Thermal Science and Engineering Progress 4 (2017) 7184

Contents lists available at ScienceDirect

Thermal Science and Engineering Progress

journal homepage: www.elsevier.com/locate/tsep

Singular perturbation and differential transform methods to

two-dimensional flow of nanofluid in a porous channel with
expanding/contracting walls subjected to a uniform transverse
magnetic field
M.G. Sobamowo
Department of Mechanical Engineering, University of Lagos, Akoka, Lagos, Nigeria

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

Article history: This paper presents a comparative study of two approximate analytical methods for the analysis of two-
Received 17 April 2017 dimensional unsteady flow of nanofluid in a porous channel through expanding/contracting walls with
Received in revised form 9 August 2017 large injection or suction under the influence of a uniform transverse magnetic field. Similarity transfor-
Accepted 2 September 2017
mations are used to reduce the systems of the governing partial differential equations to a nonlinear
fourth-order ordinary differential equation which is solved using method of matched asymptotic expan-
sion and differential transform method. From the verification of the approximate analytical solutions, it is
Keywords:
established that the results obtained using differential transformation method demonstrate remarkable
Nanofluid
Porous channel
accuracy and better agreements with the results of numerical method than the results obtained using
Expanding or contracting walls method of matched asymptotic expansion. Therefore, the differential transform method gives more accu-
Transverse magnetic field rate results than method of matched asymptotic expansion. Also, due to the higher accuracy of the dif-
Method of matched asymptotic expansion ferential transformation method than the method of matched asymptotic expansion, the approximate
Differential transformation method analytical solutions through the differential transformation method are used to investigate the effects
of permeation Reynolds number, wall expansion ratio and magnetic parameter on the flow behavior of
the fluid. From the results, it is established that increase in the Reynolds number decreases the axial
velocity at the center of the channel during the expansion while the axial velocity increases slightly near
the surface of the channel when the wall contracts at the same rate. Also, as the wall expansion ratio
increases, the velocity at the center decreases and increases near the wall. For every level of injection
or suction, in the case of expanding wall, increasing the wall dilation rate leads to increase in axial veloc-
ity near the center and decrease in the axial velocity near the wall. The results present in this work can
serve as references for the other methods of analysis of the problem.
2017 Published by Elsevier Ltd.

238
1. Introduction U by gaseous diffusion and also flow in multichannel filtration
systems such as the wall flow monolith filter used to reduce emis-
The study of laminar flow through a porous channel or pipe sions from diesel engines [18] etc. The equations governing such
with contracting or expanding permeable walls has received con- flow processes are generally nonlinear equations which are diffi-
siderable amount of research efforts in the past few decades due cult to solve exactly and analytically. In order to predict and deter-
to its various biological and engineering applications. These appli- mine the actual flow behavior, different approximate analytical
cations are evident in the transport of biological fluids through and numerical methods have been employed to solve the nonlinear
contracting or expanding vessels, filtration in kidneys and lungs, governing equations. In the past studies of the laminar flow
flow inside lymphatics, the synchronous pulsation of porous dia- through a porous channel, Majdalani [5] and Majdalani and Roh
phragms, modeling of air circulation in respiratory system, model [6] employed asymptotic methods using Wentzel-Krammers-
of the regression of the burning surface in the solid rocket motors, Brillouin (WKB) and multiple-scale techniques to study the oscilla-
binary gas diffusion, chromatography, ion exchange, ground water tory channel flow with wall injection. Majdalani and van Moorhem
movement, transpiration cooling and the separation of 235U from [7] adopted the same method of multiple-scales to develop an
approximate analytical solution to the laminar flow problem while
E-mail addresses: [email protected], [email protected] Oxarango et al. [8] presented a new model for the laminar flow in

http://dx.doi.org/10.1016/j.tsep.2017.09.001
2451-9049/ 2017 Published by Elsevier Ltd.
72 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

Nomenclature

a_ time-dependent rate, m/s x coordinate axis parallel to the channel walls, m

Bo electromagnetic induction, T or kg/s
2 A
1 
y coordinate axis perpendicular to the channel walls, m
Ha Hartmann number qnf density of the nanofluid, kg/m3

p pressure, N/m2 qf density of the fluid, kg/m3
Re permeation Reynolds number lnf dynamic viscosity of the nanofluid, kg/ms or Ns/m2
t time, s qs density of the nanoparticles, kg/m3

u velocity component in x-direction, m/s / fraction of nanoparticles in the nanofluid
v velocity component in y-direction, m/s r electrical conductivity, S/m or kg
1 m
3 s3 A2
Vw fluid inflow velocity at the wall, m/s a dimensionless wall dilation rate

