(This is a sample cover image for this issue. The actual cover is not yet available at this time.

)
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Numerical analysis of laminar forced convection recess ow with two inclined
steps considering gas radiation effect
Meysam Atashafrooz, Seyyed Abdolreza Gandjalikhan Nassab

Mechanical Engineering Department, School of Engineering, Shahid Bahonar University, Kerman, Iran
a r t i c l e i n f o
Article history:
Received 29 September 2011
Received in revised form 18 April 2012
Accepted 19 June 2012
Available online 28 June 2012
Keywords:
Laminar convection ow
Recess
Radiation
Blocked-off method
DOM
a b s t r a c t
In this work, a numerical analysis of combined convectionradiation heat transfer over a recess including
two inclined backward and forward facing steps in a horizontal duct is presented. The uid is treated as a
gray, absorbing, emitting and scattering medium. Discretized forms of the continuity, momentum and
energy equations are obtained by the nite volume method and solved using the SIMPLE algorithm, while
the blocked off method is employed in simulation of the inclined surfaces. The radiative transfer equation
(RTE) is solved numerically by the discrete ordinates method (DOM). Numerical results reveal that the
radiationconduction parameter, optical thickness, step inclination angle and recess length have consid-
erable effects on heat transfer behavior of the recess ow. Comparison of numerical results with the
available data published in open literature shows a good consistency.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
Forced convection ow over a backward or forward facing step
in a channel is widely encountered in engineering applications. In
these types of convection ow, separating and reattaching regions
exist because of the sudden changes in ow geometry. Separation
ows accompanied with heat transfer are frequently encountered
in many systems, such as cooling of electronic systems, power gen-
erating equipments, gas turbine blades, heat exchangers, combus-
tion chamber and ducts ows used in industrial applications. In
some of the mentioned devices, specially, when soot particles exist
in the combustion product, the radiation effect may be important.
Besides, the trend toward increasing temperature in modern tech-
nological systems has promoted concerted effort to develop more
comprehensive and accurate theoretical methods to treat radia-
tion. Therefore, for having more accurate and reliable results in
the analysis of these types of ow, the owing uid must be con-
sidered as a radiating medium and all of the heat transfer mecha-
nisms including convection, conduction and radiation, must be
taken into account. The ow over backward facing step (BFS) or
forward facing step (FFS) has the most features of separated ows.
Although the geometry of BFS or FFS ow is very simple, the heat
transfer and uid ow over these types of step contain most of
complexities. Consequently, it has been used in the benchmark
investigations. There are many studies about laminar convection
ow over BFS in a duct by several investigators [14].
Kondoh et al. [5] studied laminar heat transfer in a separating
and reattaching ow, numerically, by simulating the ow and heat
transfer downstream of a backward-facing step. The effects of
channel expansion ratio, Reynolds number and Prandtl number
on heat transfer behavior were investigated. Erturk [6] investigated
the characteristics of ow over a two dimensional BFS in a wide
range of Reynolds numbers. The two-dimensional NavierStokes
equations for incompressible steady ows were solved with a very
efcient nite difference numerical method which proved to be
highly stable even at very high Reynolds numbers. Abu-Nada
[79] analyzed the convection ow over a backward facing step
in a duct to investigate the amount of entropy generation in this
type of ow. In those works, the set of governing equations were
solved by the nite volume method and the distributions of entro-
py generation number, friction coefcient and Nusselt number on
the duct walls were calculated. Moreover, the effect of suction
and blowing on the entropy generation number and Bejan number
were presented.
Although there are many research studies about BFS geome-
tries, but the uid ow with heat transfer over FFS received less
attention. In a recent study, Bahrami and Gandjalikhan Nassab
[10] analyzed the convection ow over FFS in a duct to investigate
the amount of entropy generation in this type of ow. A review of
research on laminar convection ow over backward and forward
facing steps was done by Abu-Mulaweh [11]. In that study, a com-
prehensive review of such ows, those have been reported in sev-
eral studies in the open literature was presented. The purpose was
to give a detailed summery of the effect of several parameters such
as step height, Reynolds number, Prandtl number and the
0045-7930/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compuid.2012.06.020

