boundary layers. It is seen that as x̄ increases, the Transfer 29 907–18
entropy generation rapidly decreases along the channel. [10] Mahmud S, Tasnim S H and Mamun S A A 2003 Int. J. Therm.
The main feature of this study that distinguishes it from Sci. 42 731–40
[11] Saouli S and Aiboud-Saouli S 2004 Int. Commun. Heat Mass
similar studies carried out earlier is the account of initial Transfer 31 879–86
transition and entrance effects. The entropy generated in such [12] Polyanin A D 2001 Handbook of Linear Partial Differential
a small section is dominant on the total entropy produced Equations for Engineers and Scientists (Boca Raton, FL:
in the entire system. Therefore, exclusion of this section in Chapman and Hall)
entropy generation calculation results in a serious error [13] Bejan A 1996 Entropy Generation Minimization (New York:
CRC Press)
in determining the correct efficiency of the system. This [14] Ibrahim O, Komurgoz G and Arikoglu 2007 Proc. Inst. Mech.
study might encourage further research on entropy generation Eng. A: J. Power Energy 221 609–16
minimization for the inclined channel problem, including [15] Mahmud S and Fraser R A 2003 Int. J. Therm. Sci.
entrance effects. 42 177–86