channels with wall suction or injection in multi-channel filtration in many real life and practical problems) and weak nonlinearities.
systems. Jankowski and Majdalani [9] applied method of Consequently, there have been various attempts in recent times to
multiple-scale to analyze oscillatory channel flow with arbitrary develop non-perturbation methods such as homotopy analysis
suction. In a previous work [10], the same authors developed an method (HAM), Adomian decomposition method (ADM), varia-
analytical solution by means of the Liouville-Green transformation tional iteration method (VIM), differential transformation method
for laminar flow in a porous channel with large wall suction and a (DTM), variation of parameter method (VPM), double decomposi-
weakly oscillatory pressure while Zhou and Majdalani [11] used tion method (DDM), optimal homotopy asymptotic method
finite difference method and asymptotic technique (variation of (OHAM), methods of weighted residuals (MWR), Daftardar-Gejiji
parameters and small parameter perturbations) to investigate the and Jafari method (DJM), Temini and Ansari method (TAM) etc.
meanflow for slab rocket motors with regressing walls. The results to overcome the deficiencies of perturbation methods. The applica-
from the two methods were compared for different Reynolds num- tions of these non-perturbation methods to boundary value prob-
bers Re and the wall regression rate a, and it was observed that lems require the search for a value or values that will satisfy the
accuracy of the analytical solution deteriorates for small Re and end boundary condition(s). Such quest necessitates the use of soft-
large a. A good agreement between the solutions was observed ware (Maple, Mathematica or MatLab) which consequently results
for large values of Re. A similar analysis was done by Majdalani in additional computational cost in the generation of solutions to
and Zhou [12] for moderate-to-large injection and suction driven the problems. Also, the solutions reported for the other relatively
channel flows with expanding or contracting walls. Multiple solu- sophisticated methods to nonlinear problems have good accuracy,
tions associated with this problem have been reported by Robinson but they are more complicated in applications and analysis than
[13], Zarturska et al. [14] and Si et al. [1516]. Majdalani et al. [17] perturbation methods. Therefore, over the years, the relative sim-
applied regular perturbation method to study two-dimensional plicity and high accuracy in the limit of small parameter have
viscous flow between slowly expanding or contracting walls with made perturbation methods (regular perturbation, method of
weak permeability. In a recent study, Dinarvand et al. [18] adopted asymptotic expansion, method of matched asymptotic expansion,
homotopy analysis method (HAM) and homotopy perturbation method of multiple scales, Method of WKB approximation, Method
methods (HPM) to solve Bermans model of two-dimensional vis- of JWKB (Jeffery-Wentzel-Krammers-Brillouin) approximation,
cous flow in porous channels with wall suction or injection. The Poincare-Lindstedt method, Periodic averaging method, homotopy
authors concluded that the HPM solution is not valid for large Rey- perturbation method, parameterized perturbation method etc.)
nolds numbers, a weakness that has been earlier observed in the interesting tools among the most frequently used approximate
other perturbation techniques. Effects of nanofluid and magnetic analytical methods. Perturbation methods provide in general,
field on the fluid flow process have been extensively analyzed accurate results for small perturbation parameters with handy
[1929]. Using the homotopy analysis method, Xu et al. [30] devel- mathematical formulations. The approximate analytical methods
oped highly accurate series approximations for two-dimensional such as HAM, ADM, DTM, VIM etc. have their own operational
viscous flow between two moving porous walls and obtained mul- restrictions that severely narrow their functioning domain as they
tiple solutions associated with this problem. Also, the same are limited to small domain. Using DTM, HAM, ADM, VIM for large
method was adopted by Dinarvand and Rashidi [31] to analyze or infinite domain is done with either the application of before-
two dimensional viscous flow in a rectangular domain bounded treatment techniques such as domain transformation techniques,
by two moving porous walls. Although, the homotopy analysis domain truncation techniques and conversion of the boundary
method is a reliable and efficient semi-analytical technique, it value problems to initial value problems or the use of after-
however suffers from a number of limiting assumptions such as treatment techniques such as Pade-approximants, basis functions,
the requirements that the solution ought to conform to the so- cosine after-treatment technique, sine after-treatment technique
called rule of solution expression and the rule of coefficient ergod- and domain decomposition technique. This is because many of
icity. Moreover, in many cases, the approximate analytical solu- them were initially established for initial value problems. Amend-
tions through the homotopy analysis method come with large ing the methods to boundary value problems especially for large or
number of terms. In practice, analytical solutions with large num- infinite domains boundary value problems leads to the search for
ber of terms and conditional statements for the solutions are not values (s) of the unknown parameter(s) that will satisfy the end
convenient for use by designers and engineers [32]. boundary condition(s). This drawback in the other approximation
The importance of perturbation methods (PMs) in providing analytical methods is not experienced in perturbation methods
approximate analytical solution to nonlinear problems has been as such tasks of before- and after-treatment techniques might
well established in literatures [33]. However, the applications of not be necessary in the perturbation methods.
the perturbation methods (regular and singular perturbation In the class of the perturbation methods for solving singular
methods) and the validity of their approximate analytical solutions perturbation problems (problems where a very small parameter
are limited to small perturbation parameter (which may not exist multiplies the highest derivative of a differential equation), the
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 73

method of matched asymptotic expansion is mostly used to deter- bounded by two permeable surfaces or walls that enable the nano-
mine a uniform and accurate approximation to the solution of sin- fluid to enter or exit during successive expansions or contractions
gularly perturbed differential equations and to find global as shown in Fig. 1. One end of the channel is closed by a compliant
properties of differential equations. solid membrane. Both walls are assumed to have equal permeabil-
The relatively new approximate analytical method, differential _
ity and to expand uniformly at a time dependent rate, at. A coor-
transformation method (DTM) has proven to be more effective dinate system is chosen with the origin at the center of the channel
than most of the other approximate analytical solutions as it does as shown in the figure.
not require many computations as carried out in Adomian decom- Following the assumptions, the equations for continuity and
position method, homotopy analysis method, regular and singular momentum are
perturbation methods, homotopy perturbation method, optimal  @ v
@u
homotopy asymptotic method, optimal homotopy perturbation 0 1
@ x @ y

method, variational iteration method, variation of parameter
method, double decomposition method, Daftardar-Gejiji and Jafari
 !

@u @u 
@u @p @2u  @2u
method, Temini and Ansari method etc. Also, the differential trans- qnf 
u v

lnf 
2
rB2o u 2
formation method as introduced by Zhou [34] has fast gained @t @ x 
@y @ x @ x2 @ y

ground as it appears in many engineering and scientific research

 !
papers because of its comparative advantages over the other @ v @ v @ v 
@p @ 2 v @ 2 v
approximate analytical methods. It is a method that solves differ- qnf 
u v
lnf 2 3
@t @ x @y  
@y @ x2 @ y
ential equations, difference equation, differential-difference equa-
tions, fractional differential equation, pantograph equation and where
integro-differential equation without linearization, discretization,
restrictive assumptions, perturbation and discretization or round-
qnf qf 1
/ qs / 4
off error. It reduces complexity of expansion of derivatives and
lf
the computational difficulties of the other traditional and approx- lnf 5
imation analytical methods. Using DTM, a closed form series solu- 1
/2:5
tion or approximate solution can be obtained as it provides Assuming no slip condition, the appropriate boundary condi-
excellent approximations to the solution of non-linear equation tions are given as
with high accuracy. It is capable of greatly reducing the size of
a
computational work while still accurately providing the series  at;
y  0;
u v
V w

solution with fast convergence rate. It is a more convenient c
method for engineering calculations than other approximate ana- 
@u
lytical or numerical methods. It appears more appealing than the  0;
y

0; v 0
@y
numerical solution as it comparatively reduces the computation
costs, simulations and task in the analysis of nonlinear problems. x 0; 0
u 6
Moreover, the need for small perturbation parameter as required
in traditional perturbation methods (PMs), the rigour of the deter- where c Va_w is the wall presence or injection/suction coefficient i.e.
mination of Adomian polynomials as carried out in ADM, the which is the measure of permeability
restrictions of HPM to weakly nonlinear problems as established On introducing the following stream functions and the mean
in literatures, the lack of rigorous theories or proper guidance for flow vorticity
choosing initial approximation, auxiliary linear operators, auxiliary

@w @w
functions, auxiliary parameters, and the requirements of confor- 
u

; v 7
mity of the solution to the rule of coefficient ergodicity as done @y @ x
in HAM, the search Langrange multiplier as carried in VIM, the The pressure term in Eqs. (2) and (3) can be eliminated and the
challenges associated with proper construction of the approximat- vorticity transport equation is obtained as
ing functions for arbitrary domains or geometry of interest as in
  !
Galerkin weighted residual method (GWRM), least square method @n @ n @ n @ 2 n @ 2 n 
@u
qnf  v
u lnf 
2 rB2o u 8
(LSM) and collocation method (CM) are some of the difficulties that @t @ x @y  @ x2 @ y 
@y
DTM overcomes.
The various applications of the reviewed laminar flow in a por- where

n @ v
@ u
ous channel with deformable boundaries require accurate predic-  
tions and determinations of the actual flow behavior. Therefore, 9
@ x @ y
in this work, a comparative study of a singular perturbation
method (method of matched asymptotic expansion) and differen-
tial transform method is carried out for the analysis of two-
dimensional unsteady flow of nanofluid in an expanding/contract-
ing porous channel with large suction and injection under the
influence of transverse magnetic field. Also, the developed solu-
tions are used to study the effects of the flow parameters on the
two-dimensional laminar flow of the nanofluid through expanding
or contracting porous channel with large suction and injection.