Corresponding author.
E-mail address: ganjali2000@yahoo.com (S.A. Gandjalikhan Nassab).
Computers & Fluids 66 (2012) 167176
Contents lists available at SciVerse ScienceDirect
Computers & Fluids
j our nal homepage: www. el sevi er. com/ l ocat e/ compui d
Author's personal copy
buoyancy force on the ow and temperature distributions down-
stream of the step. Also at different points of view, this geometry
(FFS) was studied by several researchers [12,13].
There are many engineering applications, in which the forward
or backward-facing step is inclined. Simulations of three-dimen-
sional laminar forced convection adjacent to inclined backward-
facing step in rectangular duct were presented by Chen et al.
[14] to examine the effects of step inclination on ow and heat
transfer distributions. Velocity, temperature, Nusselt number and
friction coefcient distributions were presented in that study.
The effects of step inclination angle on Nusselt number and friction
coefcient distributions were showed by plotting many gures. In
a recent study, Gandjalikhan Nassab et al. [15] studied the turbu-
lent forced convection ow adjacent to inclined forward facing
step in a duct. In that study, the NavierStokes and energy equa-
tions were solved in the computational domain by CFD method
using conformal mapping technique and the effects of step inclina-
tion angle on ow and temperature distributions were determined.
In all of the mentioned studies, the effect of radiative heat trans-
fer in uid ow was not studied, such that the gas energy equation
only contains the convection and conduction terms. In a forced
convection problem, when the owing gas behaves as a participat-
ing medium, its complex absorption, emission and scattering intro-
duce a considerable difculty in the simulation of these ows.
There are limited numbers of literatures about the radiative trans-
fer problems in convection ows with complex 2-D and 3-D
geometries.
Bouali and Mezrhab [16] studied heat transfer by laminar
forced convection with considering surface radiation in a divided
vertical channel with isotherm side walls. They found that the sur-
face radiation has important effect on the Nusselt number in con-
vective ow with high Reynolds numbers. Azad and Modest [17]
investigated the problem of combined radiation and turbulent
forced convection in absorbing, emitting and linearly anisotropic
scattering gas particulate ow through a circular tube.
Two-dimensional forced convection laminar ow of radiating
gas over a BFS in a duct was analyzed by Ansari and Gandjalikhan
Nassab [18]. Effects of wall emissivity, Reynolds number and its
interaction with the conductionradiation parameter on heat
transfer behavior of the system were investigated. Also, the same
authors [19,20] studied the laminar forced convection ow of a
radiating gas adjacent to two inclined backward and forward fac-
ing steps in a duct. The two-dimensional Cartesian coordinate sys-
tem was used to simulate the inclined surfaces by considering the
Blocked-off region in regular grid. The uid was treated as a gray,
absorbing, emitting and scattering medium. The effect of radiative
properties on heat transfer behavior of uid ow was investigated.
The study of mixed convection heat transfer in 3-D horizontal
and inclined ducts with considering gas radiation effects has been
numerically examined in detail by Chiu et al. [21,22]. Those works
were primarily focused on the interaction of the thermal radiation
with mixed convection for a gray uid in rectangular ducts. The
vorticityvelocity method was employed to solve the three-dimen-
sional NavierStokes equations while the integro-differential radi-
ative transfer equation was solved by the discrete ordinates
method. Results revealed that radiation effects have a considerable
impact on the heat transfer and would reduce the thermal buoy-
ancy effects. Besides, it was revealed that the development of tem-
perature was accelerated by the radiation effects.
Although there are limited studies about laminar forced convec-
tion ow of radiating gas over a BFS and FFS, but base on the
authors knowledge, combined convection and radiation heat
transfer over two inclined backward and forward facing steps,
which provides a recess, is still not studied theoretically or exper-
imentally. Since, this ow geometry has many engineering applica-
tions, therefore, the present research work deals the 2-D analysis of
an incompressible laminar forced convection ow of a radiating
gas over a recess in a horizontal duct, while the well known
block-off method and DOM are employed to solve the uid me-
chanic and radiation problems.
Nomenclature
A
x
, A
y
areas of control volumes faces normal to the x- and y-
directions, respectively (m
2
)
C
p
specic heat (J kg
1
K
1
)
CR contraction ratio
D
h
hydraulic diameter (m)
ER expansion ratio
I radiation intensity (W m
2
)
I

dimensionless radiation intensity

Nu
t
total Nusselt number
Nu
r
radiative Nusselt number
Nu
c
convective Nusselt number
q
t
total heat ux (W m
2
)
q
r
radiative heat ux (W m
2
)
q
c
convective heat ux (W m
2
)
Pe Peclet number
Pr Prandtl number
Re Reynolds number
RC radiationconduction parameter
S radiation source function
S

dimensionless radiation source function

T temperature (K)
U
o
average velocity of the incoming ow at the inlet sec-
tion (m/s)
x, y horizontal and vertical distance, respectively (m)
X, Y dimensionless horizontal and vertical coordinate,
respectively
x
r
reattachment length (m)
Greek symbols
a thermal diffusivity (m
2
s
1
)
r Stefan Boatsmans constant = 5.67 10
8
, (W m
2
K
4
)
r
a
absorbing coefcient (m
1
)
r
s
scattering coefcient (m
1
)
e wall emissivity
/ step inclination angle
j thermal conductivity (W m
1
K
1
)
l dynamic viscosity (N s/m
2
)
m kinematic viscosity (m
2
/s)
q density (kg/m
3
)
s optical thickness
H dimensionless temperature
H
b
mean bulk temperature
h
1
, h
2
dimensionless temperature parameters
Subscripts
c convective
in inlet section
r radiative
t total
w wall
168 M. Atashafrooz, S.A. Gandjalikhan Nassab/ Computers & Fluids 66 (2012) 167176
Author's personal copy
2. Problem description
Two-dimensional combined convection and radiation heat
transfer in a horizontal heated rectangular duct over a recess is
numerically simulated. Schematic of the computational domain
is shown in Fig. 1.
The upstream and downstream heights of the duct are h
1
and
h
2
, respectively. The height of the duct inside the recess region is
H such that this geometry provides the step height of s, with
expansion (ER = H/h
1
) and contraction (CR = h
2
/H) ratios of 2 and
0.5, respectively. The upstream length of the duct is considered
to be L
1
= 4H and the rest of the channel length is equal to
L
2
= 28H. This is made to ensure that the ows at the inlet and out-
let sections are not affected signicantly by the sudden changes in
the geometry and ow at the exit section becomes fully developed.
The length of recess depicted by D is considered to be 20H and 5H
in different test cases. The steps are considered to be inclined with
inclination angle of U, which is evaluated from the horizontal
sense.
3. Theory
For incompressible, steady and two-dimensional laminar ow,
the governing equations are the conservations of mass, momentum
and energy that can be written as follows:
@u
@x