2. Problem formulation

Consider a fully developed unsteady, laminar, isothermal, and

incompressible flow in a two-dimensional porous channel Fig. 1. The model of the porous channel with expanding or contracting walls.
74 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

Also, the above partial differential equation can be converted to singularly perturbed differential equations and to find global prop-
ordinary differential equation using the following similarity trans- erties of differential equations such as eigenvalues [33]. In apply-
formations based on the following similarity variables ing the MMAE for a singularly perturbed boundary value
lnf x 0
lnf  problem, the interval is divided into two overlapping subintervals,

u f g; t; v f g; t 10 an inner region and an outer region. Each subinterval is used to
qnf H2 qnf H
obtain an asymptotic approximation to the solution of differential
where equation valid on the interval of each region. Finally, the outer and
inner solutions are then combined through a process called

y @ f g; t matching in such a way that an approximate solution for the
g ; f 0 g; t
H @g whole domain is obtained [33]. It should also be pointed out that
the matching process assures that the asymptotic approximations
Substituting Eq. (9) and (10) into Eq. (8), one arrives at a fourth
have the same functional form on the overlap of every pair of inter-
order ordinary differential equation of the form
vals and that it satisfies all the boundary conditions given at vari-
2
f    qnf a  ous points on the interval.
gggg atgf ggg 3f gg f f ggg
f g f gg
f ggt
Ha2 f gg 0 For small injection and suction at the walls where the perme-
lnf
ation Reynolds number is small, the above Eq. (16) can easily be
11
solved using regular perturbation method. Following the regular
And the boundary conditions becomes perturbation method (RPM) procedures, it can easily be shown that
the series solution of Eq. (16) which satisfies the boundary condi-
g 0; f 0; f gg 0 tions in Eq. (15) is given as
12
g 1; f Re; f g 0 g


1 7 1 11 3 1

f g 3
g2 a g
g5 g g
_
qnf aat 2 560 80 560 112
where at lnf is the non-dimensional wall dilation rate which ( )
1
g 7

3
g3
1
g
is positive for expansion and negative for contraction.and Re nfl w
q aV Re 280  1 280 11 140

nf
a 184800 g
20161
g9 4900
3
g7 1400
3
g5
776160
3233
g3 1293600
2459
g
is the permeation Reynolds number, which is positive for injection 17
and negative for suction.
Using the following variables, However, application of regular perturbation method to the
nonlinear Eq. (16) breaks down when the injection and the suction

w 
u v f xf at the walls are large. This is because, when the permeation Rey-
w ; u ; v ; f ) w ;
aa_ a_ a_ Re c nolds number is large, the problem becomes a singular perturba-
0
xf
f a tion problem. In order to obtain the analytical solution that is
u ; v ; c 13 uniformly valid for the channel, a singular perturbation method
c c Re
i.e. method of matched asymptotic expansion is applied in this
Eqs. (11) and (12) are normalized as
work.
qnf a2 If we integrate Eq. (16), we have
f gggg atgf ggg 3f gg Reffggg
f g f gg
f
Ha2 f gg 0 h i
lnf ggt 2
f ggg agf gg 2f g Re ffgg
f g
Ha2 f g co 18
14
where co is space-invariant parameter and the boundary conditions
with the boundary conditions as
are
g 0; f 0; f gg 0
g 0; f 0; f gg 0
g 1; f 1; f g 0 15
g 1; f 1; f g 0 19
_
q aat
Assuming at nfl remains constant during the flow process, In order to find the outer and inner expansions of Eq. (18) sub-
nf

andf f g, then f ggt 0 and Eq. (14) reduces to ject to the boundary conditions of Eq. (19), Eq. (18) is divided by
Re and Eq. (20) is obtained as
f gggg agf ggg 3f gg Reffggg
f g f gg
Ha2 f gg 0 16
1 a 2 Ha2 c
The boundary conditions still remain as Eq. (15). f ggg gf gg 2f g ffgg
f g
f o 20
Re Re Re g Re
If a = 0 in the above Eq. (16), the Bermans model [1] for chan-
nels with stationary walls is recovered. where 1
Re
 1.Taking e Re1 , Eq. (21) is arrived at as
ef ggg aegf gg 2f g ffgg
f g 2 eco 21
3. Approximate analytical methods of solutions
Eq. (21) can be written as
The main focus of this paper is to present comparative study of
1
 3

singular perturbation and differential transform methods. In order ef ggg aegf gg 2f g ffgg
f g 2
e
2 Ha2 e2 f g
to achieve this purpose, the developed Eq. (16) is solved using the h  3
 1 i
two methods as the procedures are given in the subsequent
k2 k Ha2 e2 e
2
2ake 22
sub-sections.
where
3.1. Method of matched asymptotic expansion
 3
 1
ec0 k2 k Ha2 e2 e
2
2ake
The method of matched asymptotic expansion (MMAE) is a gen- 1
 3
 1
 3

eralization of Prandtls boundary layer theory and is mostly used to Since e
2 Ha2 e2  1, the term containing a multiple of e
2 Ha2 e2
determine a uniform and accurate approximation to the solution of will not disappear in the limiting case as e ! 0:
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 75

The above problem in Eq. (22) is a singular perturbation which X

1
n1
f s 1 e 2 zn s 32
is solved by method of matched asymptotic expansion as the
n0
procedures are given as follows.
Substituting Eq. (32) into Eq. (30), one obtains
3.1.1. The outer solution
k0
z0 z_ 0

1
It is assumed that the outer solution take the form of e2 : 33a
r
1 3
f f 0 f 1e f 2e f 3e . . . 23

2 !
2 2

2
k1 k0 k0
and e :
1 z1 z_ 1
k0
e
rs
r se
rs
r r r
1 3
k k0 k1 e k2 e k3 e . . .
2 2 24
33b
The coefficients ki are to be determined by matching the outer
solutions with the inner solutions.
3
3
z2 z_ 2

k2 ak0
rs k0
On substituting Eq. (23) into Eq. (24), one arrives at (after e2 : e e
2rs
r r r
equating the same power of the coefficient e): 2 3 33c
k2
k1 k0 rrs
2
2r0 3r2 s2
2rs
2
e0 : 0 2
f 0

00
f 0f 0 k20 25a 4 3
5
rs
 e

2kr02k1 rs 1 12 kr0 r3 s3 r2 s2 2rs
2
1 0 0 00 00
e2 : 2f 0 f 1
f 0 f 1
f 0 f 1 2k0 k1 25b The boundary conditions for these inner solutions are given as:

e1 : 0 0 00 00 000
2f 0 f 2
f 2 f 0
f 2 f 0
f 0
agf 2f 0
f 0 f 1 f 1
00 0 2 00 z0 0 z1 0 z2 0 . . . zn 0 0;
0 z_ 0 0 z_ 1 0 z_ 2 0 . . . z_ n 0 0 34
Ha2 f 0 2k0 k2 k21 2ak1 25c
On solving Eqs. (33a), (33b) and (33c) using the boundary con-
3 0 0 00 00 000 00 00 00 0 00 ditions in Eq. (34), one arrived at
e :
2 2f 0 f 3
f 0 f 3
f 0 f 3
f 1
f 2 f 2
f 1 f 1 g ag
f 3f 0
00 00 00 0 0 0 0

f 0 f 3
f 1 f 2
f 2 f 1 2f 1 f 2 2f 0 f 3 Ha2 f 1 2 k1 a 2k1 k2 k0
z0 1
rs
e
rs 35a
25d r
Subject to the following boundary conditions
2
k0 k0
f 0 0 f 1 0 f 2 0 f 3 0 0 26 z1 1
rs k0
2
r r
(
2 )
00 00 00 00 k1 1 k0
f 0 0 f 1 0 f 2 0 f 3 0 0
k 1 rs
2 rs e
rs
2
35b
r 0 2 r
Using the above boundary conditions, it can easily be shown
that the solutions to Eqs. (25a), (25b), (25c) and (25d) are  3 k2
z2 kr2 1
rs 92 kr0 k1
k0 r
r0
4kr02k1 ark0
ark0 1 rse
rs
f 0 g k0 g; f 1 g k1 g; f 2 g k2 g; f 3 g k3 g 27 n o h i
k2

kr2 k1 1 rs 12 k0 rrs2
2rs
2 e
rs
2r0 rs3 2rs2 2rs 2
Substituting Eq. (27) into Eq. (23), one arrives at the outer solu- h i   h i
1 k0 3

kr0 k21 rs2 4rs 4 24 r 3rs4 16rs3 60rs2 96rs 120 e
rs
tion as  
1 k0 3
2rs
 1 3
 24 r e
f k0 k1 e2 k2 e k3 e2 . . . g 28 35c
The above Eq. (28) is the outer solution of Eq. (18) which is only Therefore, the second-order inner solution which is valid in the
valid in the region between the edge of the boundary layer and boundary layer region is given as
the center of the channel. Again, it should be stated that the coeffi- 8 8 k 2 9 9
>
> < r0 1
rs k0
2 r0
k = >
>
cients ki will be determined by matching the outer solutions with > k0
>
rs n o 1 >
>
>
> r 1
rs
e e2 >
>
>
> : k0 2

r k0 1 rs
2 r 2 rs e
k1 1 2
rs ; >
>
>
> >
>
the inner solutions. >
>
> 8 9 >
>
>
>
< > 9 k0 3 k20 ak0 ak0 > >
=
>
>
k2 4k0 k1
1
rs 2 r k1
k0 r
r
r2 r
r 1 rse
rs
>
>
1
f s 1 e2 >
> r
h i h >
i>
> < >
> 2 >
> >
>
>
f k2
k 1 rs 1
k r rs 2

2 rs
2 ge
rs


k0
rs 3
2rs2
2rs 2 = >
>
>
3.1.2. The inner solution e>
1 2 0
>
>
r 2r >
>
>
> > >
>
> >
> 1 k0 3
rs >
>
>
>
>
> >
2 4 3 2
In order to develop the inner solution, knowing that the viscous > > >
k0 k1
rs 4 rs 4 3rs 16 rs 60 rs 96rs 120e >
>
> > r2 24 r >
> >
>
: >
> :  
1 k0 3
2rs
>
; >
;
1 1 24 r e
layer is of the order of e2 , a stretching transformation s 1
ge
2
is introduced. This implies that 36
1
g 1
se 2 29
3.1.3. Matching criteria
Substituting Eq. (29) into Eq. (22), gives The constants or the coefficients ki in the outer solution can be
 
e2 a2z_ sz_
z1 ea e2 z_ z_ 2
zz
rz_ determined and the final complete solution can be found by
3 1

matching the outer solutions with the inner solutions at the edge
1 3
k2 e2 kr
2ake2 30 of the boundary layer i.e. by imposing continuity between the solu-
tions at different scales. There are different matching criteria or
and the inner boundary conditions as principles such as Prandtls, van Dyke and Johnson (a modified
z0 0; z_ 0 0 31 van Dykes matching principle) used in literatures to determine
the coefficients of in the outer solution and to find the final/com-
3
where r Ha2 e It should be noted that the dot shows that the
2 plete solution of the differential equation.
derivative is with respect to s.Following the boundary condi- The matching of the outer solution with the inner solution can
tionf g 1 in Eq. (19), assuming an inner solution of the form be carried out in different ways using different approaches.
76 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

i. Expressing the outer solution in terms of the inner variable s For x xi , then ux; p uxi ; p, where p belongs to the set of non-
and matching it with the inner solution as s ! 1 negative integers, denoted as the p-domain. Therefore Eq. (40) can
ii. Assuming that the inner and outer expansions are both valid be rewritten as
for intermediate times of the order e2  g  1, it is required
1
 p
d ux
that the expansions agree asymptotically in this regime, Up uxi ; p p 41
dt xxi
where s ! 1 and g ! 0 ase ! 0 . Hence, the matching cri-
teria/conditions are given as lims!1 f s limg!0 f g The Where U p is called the spectrum of ux at x xi
above equation means that the inner solution evaluated as If ux can be expressed by Taylors series, the ux can be rep-
the outer solution is approached is equal to the outer solu- resented as
tion evaluated as the inner solution is approached.
1 
X
iii. Also, in order to match the solution near y = 0, an intermedi- x
xi p
ate layer can be introduced Which is expressed in terms of
ux Up 42
p
p!
1
the variable y , where g wey , e < we 6 1, and requires 2