@v
@y
0 1
u
@u
@x
v
@u
@y

1
q
@p
@x

l
q
@
2
u
@x
2

@
2
u
@y
2
_ _
2
u
@v
@x
v
@v
@y

1
q
@p
@y

l
q
@
2
v
@x
2

@
2
v
@y
2
_ _
3
@
@x
quc
p
T
@
@y
qvc
p
T j
@
2
T
@x
2

@
2
T
@y
2
_ _
r
~
q
r
4
In the above equations, u and v are the velocity components in
x- and y-directions, respectively, q the density, p the pressure, T
the temperature, l the dynamic viscosity, c
p
the specic heat, j
the thermal conductivity and ~q
r
is the radiative ux vector. The
boundary conditions are treated as no slip condition at the solid
walls (zero velocity) and constant temperature of T
w
at all boundary
surfaces. At the inlet duct section, the ow is fully developed with
uniform temperature of T
in
, which is assumed to be lower than
T
w
. At the outlet section, zero axial gradients for velocity compo-
nents and gas temperature are employed.
3.1. Gas radiation modeling
In the energy equation, besides the convective and conductive
terms, the radiative term as the divergence of the radiative heat
ux, i.e. r ~q
r
also presented. This radiative term can be computed
as follow [23]:
r
~
q
r
r
a
4pI
b
~r
_
4p
I~r;~sdX 5
In the above equation, I~r;~s is the radiation intensity at the situa-
tion ~r and in the direction ~s and I
b
~r
rT~r
4
p
is the black body
radiation intensity where r
a
is the absorption coefcient. For calcu-
lation of r ~q
r
, the radiation intensity eld is primary needed. To
obtain this term, it is necessary to solve the RTE. This equation for
an absorbing, emitting and scattering gray medium can be
expressed as [23]:
~s rI~r;~s bI~r;~s r
a
I
b
~r
r
s
4p
_
4p
I~r;~s/~s;~s
0
dX
0
6
where r
s
is the scattering coefcient, b r
a
r
s
the extinction
coefcient and /~s;~s
0
is the scattering phase function for the
radiation from incoming direction ~s
0
and conned within the so-
lid angle dX
0
to scattered direction ~s conned within the solid
angle dX. In this study, the phase function is equal to unity be-
cause the assumption of isotropic scattering assumption. The
boundary condition for a diffusely emitting and reecting gray
wall is:
I~r
w
;~s e
w
I
b
~r
w

1 e
w

p
_
~nw~s
0
<0
I~r
w
;~sj~n
w
~s
0
jdX
0
~n
w
~s > 0
7
in which e
w
is the wall emissivity, I
b
~r
w
the black body radiation
intensity at the temperature of the boundary surface and ~n
w
is the
outward unit vector normal to the surface. Since, the RTE depends
on the temperature elds through the emission term I
b
~r
w
, thus
it must be solved simultaneously with overall energy equation.
RTE is an integro-differential equation that can be solved with dis-
crete ordinates method.
In the DOM, Eq. (6) is solved for a set of n different directions, ~s
i
,
i = 1, 2, 3, . . . , n and integrals over solid angle are replaced by the
numerical quadrature, that is,
_
4p
f ~sdX

n
i1
w
i
f ~s
i
8
where w
i
are the quadrature weights associated with the directions
~s
i
. Thus, according to this method, Eq. (6) is approximated by a set of
n equations, as follows:
Fig. 1. Schematic of computational domain.
M. Atashafrooz, S.A. Gandjalikhan Nassab / Computers & Fluids 66 (2012) 167176 169
Author's personal copy
~s
i
rI~r;~s
i
bI~r;~s
i
r
a
I
b
~r
r
s
4p

n
j1
I~r;~s
j
/~s
j
;~s
i
w
j
i 1; 2; 3; . . . ; n 9
subjected to the boundary conditions:
I~r
w
;~s
i
e
w
I
b
~r
w

1 e
w

~nw~s
j
<0
I~r
w
;~s
j
j~n
w
~s
j
jw
j
~n
w
~s
i
> 0
10
Also r ~q
r
is represented as:
r
~
q
r
r
a
4pI
b
~r

n
i1
I~r;~s
i
w
i
11
At any arbitrary surface, heat ux may also be determined from
the surface energy balance as:
~
q nr
w
e
w
pI
b
r
w

~nw~s
j
<0
I
i
r
w
jn
w
s
i
jw
i
12
In 2-D Cartesian coordinate system, Eq. (9) becomes as follows
[23]:
n
i
@I
i
@x
g
i
@I
i
@y
bI
i
bS
i
i 1; 2; 3; . . . ; n 13
where
S
i
1 xI
b

x
4p

n
j1
I~r;~s
j
/~s
j
;~s
i
w
j
i 1; 2; 3; . . . ; n 14
In fact S
i
is a shorthand for the radiative source function. In Eq.
(14), xis the albedo coefcient, dened as x
rs
b
. The nite differ-
ence form of Eq. (13) gives the following relation for computation
of the radiant intensity [23]:
I
p
i

jn
i
jA
x
I
xi
i=c
x
jg
i
jA
y
I
yi
i=c
y
b8S
pi
b8 jn
i
jA
x
=c
x
jg
i
jA
y
=c
y
15
in which n
i
and g
i
are the direction cosines for the direction~s
i
and 8
is the element cell volume.
The details of the numerical solution of RTE by DOM were also
described in the previous work by the second author, in which the
thermal characteristics of porous radiant burners were investi-
gated [24].
For the radiative boundary conditions, the walls are assumed to
emit and reect diffusely with constant wall emissivity, e
w
= 0.8. In
addition, the inlet and outlet sections are considered for radiative
transfer as black walls at the uid temperatures in inlet and outlet
sections, respectively.
3.2. Non-dimensional forms of the governing equations
In numerical solution of the set of governing equations includ-
ing the continuity, momentum and energy, the following dimen-
sionless parameters are used to obtain the nondimensional forms
of these equations:
X; Y
x
D
h
;
y
D
h
_ _
; U; V
u
U