f outer
f inner
that lime!0;y fixed R 0; R 0; 1; 2 . . . where Eq. (42) is called the inverse of Uk using the symbol D
e2
iv. A similar procedure is followed to match the solution near denoting the differential transformation process and combining
g = 1. In this case, one introduces variabley , where Eq. (41) and Eq. (42), it is obtained that
g 1
wey ; and requires that lime!0;y fixed
1 
X
f outer
f inner
0; R 0; 1; 2 . . . x
xi p
e2
R
ux Up D
1 Up 43
p0
p!
The matching could also be done by using the fact that the outer
o
solution f g should tend to the value given by the inner solution
i
f gas g ! 1 3.2.1. Operational properties of differential transformation method
Using the first approach by expressing the outer solution in If ux and v x are two independent functions with space (x)
terms of the inner variable s i.e. where Up and Vp are the transformed function corresponding
1
to ux and v x , then it can be shown from the fundamental math-
f g k0 k1
k0 se2 k2
k1 se . . . 37 ematics operations performed by differential transformation that.
and matching the above Eq. (37) of the outer solution with the inner
solution as s ! 1 gives i. If zx ux  v x; then Zp Up  Vp
n
ii. If zx aux; then Zp aUp If zx d dxux n ; then
1 1
k0 1; k1 ; k2 1
38 Zp p 1 p 2p 3 . . . p nUp n
r r2 P
iii. If zx uxv x; then Zp pr0 VrUp
r
Therefore, the outer solution can be written as P


  iv. If zx um x , then Zp pr0 U m
1 rUp
r
P
1 1 1 v. If zx uxv x; then Zp pr0 r 1Vr 1Up
r
f g k0 e2 1
2 e . . . g 39 P
r r n
vi. If zx xm d dxux
n ; then Zp pl0 d l
m
1p
l 1
Hence, the complete solution is given by Eqs. (36) and (39). p
l 2p
l 3 . . . p
l nUp
l n
Since the inner solution contains an exponential term in g, it could d3 ux P
vii. If zx dux
dx dx3
; then Zp pl0 Up
ll 1
be stated that the inner solution given by Eq. (36) is valid even out-
l 2l 3Ul 3
side the boundary layer region. In the outside boundary layer d2 ux Pp
region, the contribution from the exponential term is negligible viii. If zx dux
dx dx2
; then Zp l0 p
l 1Up
l 1
and the inner solution reduces to a linear solution (in terms of l 1l 2Ul 2
g) which is equivalent to the outer solution. Therefore, the inner 2 Pp
ix. If zx dux
dx
then Zp l0 p
l 1Up
l 1
solution is valid throughout the region of 0 < g < 1 i.e. it is uni- l 1Ul 1
formly valid for the channel. Although, multiple solutions associ- Pp
x. If zx u dux ; then Zp
ll 1Ul 1
l0 Up
ated with this problem have been reported by Robinson [13], dx
2 2 Pp
Zaturska et al. [14] and Si et al. [15,16], this is not the focus of xi. If zx d dx
ux
2  ; then Zp l0 p
l 1p
l 2
the present study. Up
l 2l 1 l 2Ul 2

3.2. Differential transform method Using the above operational properties of DTM, the recursive
relation for Eq. (16) in DTM domain is given as
The differential transform method has widely been used to solve
X
p
both singular and non-singular perturbed boundary values prob- p 1p 2p 3p 4Fp 4 a dlp
l 1p
l 2p
l 3Fp
l 3
lems. It gives analytical solution to differential or integral solutions 2
l0
3
Xp
in the form of a polynomial by transforming each term in the differ- Fp
ll 1l 2l 3Fl 3
6 7
ential equation or integral into a recursive form or relation of the 6 l0 7
3ap 1p 2Fp 2 Re6
6 X p
7
7
equation which follows an iterative procedure for obtaining analyt- 4 5

p
l 1Fp
l 1l 1l 2Fl 2
ical series solutions of differential or integral equations. l0

The basic definitions of the method is as follows.
Ha2 p 1p 2Fp 2 0

If ux is analytic in the domain T, then it will be differentiated 44
continuously with respect to space x.
where
p

d ux 1 l1
p ux; p for all x 2 T 40 dl
dx 0 l0
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 77

and the recursive relations for the boundary conditions in Eq. (15) Using p = 0, 1, 2, 3. . . in the above recursive relations, we arrived
are at
F0 0; p 1p 2Fp 2 0 ) p 0; F2 0 F4 0
Xp F5 20
1
Ha2 k2
3ak1
Fp 1; ) p 1; F1 k1 ;
F6 0
l0 n o
Xp F7 840
1 2
Ha2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2
p 1Fp 1 0; ) p 2; F3 k2
F8 0
l0 8 9
45 < 2
Ha2 fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g =
F9 60480
1
: 2 2 2 ;

27a3 k2 108aRek2 54Rek1 k2 a2 48k1 k2 a2 48Rek1 k2 24Reak1 k0
Therefore, the following boundary conditions in DTM domain
are arrived at F10 0
:
F0 0; F2 0; F1 k1 ; F3 k2 46
:
where k1 and k2 are the constants which will be determined :
through the boundary conditions
48
1
Fp 4 Following the definition in Eq. (43), the solution of Eq. (16)
p 1p 2p 3p 4
8 9 using DTM is given as
>
> Xp
>
>
>
> Ha2 p 1p 2Fp 2
a dlp
l 1p
l 2p
l 3Fp
l 3 >>
>
> n o
>
> >
> f g k1 g k2 g3 20 Ha2 k2
3ak1 g5 840
2
Ha2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g7
> 3>
1 1
l0
>
> 2 p >
> ( )
< X = 2
Ha2 fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g
6 Fp
ll 1l 2l 3Fl 3 7 60480
1
g9 ...
>
> 6 l0 7>> 2 2

27a3 k2 108aRek2 54Rek1 k2 a2 48k1 k2 a2 48Rek1 k2 24Reak1 k0
2
> 7>
>
3ap 1p 2Fp 2
Re6
>
> 6 X 7>>
>
>
>
> 4
p
5 >
>
>
49
>
:
p
l 1Fp
l 1l 1l 2Fl 2 >;
l0
In order to find constants k1 and k2, recall from Eq. (15) that at
47 g 1; f 1; f 0 g 0.

Table 1
Comparison of results of flow for large Reynolds number, injection and suction.

F Injection Re = 100 Ha = 0 Suction Re =
100 Ha = 0

g NM MMAE DTM NM MMAE DTM
0.0 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.1 0.16377 0.16414 0.16378 0.10117 0.10116 0.10117
0.2 0.32193 0.32254 0.32193 0.20234 0.20232 0.20234
0.3 0.46995 0.47078 0.46994 0.30351 0.30349 0.30349
0.4 0.60424 0.60523 0.60424 0.40468 0.40465 0.40467
0.5 0.72190 0.72292 0.72190 0.50585 0.50581 0.50584
0.6 0.82063 0.82154 0.82061 0.60702 0.60697 0.60702
0.7 0.89871 0.89938 0.89870 0.70819 0.70813 0.70818
0.8 0.95498 0.95535 0.95497 0.80937 0.80930 0.80938
0.9 0.98878 0.98890 0.98878 0.91053 0.91045 0.91053
1.0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

Table 2
Comparison of results of skin friction for large Reynolds number with suction.
00
f 1 Suction with a = 0
Re Ha NM MMAE DTM