;
v
U

; P
p
qU
2

;
H
T T
in
T
w
T
in
; h
1

T
in
T
w
T
in
; h
2

T
w
T
in
; I

I
rT
4
w
;
S

S
rT
4
w
; s b D
h
; 1 x
r
a
b
; Pr
m
a
;
Re
qU
0
D
h
l
; Pe Re Pr; RC
rT
3
w
D
h
k
; q

r

q
r
rT
4
w
16
where D
h
is the hydraulic diameter which is equal to 2h
1
.
The non-dimensional forms of the governing equations are as
follows:
@U
@X

@V
@Y
0 17
@
@X
U
2

1
Re
@U
@X
_ _

@
@Y
UV
1
Re
@U
@Y
_ _

@P
@X
18
@
@X
UV
1
Re
@V
@X
_ _

@
@Y
V
2

1
Re
@V
@Y
_ _

@P
@Y
19
@
@X
UH
1
Pe
@H
@X
_ _

@
@Y
VH
1
Pe
@H
@Y
_ _

s1 xRCh
1
h
2
Pe
4
h
4
2
H
h
1
1
_ _
4

n
i1
I

i
w
i
_ _
0 20
3.3. The main physical quantities
The main physical quantities of interest in heat transfer study
are the mean bulk temperature and Nusselt number.
The mean bulk temperature along the channel was calculated
using the following equation:
H
b

_
1
0
HUdY
_
1
0
UdY
21
In the combined convectionradiation heat transfer, the energy
transport from the duct wall to the gas ow depends on two re-
lated factors:
1. Fluid temperature gradient on the wall.
2. Rate of radiative heat exchange on boundary surface.
Therefore, total heat ux on the wall is the sum of convective
and radiative heat uxes such that q
t
q
c
q
r
k
@T
@y
_ _
q
r
.
Therefore, the function of total Nusselt number Nu
t

q
t
H
kT
W
T
b

_ _
is
the sum of local convective Nusselt number (Nu
c
), and local radia-
tive Nusselt number (Nu
r
).
Accordingly, the total Nusselt numbers is given as follows [19]:
Nu
t
Nu
c
Nu
r

1
H
w
H
b
@H
@Y

Y0

RC h
1
h
2
H
w
H
b
q

r
22
Eq. (22) contains two parts. The rst term on the right hand side
represents the convective Nusselt number, whereas the second
term is the radiative one. It should be considered that for pure con-
vective hat transfer, total Nusselt number is equal to the convective
part.
4. Numerical procedure
Finite difference forms of the continuity, momentum and en-
ergy Eqs. (17)(20) were obtained by integrating over an elemental
Table 1
Grid independence study, Re = 400, D = 20H, RC = 100, x = 0.5, s = 0.005.
Grid size Value of the maximum
total Nusselt number
x
r
/D
h
150 14 3.542
260 20 5.41 3.624
320 25 5.69 3.782
400 30 5.88 3.890
480 36 6.07 4.014
530 40 6.09 4.025
170 M. Atashafrooz, S.A. Gandjalikhan Nassab/ Computers & Fluids 66 (2012) 167176
Author's personal copy
cell volume with staggered control volumes for the x- and y- veloc-
ity components. Other variables of interest were computed at the
grid nodes. The discretized forms of the governing equations were
numerically solved by the SIMPLE algorithm of Patankar and Spal-
ding [25]. Numerical solutions were obtained iteratively by the
line-by-line method. Numerical calculations were performed by
writing a computer program in FORTRAN. Based on the result of
grid tests for obtaining the grid-independent solutions, six differ-
ent meshes were used in the grid independence study. The corre-
sponding maximum value of the total Nusselt number on the
bottom wall and also the location of reattachment point are calcu-
lated and tabulated in Table 1. As it is seen, a grid size of 480 36
can be chosen for obtaining the grid independent solution, such
that the subsequent numerical calculations are made based on this
grid size. It should be mentioned that near the top, bottomand step
walls clustering is employed in the x- and y-directions for obtain-
ing more accuracy in the numerical calculations. To simulate the
incline surfaces in the computational domain, the blocked off
method is used in this study (see Fig. 2). Also, for computing the
divergence of radiative heat ux, which is needed for the numerical
solution of the energy equation, S
4
approximation has been used in
this study (Since, in the DOM, different numbers of discrete direc-
tions can be chosen during S
N
approximation, the results obtained
by S
4
, S
6
and S
8
approximations were compared and there was a
small difference, less 1% error, between S
4
and S
6
approximations).
Numerical solutions are obtained iteratively such that iterations
are terminated when sum of the absolute residuals is less than
10
4
for momentum and energy equations. But in the numerical
solution of RTE, the maximum difference between the radiative
intensities computed during two consecutive iteration levels did
not exceed 10
6
at each nodal point for the converged solution.
By this numerical strategy, the velocity, temperature and radiation
intensity distributions inside the computational domain can be
obtained.
5. Blocked-off method
In many cases, a computer program written for a regular grid
can be improved to handle an irregularly shaped computational
domain using the blocked-off method [26]. In this technique, the
whole 2-D region is divided into two parts: active and inactive or
blocked-off regions. The region where solutions are done is known
as the active region and the remaining portion is known as the
inactive or the blocked-off region. Therefore, by rendering inactive
some of the control volumes of the regular grid, the remaining ac-
tive control volumes form the desired irregular domain with com-
plex boundary. By this technique, the surface of inclined step in the
present analysis is approximated by a series of ne rectangular
steps, Fig. 2. It should be mentioned that the control volumes,
which are inside the active region, are designated as 1 and other-
wise they are 0, as shown in Fig. 2. It is obvious that using ne grids
in the interface region between active and inactive zones causes to
have an approximated boundary which is more similar to the true
boundary.
According to the blocked-off technique, known values of the
dependent variables must be established in all inactive control vol-
umes. If the inactive region represents a stationary solid boundary
as in the case, the velocity components in that region must be
equal to zero, and if the region is regarded as isothermal boundary,
the known temperature must be established in the inactive control
volumes.
6. Code validation
As it was mentioned before, there are not any theoretical and
experimental results about laminar forced convection ow of a
radiating gas over a recess including two inclined backward and
forward facing steps in a horizontal duct. Therefore, the numerical
results of this work are compared by the results of several investi-
gators for separately BFS or FFS convection ows.
To verify the accuracy of computations in obtaining the heat
transfer and ow characteristics of forced convection ow of a
radiating gas over BFS, the numerical implementation is validated
by reproducing the results of two investigators.
The results of reattachment point in ow over BFS with an
expansion ratio of 2 and for different Reynolds numbers are com-
pared with the experimental data, obtained by Armaly et al. [1],
in Fig. 3. As this gure shows, the reattachment point moves
Fig. 2. Blocked-off region in a regular grid.
M. Atashafrooz, S.A. Gandjalikhan Nassab / Computers & Fluids 66 (2012) 167176 171
Author's personal copy
toward the downstream side as the Reynolds number increases.
However, a good consistency is seen between experimental and
present numerical results. To compare the results of combined
convectionradiation heat transfer using DOM, another test case
is studied based on the results of Ansari and Gandjalikhan Nassab
[18] in which a 2-D laminar forced convection of gas ow over a
BFS was analyzed. In this test case, the boundary conditions are
treated as no slip condition at the solid walls (zero velocity) and
constant temperature of T
w
at all boundary surfaces. At the inlet
duct section, the ow is fully developed with uniform temperature
of T
in
, which is assumed to be lower than T
w
.
Distribution of convective Nusselt number along the bottom
wall is compared with that reported in Ref. [18]. The results are
presented graphically in Fig. 4. As it is seen from this gure, a good
agreement exists between the numerical results of the present
work with those obtained by Ansari and Gandjalikhan Nassab [18].
To check the performance and accuracy of the results obtained
by the blocked-off technique in solving the governing equations,
the numerical computation is validated by the results of Ansari
and Gandjalikhan Nassab [20] over a FFS. In Ref. [20], the block-
off method was considered for simulating incline surface of a
FFS, with inclination angle of U= 45. Distribution of mean bulk
temperature calculated in the present study is compared with that
obtained by Ansari and Gandjalikhan Nassab [20] in Fig. 5. This
gure shows that how the uid ow warms up from the inlet to
outlet section. However, a good agreement is achieved between
the reported results.
7. Results and discussion
In this study, numerical results are presented for combined
convection and radiation heat transfer over a recess including
two inclined backward and forward facing steps in a horizontal
duct with two subsequent expansion ratio of ER = 2 and contrac-
tion ratio of CR = 0.5 at different conditions. In numerical calcula-
tions, the Reynolds number is equal to 400, while the Prandtl
number is kept constant at 0.71 to guarantee the constant uid
physical properties.
First, in order to showthe owpattern, the streamlines are plot-
ted in Fig. 6 while the recess length is equal to D = 20H. The effect
of sudden expansion and contraction along two steps is clearly
seen from the curvatures of streamlines. Fig. 6a shows that three
main recirculation zones are encountered in the ow domain for
the step inclination angle of / = 90 at Re = 400. The primary recir-
culation region occurs downstream the backward step adjacent the
bottomwall, whereas the secondary recirculation zone exists along
the top wall and the third recirculation region with small extent
occurs on the bottom wall before the forward step and adjacent
to the step corner. It should be noted that for small values of the
Reynolds number (say for Re < 350 for this test case), the secondary
recirculation zone disappears. To see the effect of step inclination
angle on the uid ow behavior, Fig. 6be show the streamline
contours for / = 60, 30, 20 and 10, respectively. It can be seen
from these gures that by decreasing step inclination angle, the
length of reattachment point in the primary recirculated zone de-
creases while the extents of second and third recirculation regions
decrease and nally these domains disappear for very low inclina-
tion angles.
In the convection ow of a radiating uid, the radiationcon-
duction parameter (RC), the optical thickness (s) and the albedo
coefcient (x) are the main parameters that affect the thermal
behavior of the radiationconvection system. In the next gures,
an attempt is made to study the effect of these parameters on ther-
mal behavior of the thermal system. Also, the effects of step incli-
nation angle and recess length on the Nusselt number are studied
in the present work.
One of the main parameters in the combined radiationconduc-
tion systems, which show the relative importance of the radiation
Re
s
=U
0
s/
x
r
/
s
0 50 100 150 200 250 300
0
2
4
6
8
10
Present Work
Armaly [1]
ER = 2
Fig. 3. Comparison of reattachment point.
X
N
u
c
0 5 10 15 20
0
0.5
1
1.5
2
2.5
3
Present work
Ansari and Gandjalikhan Nassab [18]
Re = 400
ER = 2
RC = 50
= 0.5
= 0.005
x
h
H
Fig. 4. Comparison of convective Nusselt number along the bottom wall.
X

b
0 5 10 15 20
0
0.2
0.4
0.6
0.8
1
Present Work
Ansari and Gandjalikhan Nassab [20]
Re = 400
= 45
o
CR = 0.5
RC = 150
= 0.5
= 0.005
T
W
5H
x
T
W
H
h

Fig. 5. Comparison of mean bulk temperature along the duct.