12.957 6.445
4.78
4.70
4.78

18.638 6.103
4.07
4.16
4.07

25.747 7.151
4.11
4.24 -4.11

41.818 11.419
5.10
5.14
5.10

50.460 9.964
4.16
4.36
4.16

62.485 11.077
4.17
4.39
4.17

76.326 12.233
4.18
4.42
4.18

Table 3
Comparison of results of skin friction for large Reynolds number with injection.
00
f 1 Injection witha=0
Re Ha NM MMAE DTM
60.175 7.825
59.90
60.16
59.90
54.265 7.438
54.10
54.25
54.10
39.380 6.370
39.26
39.46
39.26
33.946 5.923
33.54
33.93
33.54
23.006 4.924
22.59
23.01
22.59
18.154 4.879
17.95
18.46
17.95
78 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

Table 4
Comparison of results for small Reynolds number with injection, suction, expansion and contraction.

f Contraction and Suction R =
0.5, a=
1, Ha = 0 Contraction and Injection R = 0.5,a=
1, Ha = 0

g NM RPM DTM NM RPM DTM
0.0 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.2 0.27815 0.29360 0.27814 0.28068 0.29460 0.28067
0.4 0.54007 0.56250 0.54007 0.54428 0.56410 0.54428
0.6 0.76679 0.78250 0.76679 0.77097 0.78410 0.77097
0.8 0.93279 0.93010 0.93278 0.93490 0.93090 0.93490
1.0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

Table 5
Comparison of results for small Reynolds number with injection, suction, expansion and contraction.

F Contraction and Suction R =
1, a=
1, Ha = 0 Contraction and Injection R = 1, a=
1, Ha = 0

g NM RPM DTM NM RPM DTM
0.0 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.2 0.27678 0.29310 0.27677 0.28184 0.29510 0.28183
0.4 0.53779 0.56160 0.53778 0.54621 0.56500 0.54620
0.6 0.76449 0.78170 0.76448 0.77287 0.78500 0.77286
0.8 0.93161 0.92970 0.93162 0.93583 0.93130 0.93584
1.0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

Table 6
Comparison of results for small Reynolds number with injection, suction, expansion and contraction.

f Expansion and Suction R =
0.5, a = 1, Ha = 0 Expansion and Injection R = 0.5, a = 1, Ha = 0

g NM RPM DTM NM RPM DTM
0.0 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.2 0.31707 0.29170 0.31707 0.31609 0.29880 0.31609
0.4 0.59972 0.57140 0.59972 0.59818 0.57410 0.59818
0.6 0.81886 0.79940 0.81885 0.81747 0.80190 0.81747
0.8 0.95526 0.95690 0.95525 0.95430 0.95800 0.95430
1.0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

Table 7
Comparison of results for small Reynolds number with injection, suction, expansion and contraction.

f Expansion and Suction R =
1, a = 1, Ha = 0 Expansion and Injection R = 1, a = 1, Ha = 0

g NM RPM DTM NM RPM DTM
0.0 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000
0.2 0.31765 0.29620 0.31764 0.31567 0.29960 0.31568
0.4 0.60063 0.57000 0.60062 0.59753 0.57540 0.59754
0.6 0.81968 0.79820 0.81967 0.81689 0.80320 0.81689
0.8 0.95526 0.95630 0.95525 0.95405 0.95860 0.95406
1.0 1.00000 1.00000 1.00000 1.00000 1.00000 1.00000

Table 8
Comparison of results of skin friction for large Reynolds number with suction.

f g Suction with expansion, Re =
1.0, a=0.1

g NM HPM [35] HAM [35] DTM
0.00 0.000000000 0.000000000 0.000000000 0.000000000
0.05 0.075104609 0.075104330 0.075104609 0.075104609
0.10 0.149826629 0.149826076 0.149826628 0.149826628
0.15 0.223783842 0.223783029 0.223783841 0.223783842
0.20 0.296594783 0.296593728 0.296594781 0.296594780
0.25 0.367879067 0.367877796 0.367879065 0.367879065
0.30 0.437257704 0.437256246 0.437247701 0.437247702
0.35 0.504353340 0.504351733 0.504353337 0.504353337
0.40 0.568790457 0.568788740 0.568790454 0.568790454
0.45 0.630195472 0.630193690 0.630195468 0.630195470
0.50 0.688196757 0.688194959 0.688196753 0.688196755
0.60 0.792510690 0.792509019 0.792510687 0.792510686
0.70 0.878791325 0.878789990 0.878791323 0.878791324
0.80 0.944107888 0.944107058 0.944107887 0.944107886
0.90 0.985512871 0.985512584 0.985512872 0.985512870
1.00 1.000000000 1.000000000 1.000000000 1.000000000
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 79

Fig. 2. Variation of f (g) for different expansion and contraction ratio, a and different small values of Re.
80 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

Therefore, using the boundary conditions, Eqs. (50) and (51) are
written as
2
f 1 k1 k2 20
1
Ha2 k2
3ak1 g5 840
1
fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g
( 2
)
Ha2 fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g
60480
1
2 2 2
. .. 1

27a3 k2 108aRek2 54Rek1 k2 a2 48k1 k2 a2 48Rek1 k2 24Reak1 k0
50

0 2
f 1 k1 3k2 14 Ha2 k2
3ak1 120 1
fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g
( 2 )
Ha fHa 2
Ha 2
k2
3ak 1 9a2 k2 12Rek22 6Rek1 k2 a2 g
6720
1
2 2 2
. .. 0

27a k2 108aRek2 54Rek1 k2 a 48k1 k2 a 48Rek1 k2 24Reak1 k0
3 2 2

51
On solving the above Eq. (50) and (51), different values for k1
and k2 for respective different values of Re and a were obtained.
For the purpose of comparison of results, the second-order
derivatives of f (g) at the wall, which represents the skin friction
is developed as
00 2
f 1 6k2 Ha2 k2
3ak1 20 1
fHa2 Ha2 k2
3ak1 9a2 k2 12Rek2 6Rek1 k2 a2 g
( 2
)
2 2 2
Ha fHa Ha k
3 a k 1 9a k2 12Rek2 6Rek1 k2 a g
2 2
840
1 2
. .. Fig. 3. Variation of f (g) for different expansion and contraction ratio, a and large
2 2 2