172 M. Atashafrooz, S.A. Gandjalikhan Nassab/ Computers & Fluids 66 (2012) 167176
Author's personal copy
mechanism compared with its conduction counterpart is the radi-
ationconduction parameter (RC). High value of RC parameter
shows the radiation dominance in a thermal system.
For the convective ow with radiating heat transfer in a 2-D
channel including two inclined backward and forward facing steps,
as shown in Fig. 1, the distributions of convective, radiative and to-
tal Nusselt numbers along the bottom wall at different values of
the RC parameter are presented in Fig. 7ac, respectively.
Fig. 7a shows that the minimum value of convective Nusselt
number occurs on the bottom wall at the backward step corner,
where the uid is at rest. Also, it is seen that after this point, the
value of Nu
c
increases sharply in the primary recirculation region
because of the ow vortices, such that the maximum Nu
c
on the
bottom wall coincides with the point of reattachment after which
Nusselt number decreases and approaches to a constant value as
the distance continues to increase in the stream-wise direction.
Then, the Nusselt number decreases along the ow direction such
that the value of Nusselt number vanishes at the forward step
X
0
0.5
1
Y
5 10 15 20 0
(a)

90 =
X
0 5 10 15 20
0
0.5
1
Y
(b)

60 =
X
0 5 10 15 20
0
0.5
1
Y
(c)

30 =
X
0 5 10 15 20
0
0.5
1
Y
(d)

20 =
X
0 5 10 15 20
0
0.5
1
Y
(e)

10 =
Fig. 6. Distribution of streamlines contours, Re = 400, (a) / = 90, (b) / = 60, (c) /
= 30, (d) / = 20, (e) / = 10.
X
N
u
c
0 5 10 15 20
0
1
2
3
4
5
6
RC = 200
RC = 150
RC = 100
RC = 50
No Radiation
Re = 400
= 45
o
= 0.5
= 0.005
(a) Convective Nusselt number
X
N
u
r
0 5 10 15 20
0
1
2
3
4
RC = 200
RC = 150
RC = 100
RC = 50
Re = 400
= 45
o
= 0.5
= 0.005
(b) Radiative Nusselt number
X
N
u
t
0 5 10 15 20
0
1
2
3
4
5
6
7
8
RC = 200
RC = 150
RC = 100
RC = 50
No Radiation
Re = 400
= 45
o
= 0.5
= 0.005
(c) Total Nusselt number
Fig. 7. Effect of RC on the Nusselt number distribution along the bottom wall, (a)
Convective Nusselt number. (b) Radiative Nusselt number. (c) Total Nusselt
number.
M. Atashafrooz, S.A. Gandjalikhan Nassab / Computers & Fluids 66 (2012) 167176 173
Author's personal copy
corner where the uid is at rest. Moreover, Fig. 7a illustrates that
the convective Nusselt number decreases by increasing in RC
parameter. This is due to this fact that under the effective presence
of radiation mechanism at high values of RC, the temperature eld
in the participating media becomes more uniform. Consequently,
the amount of temperature gradient inside the ow domain de-
creases that cases a decrease in the value of convective Nusselt
number. But a different trend is seen from Fig. 7b for the variation
of radiative Nusselt number with RC parameter. This gure shows
that Nu
r
starts from a minimum value at the backward step corner
where the minimum radiative heat ux leave the bottom wall. As
the distance increases from the step corner, the amount of radia-
tive heat ux and consequently radiative Nusselt number increases
sharply, which is due to a decrease in bottom wall incident radia-
tive heat ux incoming from the inclined stepped wall. After this
sharp slope, Nu
r
increases by a smooth slope until it reaches to a
local maximum value as the distance continues to increase in the
stream-wise direction Then, the radiative Nusselt number de-
creases along the ow direction up to the forward step corner on
the bottom wall. Furthermore, Fig. 7b shows that the radiative
Nusselt number increases by increasing in RC parameter, which
is due to the increase in bottom walls outgoing radiative heat ux.
It is seen that RC parameter has a greater inuence on the radiative
Nusselt number than the convective Nusselt number such that to-
tal Nusselt number increases by increasing in RC, as shown in
Fig. 7c.
Another important parameter in participating medium is the
optical thickness that affects the temperature distribution. This
parameter is a well-known radiation property, such that high val-
ues for s means that the medium has great ability to absorb and
emit radiant energy.
To study the effect of optical thickness on the mean bulk tem-
perature distribution along the duct Fig. 8 is plotted. This gure
shows that the mean bulk temperature increases along the duct
because of both convection and radiation mechanisms. As it is seen
from Fig. 8, increase of optical thickness and consequently increase
of radiation heat transfer mechanism causes an increase in the
amount of the gas mean bulk temperature.
For more study about the effect of optical thickness on the ther-
mal behavior of convectionradiation system, the distribution of
total Nusselt number on the bottom wall at different values of
the optical thickness is plotted in Fig. 9. It is seen that as the med-
iums ability to absorb and emit thermal radiation becomes greater at high values of the optical thickness, such systems have high val-
ues for the total Nusselt number.
The scattering albedo is an important parameter in radiating
systems that can show the ability of participating medium to scat-
ter thermal radiation. As it was mentioned before, scattering albe-
do, x, is dened as x
rs
b
. The extreme values of scattering albedo,
i.e., x= 1.0 and x= 0.0 correspond to pure scattering and non-
scattering cases, respectively. Therefore, the medium changes from
pure absorption to pure scattering by increasing x from zero to
unity.
The effect of scattering albedo on the mean bulk temperature
is presented in Fig. 10. This gure shows that in the case of hav-
ing participating medium with high scattering effect, the rate of
heat transfer from the heated walls into convection ow de-
creases, such that Fig. 10 depicts that the mean bulk temperature
decreases by increasing in x. Because, less radiative heat ux is
converted to gas thermal energy in a pure scattering case com-
pared to a pure absorption one. Beside, it can be found from
Fig. 10 that when the radiation term is omitted from the energy
equation in the case of no-radiation problems, the convective
system has the same trend and behavior as pure scattering case
with x= 1.0.
One of the main factors in controlling the rate of convection
heat transfer is the step inclination angle. Fig. 11 obviously shows
X