27a3 k2 108aRek2 54Rek1 k2 a2 48k1 k2 a2 48Rek1 k2 24Reak1 k0 value of Re.
52

4. Results and discussion Re, is no longer large and the error between the NM and MMAE
becomes increasing large when Re is not large enough for the per-
In order to verify the two approximate analytical methods used turbation solution. For the large injection case of Re 500, the
in this work, the fourth order ordinary differential equation is also solution becomes much less sensitive to a. While there is no addi-
solved numerically using shooting method coupled with Runge- tional computational cost in the determination of the coefficients
Kutta method. For the purpose of demonstrating the efficiency or constant of the outer solution, the determination of the
and accuracy of MMAE, RPM and DTM, the obtained results for unknown constants in the solutions of DTM necessitated the use
the velocity distribution for different permeation Reynolds number of software which shows that the cost of computation of approxi-
and non-dimensional wall dilation rate are compared with the mate analytical solution using PM is lower than using DTM. How-
solutions of the numerical method (NM) and some other approxi- ever, the solution of DTM is stable and reliable for all cases of Re
mate analytical methods used in the past published works for the and a as it does not require any small parameter in the nonlinear
flow problem as shown in Tables 18. differential equation. Unlike the perturbation methods, the DTM
The tables present the comparison of results for small and large does not depend on any small physical parameters. Thus, it is valid
values of Reynolds number for both weak injection and suction. for both weakly and strongly nonlinear problems. This shows that
Also, for small weak injection or suction i.e. for small value of Rey- the limitations of the traditional perturbation methods have been
nolds number, the discrepancy between the results of the numer- eliminated through the use of DTM. Also, DTM reduces the size
ical method (NM) and that of regular perturbation method (RPM) of computational domain. Comparing the solution of DTM with
decreases as the net permeation Reynolds decreases. i.e. the accu- the numerical solutions shows that the obtained solutions through
racy of the RPM increases or proportional to the smallness of Rey- the DTM demonstrate remarkable accuracy with the numerical
nolds number. It is also established that the solution of the RPM is solutions. Therefore, it can be deduced that differential transform
a fair approximation to the solution of the NM for
5  Re  5. The method is more precise and accurate than the MMAE and RPM.
asymptotic error increases as Re increases. Combining injection This accuracy gives high confidence about validity of DTM in pro-
(positive Re) with wall contraction (negative a) gives a smaller viding solutions to the problems. The high accuracy of the DTM
effective Reynolds number due to the doubly perturbed nature of for the analysis of problem reveals an excellent agreement of engi-
the asymptotic solution, a smaller error i.e. expected when the sec- neering accuracy and precision.
ondary perturbation parameter, a is much smaller than the pri- Fig. 2 show the effects of the permeation Reynolds number and
mary perturbation parameter |Re|. Therefore, an improved non-dimensional wall dilation rate on the dimensionless flow
approximation is achieved as expected at smaller values of |a|. velocities. Figs. 2e, f, g, h, 5e, 6a and b show the effects of Reynolds
Additionally, the approximate analytical solutions using RPM are number, Re on the velocity at constant non-dimensional wall dila-
practically equivalent to the exact solution for sufficiently small | tion rate and on the dimensionless axial velocity. Increase in the
Re| and |a|. It should be pointed out that the asymptotic solutions Reynolds number decreases the axial velocity at the center of the
deteriorate when the expansion or contraction ratio, a is of the channel during the expansion while the axial velocity increases
same order as the permeation Reynolds number, Re. i.e. as a slightly near the surface of the channel when the wall contracts
approaches Re, the approximate analytical solution becomes less at the same rate. The behavior of axial velocity for different perme-
reliable and the regular perturbation solution breaks down when ation Reynolds number, over a range of non-dimensional wall dila-
the perturbation parameter, Re becomes large. It also breaks downs tion rate are shown in Figs. 2ad, 3, 4af, 5ad. The figures depict
when second term in the asymptotic solution is of the same mag- that, for every level of injection or suction, the velocity is maxi-
nitude with the first term or when the third term in the solution is mum at the center of the channel and near the point, the velocity
quantitatively the same as the second term. It was observed that is increased when the channel is expanding and decrease when the
the analytical solution given by the method of matched asymptotic channel contracts. That is because the flow toward the center
expansion (MMAE) breaks down when the perturbation parameter, becomes greater to make up for the space caused by the expansion
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 81

Fig. 4. Variation of f0 (g) for different expansion and contraction ratio, a and different small values of Re.

of the wall and as a result the axial velocity also becomes greater the axial velocity near the wall because the flow toward the wall
near the center. Such behavior is evident in the transport of biolog- becomes greater and as a result the axial velocity near the wall
ical fluids through contracting or expanding vessels, filtration in becomes greater. So, both the expansion and suction through the
kidneys and lungs, flow inside lymphatics and the synchronous wall reinforce the flow through the channel and similarly does
pulsation of porous diaphragms. It could also be seen from the fig- the wall contraction and injection through the surface. The results
ures, that as the wall expansion ratio increases, the velocity at the of the present study show that for every level of injection or suc-
center decreases and increases near the wall. Similarly, for the case tion, in the case of expanding wall, increasing a(t) leads to increase
of contracting wall as shown in Figs. 5ad and 6a, increasing con- in axial velocity near the center and decrease in the axial velocity
traction ratio decreases axial velocity near the center and increases near the wall.
82 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

Fig. 5. Variation of f0 (g) for different expansion and contraction ratio, a and different small values of Re.

5. Conclusion walls with large injection or suction under the influence of trans-
verse magnetic field has been carried out. Also, the effects of the
In this work, a comparative study of two approximate analytical permeation Reynolds number, the wall expansion ratio a and mag-
methods for the analysis of two-dimensional unsteady flow of netic parameter on the flow behavior of the fluid have been ana-
nanofluid in a porous channel through expanding/contracting lyzed and investigated. From the results, it was established that
 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184 83

Fig. 6. Variation of f0 (g) for different expansion and contraction ratio, a and different values of Re.