b
0 5 10 15 20
0
0.2
0.4
0.6
0.8
1
= 0.01
= 0.0075
= 0.005
= 0.0025
No Radiation
Re = 400
= 45
o
RC = 100
= 0.5
Fig. 8. Effect of optical thickness on the mean bulk temperature distribution along
the duct.
X
N
u
t
0 5 10 15 20
0
1
2
3
4
5
6
7
8
= 0.01
= 0.0075
= 0.005
= 0.0025
No Radiation
Re = 400
= 45
o
RC = 100
= 0.5
Fig. 9. Effect of optical thickness on the total Nusselt number distribution along the
bottom wall.
X

b
0 5 10 15 20
0
0.2
0.4
0.6
0.8
1
= 0.0
= 0.25
= 0.5
= 0.75
= 1.0
No Radiation
Re = 400
= 45
o
RC = 100
= 0.005
Fig. 10. Effect of albedo on the mean bulk temperature distribution along the duct.
174 M. Atashafrooz, S.A. Gandjalikhan Nassab/ Computers & Fluids 66 (2012) 167176
Author's personal copy
that how the convective Nusselt number affected by this
parameter. As it is seen from this gure, the amount of maximum
Nusselt number increases by increasing the step inclination angle,
which is due to high rates for ow vortices in the primary recircu-
lated domain for more inclined steps. In addition, Fig. 11 reveals
that by increasing the step inclination angle, the reattachment
point where the maximum Nusselt number occurs moves toward
the downstream side.
To study the effect of recess length on the ow and heat transfer
behaviors of convection ow, another test case with the same Rey-
nolds number (Re = 400) is also analyzed for a short recess length
with D = 5H. To have a better view of the ow pattern, the stream-
lines contour for D = 5H and / = 90 is plotted in Fig. 12. If one com-
pares the streamlines in Fig. 12 with those plotted in Fig. 6, it can
be concluded that the ow eld has different patterns in long and
short recess lengths, such that it is seen that the recess length has a
great effect on the ow pattern. In this test case, because of the
short recess length D, the secondary recirculation region disap-
pears while the primary and third recirculation regions affect each
other. The effect of RC parameter on the distributions of convec-
tive, radiative and total Nusselt numbers along the bottom wall
in this test case with short recess length is showed in Fig. 13.
Comparison of this gure with Nu distribution shown in Fig. 7
illustrates that Nu distribution has different trends in short and
long recess. Fig. 13a indicates that convective Nusselt number in-
creases after the backward step in the primary recirculation region
and reaches to its maximum value nearly at the end of primary
recirculation region and then decreases to its minimum value at
the forward step corner where the ow is at rest. As it is seen from
Fig. 13b, radiative Nusselt number increases after the backward
step corner and reaches to its maximum value and then decreases
to its minimum value at the forward step corner. Moreover, Fig. 13
presents that the effect of RC parameter on Nusselt number distri-
butions in this recess length is similar to those found in Fig. 7 that
were explained before.
X
N
u
c
0 5 10 15 20
0
1
2
3
4
5
6
= 30
o
= 60
o
= 90
o
Re = 400
RC = 100
= 0.5
= 0.005
Fig. 11. Effect of inclination angle on the convective Nusselt number distribution
along the bottom wall.
X
0
0.5
1
Y
0 5
Fig. 12. Distribution of streamlines contour, D = 5H, Re = 400, / = 90.
X
N
u
c
1 2 3 4
0
1
2
3
4
RC = 200
RC = 150
RC = 100
RC = 50
No Radiation
Re = 400
= 90
o
= 0.5
= 0.005
0 5
(a) Convective Nusselt number
X
N
u
r
1 2 3 4
0
1
2
3
RC = 200
RC = 150
RC = 100
RC = 50
Re = 400
= 90
o
= 0.5
= 0.005
0 5
(b) Radiative Nusselt number
X
N
u
t
1 2 3 4
0
1
2
3
4
5
RC = 200
RC = 150
RC = 100
RC = 50
No Radiation
0 5
Re = 400
= 90
o
= 0.5
= 0.005
(c) Total Nusselt number
Fig. 13. Effect of RC on the Nusselt number distribution along the bottom wall,
D = 5H (a) Convective Nusselt number. (b) Radiative Nusselt number. (c) Total
Nusselt number.
M. Atashafrooz, S.A. Gandjalikhan Nassab / Computers & Fluids 66 (2012) 167176 175
Author's personal copy
8. Conclusion
Analysis of combined convection and radiation heat transfer
over a recess including two inclined backward and forward facing
steps in a horizontal duct is studied in this research work. The set
of governing equations consisting of the continuity, momentum
and energy is solved numerically by the CFD techniques in the
Cartesian coordinate system. The blocked off method is used to
simulate the incline surfaces of FFS and BFS. For calculating the
radiative term in the energy equation, the RTE is solved by the
DOM to obtain the distribution of radiant intensity inside the radi-
ating medium. The effects of radiationconduction parameter,
optical thickness, albedo coefcient, step inclination angle and
recess length on the thermal behavior of the convective system
are thoroughly explored by plotting the variations of streamlines
contours, Nusselt number (total, radiative and convective) and
mean bulk temperature under different conditions.
References
[1] Armaly BF, Durst F, Pereira JCF, chonung B. Experimental and theoretical
investigation of backward-facing step ow. J Fluid Mech 1983;127:47396.
[2] Vradis G, Nostrand VL. Laminar coupled ow downstream an asymmetric
sudden expansion. J Thermophys Heat Transfer 1992;6(2):28895.
[3] Tylli N, Kaiktsis L, Ineichen B. Side wall effects in ow over backward-facing
step: experiments and numerical solutions. Phys Fluids 2002;14(11):383545.
[4] Brakely D, Gabriela M, Gomes M, Henderson RD. Three - dimensional
instability in ow over a backward facing step. J Fluid Mech
2002;473:16790.
[5] Kondoh T, Nagano Y, Tsuji T. Computational study of laminar heat transfer
downstream of a backward-facing step. Int J Heat Mass Transfer
1993;36(3):57791.
[6] Erturk E. Numerical solutions of 2-D steady incompressible ow over a
backward-facing step, Part I: High Reynolds number solutions. Comput Fluids
2008;37:63355.
[7] Abu-Nada E. Numerical prediction of entropy generation in separated ows.
Entropy 2005;7:23452.
[8] Abu-Nada E. Entropy generation due to heat and uid ow in backward facing
step ow with various expansion ratios. Int J Exergy 2006;3:41935.
[9] Abu-Nada E. Investigation of entropy generation over a backward facing step
under bleeding conditions. Energy Convers Manage 2008;49:323742.
[10] Bahrami A, Gandjalikhan Nassab SA. Study of entropy generation in laminar
forced convection ow over a forward-facing step in a duct. Int Rev Mech Eng
2010;4(4):399404.
[11] Abu- Mulaweh HI. A reviewof research on laminar mixed convection ow over
backward- and forward-facing steps. Int J Therm Sci 2003;42:897909.
[12] Abu-Mulaweh HI. Turbulent mixed convection ow over a forward-facing step
the effect of step heights. Int J Therm Sci 2005;44:15562.
[13] Yilmaz I, ztop HF. Turbulence forced convection heat transfer over double
forward facing step ow. Int Commun Heat Mass Transfer 2006;33:50817.
[14] Chen YT, Nie JH, Hsieh HT, Sun LJ. Three-dimensional convection ow adjacent
to inclined backward-facing step. Int J Heat Mass Transfer 2006;49:4795803.
[15] Gandjalikhan Nassab SA, Moosavi R, Hosseini Sarvari SM. Turbulent forced
convection ow adjacent to inclined forward step in a duct. Int J Thermal Sci
2009;48:131926.
[16] Bouali H, Mezrhab A. Combined radiative and convective heat transfer in a
divided channel. Int J Numer Math 2006;16:84106.
[17] Azad FH, Modest MF. Combined radiation and convection in absorbing
emitting and anisotropically scattering gas-particulate ow. Int J Heat
Transfer 1981;24:168198.
[18] Ansari AB, Gandjalikhan Nassab SA. Thermal characteristics of convective
ows encountered in a 2-D backward facing step under the effect of radiative
heat transfer. Int Rev Mech Eng 2010;4(6):7118.
[19] Ansari AB, Gandjalikhan Nassab SA. Study of laminar forced convection of
radiating gas over an inclined backward facing step under bleeding condition
using the blocked-off method. ASME, J Heat Transfer 2011;133(7):072702.
[20] Ansari AB, Gandjalikhan Nassab SA. Numerical analysis of laminar forced
convection ow of a radiating gas over an inclined forward facing step. Int Rev
Mech Eng 2011;5(1):1207.
[21] Chiu HC, Jang JH, Yan WM. Mixed convection heat transfer in horizontal
rectangular ducts with radiation effects. Int J Heat Mass Transfer
2007;50:287482.
[22] Chiu HC, Yan WM. Mixed convection heat transfer in inclined rectangular
ducts with radiation effects. Int J Heat Mass Transfer 2008;51:108594.
[23] Modest MF. Radiative Heat Transfer. second ed. NewYork: McGraw-Hill; 2003.
[24] Keshtkar MM, Gandjalikhan Nassab SA. Theoretical analysis of porous radiant
burners under 2-D radiation eld using discrete ordinates method. J Quant
Spectrosc Radiat Transfer 2009;110:1894907.
[25] Patankar SV, Spalding DB. A calculation procedure for heat, mass and
momentum transfer in three-dimensional parabolic ows. Int J Heat Mass
Transfer 1972;15(10):1787806.
[26] Patankar SV. Numerical Heat Transfer and Fluid Flow. Philadelphia, PA: Taylor
& Francis; 1981 (Chapter 7).
176 M. Atashafrooz, S.A. Gandjalikhan Nassab/ Computers & Fluids 66 (2012) 167176