increase in the Reynolds number decreases the axial velocity at the [14] M.B. Zaturska, P.G. Drazin, W.H.H. Banks, On the flow of a viscous fluid driven
along a channel by suction at porous walls, Fluid Dyn. Res. 4 (3) (1988) 151
center of the channel during the expansion while the axial velocity
178.
increases slightly near the surface of the channel when the wall [15] X.H. Si, L.C. Zheng, X.X. Zhang, Y. Chao, Existence of multiple solutions for the
contracts at the same rate. Also, as the wall expansion ratio laminar flow in a porous channel with suction at both slowly expanding or
increases, the velocity at the center decreases and increases near contracting walls, Int. J. Miner. Metal. Mater. 11 (2011) 494501.
[16] X.H. Si, L.C. Zheng, X.X. Zhang, M. Li, J.H. Yang, Y. Chao, Multiple solutions for
the wall. For every level of injection or suction, in the case of the laminar flow in a porous pipe with suction at slowly expanding or
expanding wall, increasing a(t) leads to increase in axial velocity contracting wall, Appl. Math. Comput. 218 (2011) 35153521.
near the center and decreases in axial velocity near the wall. The [17] J. Majdalani, C. Zhou, C.A. Dawson, Two-dimensional viscous flow between
slowly expanding or contracting walls with weak permeability, J. Biomech. 35
results of the method of matched asymptotic expansion and differ- (10) (2002) 13991403.
ential transform method were compared with the numeral solu- [18] S. Dinarvand, A. Doosthoseini, E. Doosthoseini, M.M. Rashidi, Comparison of
tions. The solutions of the differential transform method HAM and HPM methods for Bermans model of two-dimensional viscous flow
in porous channel with wall suction or injection, Adv. Theor. Appl. Mech. 1 (7)
obtained in comparison with the numerical solutions showed (2008) 337347.
remarkable accuracy and better agreements than the solutions of [19] M. Sheikholeslamia, D.D. Ganji, M.M. Rashidi, Magnetic field effect on
method of matched asymptotic expansion. Therefore, the differen- unsteady nanofluid flow and heat transfer using Buongiorno model, J. Magn.
Magn. Mater. 416 (2016) 164173.
tial transform method gives more accurate results than method of [20] M. Sheikholeslami, D.D. Ganji, Nanofluid flow and heat transfer between
matched asymptotic expansion. parallel plates considering Brownian motion using DTM, Comput. Methods
Appl. Mech. Eng. 283 (2015) 651663, using DTM-FDM, Journal of Central
South University 24 (3), 675682.
[21] M. Sheikholeslamia, D.D. Ganji, M.M. Rashidi, Magnetic field effect on
References unsteady nanofluid flow and heat transfer using Buongiorno model, J. Magn.
Magn. Mater. 416 (2016) 164173.
[1] A.S. Berman, Laminar flow in channels with porous walls, J. Appl. Phys. 24 [22] M. Sheikholeslamia, M.D.D. Ganji, M.M. Rashidi, Magnetic field effect on
(1953) 12321235. unsteady nanofluid flow and heat transfer using Buongiorno model, J. Magn.
[2] R.M. Terrill, Laminar flow in a uniformly porous channel, Aeronaut. Q. 15 Magn. Mater. 416 (2016) 164173.
(1964) 299310. [23] M. Sheikholeslami, D.D. Ganji, Nanofluid hydrothermal behavior in existence
[3] R.M. Terrill, Laminar flow in a uniformly porous channel with large injection, of Lorentz forces considering Joule heating effect, J. Mol. Liq. 224 (Part A)
Aeronaut. Q. 16 (1965) 323332. (2016) 526537.
[4] E.C. Dauenhauer, J. Majdalani, Exact self-similarity solution of the Navier- [24] M. Sheikholeslami, D.D. Ganji, Transportation of MHD nanofluid free
Stokes equations for a porous channel with orthogonally moving walls, Phys. convection in a porous semi annulus using numerical approach, Chem. Phys.
Fluids 15 (6) (2003) 14851495. Lett. 669 (2017) 202210.
[5] J. Majdalani, The oscillatory channel flow with arbitrary wall injection, [25] S. Khalili, S. Dinarvand, R. Hosseini, H. Tamim, I Pop Unsteady MHD flow and
Zeitschrift fur Angewandte Mathematik und Physik 52 (1) (2001) 3361. heat transfer near stagnation point over a stretching/shrinking sheet in porous
[6] J. Majdalani, T.-S. Roh, The oscillatory channel flow with large wall injection, medium filled with a nanofluid, Chin. Phys. B 23 (4) (2014) 048203.
Proc. R. Soc. London Ser. A 456 (1999) (2000) 16251657. [26] S. Khalili, H. Tamim, A. Khalili, MM Rashidi Unsteady convective heat and mass
[7] J. Majdalani, W.K. van Moorhem, Multiple-scales solution to the acoustic transfer in pseudoplastic nanofluid over a stretching wall, Adv. Powder
boundary layer in solid rocket motors, J. Propul. Power 13 (2) (1997) 186193. Technol. 26 (5) (2015) 13191326.
[8] L. Oxarango, P. Schmitz, M. Quintard, Laminar flow in channels with wall [27] S. Khalili, A. Khalili, S. Kafashian, A. Abbassi, Mixed convection on a permeable
suction or injection: a new model to study multi-channel filtration systems, stretching cylinder with prescribed surface heat flux in porous medium with
Chem. Eng. Sci. 59 (5) (2004) 10391051. heat generation or absorption, J. Porous Media 16 (11) (2013).
[9] T.A. Jankowski, J. Majdalani, Symmetric solutions for the oscillatory channel [28] S. Dinarvand, S. Khalili, R. Hosseini, E. Damangir, I.R. Dehkordi, I. Pop, MHD
flow with arbitrary suction, J. Sound Vib. 294 (4) (2006) 880893. flow and heat transfer over a nonlinearly stretching sheet in porous
[10] A. Jankowski, J. Majdalani, Laminar flow in a porous channel with large wall medium filled with a nanofluid, Special Topics Rev. Porous Media: Int. J. 5
suction and a weakly oscillatory pressure, Phys. Fluids 14 (3) (2002) 1101 (1) (2014).
1110. [29] H. Tamim, S. Dinarvand, R. Hosseini, S. Khalili, A. Khalili, Mixed convection
[11] Zhou, J. Majdalani, Improved mean-flow solution for slab rocket motors with boundary-layer flow of a nanofluid near stagnation-point on a vertical plate
regressing walls, J. Propul. Power 18 (3) (2002) 703711. with effects of buoyancy assisting and opposing flows, Res. J. Appl. Sci. Eng.
[12] J. Majdalani, C. Zhou, Moderate-to-large injection and suction driven channel Technol. 6 (2013) 17851793.
flows with expanding or contracting walls, Zeitschrift fur Angewandte [30] J. Xu, Z.L. Lin, S.J. Liao, J.Z. Wu, J. Majdalani, Homotopy based solutions of the
Mathematik und Mechanik 83 (3) (2003) 181196. Navier-Stokes equations for a porous channel with orthogonally moving walls,
[13] W.A. Robinson, The existence of multiple solutions for the laminar flow in a Phys. Fluids 22 (2010) 053601.
uniformly porous channel with suction at both walls, J. Eng. Math. 10 (1976) [31] S. Dinarvand, M.M. Rashidi, A reliable treatment of a homotopy analysis
2340. method for two dimensional viscous flow in a rectangular domain bounded by
84 M.G. Sobamowo / Thermal Science and Engineering Progress 4 (2017) 7184

two moving porous walls, Nonlinear Anal.: Real World Appl. 11 (3) (2010) [34] J.K. Zhou, Differential Transformation and Its Applications for Electrical
15021512. Circuits, Huazhong University Press, Wuhan, China, 1986 (in Chinese).
[32] M.G. Sobamowo, Thermal analysis of longitudinal fin with temperature- [35] E. Rahimi, A. Rahimifar, R. Mohammadyari, I. Rahimipetroudi, M. Rahimi-Esbo,
dependent properties and internal heat generation using Galerkins method of Analytical approach for solving two-dimensional laminar viscous flow
weighted residual, Appl. Therm. Eng. 99 (2016) 13161330. between slowly expanding and contracting walls, Ain Shams Eng. J. 7 (2016)
[33] M.K. Parul, Methods for solving singular perturbation problems arising in 10891097.
science and engineering, Math. Comput. Modell. 54 (12) (2011) 556575.