Computers & Fluids 38 (2009) 1895–1902

Contents lists available at ScienceDirect

Computers & Fluids

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m p fl u i d

Numerical study of asymmetric laminar flow of micropolar fluids in

a porous channel
Muhammad Ashraf a,*, M. Anwar Kamal a,b, K.S. Syed a
a
Centre for Advanced Studies in Pure and Applied Mathematics, Bahauddin Zakariya University, Multan, Pakistan
b
Department of Mathematics, King Saud University, Riyadh, Saudi Arabia

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

Article history: A numerical study is presented for the two-dimensional flow of a micropolar fluid in a porous channel.
Received 29 July 2008 The channel walls are of different permeability. The fluid motion is superimposed by the large injection
Received in revised form 23 April 2009 at the two walls and is assumed to be steady, laminar and incompressible. The micropolar model due to
Accepted 23 April 2009
Eringen is used to describe the working fluid. The governing equations of motion are reduced to a set of
Available online 3 May 2009
non-linear coupled ordinary differential equations (ODEs) in dimensionless form by using an extension of
Berman’s similarity transformations. A numerical algorithm based on finite difference discretization is
employed to solve these ODEs. The results obtained are further improved by Richardson’s extrapolation
for higher order accuracy. Comparisons with the previously published work are performed and are found
to be in a good agreement. It has been observed that the velocity and microrotation profiles change from
the most asymmetric shape to the symmetric shape across the channel as the parameter R or the perme-
ability parameter A are varied between their extreme values. The results indicate that larger the injection
velocity at a wall relative to the other is, smaller will be the shear stress at it than that at the other. The
position of viscous layer has been found to be more sensitive to the permeability parameter A than to the
parameter R. The micropolar fluids reduce shear stress and increase couple stress at the walls as com-
pared to the Newtonian fluids.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction and solved the problem of flow of a viscous incompressible fluid

driven along a channel with porous uniformly accelerating walls.
Laminar flows in permeable walls have applications in the fields The above studies were carried out for Newtonian fluids.
of binary gas diffusion, filtration, ablation cooling, surface sublima- Eringen [9] for the first time developed the theory of microfl-
tion, microfluidic devices, grain regression (as in the case of com- uids, which deals with class of fluids, which exhibit certain micro-
bustion in rocket motors) and the modeling of air circulation in scopic effects, arising from local structure and micromotions of
the respiratory system. fluid elements. These fluids can support stress moments and body
The two-dimensional laminar steady flow of Newtonian fluids couples and are influenced by the spin inertia. After two years,
through a porous channel was investigated by Berman [1], which Eringen [10] introduced micropolar fluids, which is a subclass of
was driven by suction or injection. This exact solution was devel- microfluids. The micropolar fluids are those, which consist of
oped and generalized by Cox [2], Brady [3], Terrill and Shrestha bar-like elements. For example, the micropolar model has been
[4,5], Terrill [6], Waston et al. [7] and Robinson [8]. Cox [2] exam- used to describe the flow of liquid crystals, which are made up of
ined the flow by considering channel with one porous wall and dumbbell molecules. Other examples of such fluids are polymeric
other non-porous but was accelerating. Terrill and Shrestha [5] con- fluids and real fluids with suspensions, muddy fluids and fluids
sidered the laminar flow through parallel and uniformly porous with liquid crystals. The micropolar fluids have applications in
walls of different permeability. They calculated the shear stresses blood flow, lubricants, porous media, turbulent shear flows and
at the two walls to compare their series and numerical solutions. flow in capillaries and micro channels. Guram and Anwar [11] con-
The complete solution of the laminar flow of a fluid in a uniformly sidered the problem of steady, laminar and incompressible flow of
porous channel with large injection was given by Terrill [6]. He em- a micropolar fluid due to a rotating disk with uniform suction and
ployed the method of inner and outer expansions to solve the prob- injection. The equations of motion were reduced to dimensionless
lem and included the viscous layer. The resulting series solutions form and then solved by Gauss–Siedel iterative procedure with
were confirmed by numerical results. Waston et al. [7] formulated Simpson’s rule. Kelson and Desseaux [12] considered the two-
dimensional boundary layer flow of a micropolar fluid caused by
* Corresponding author. Tel.: +92 300 7354190. a moving plate or a stretching sheet and showed that a complete
E-mail address: [email protected] (M. Ashraf). similarity solution is only possible if the boundary condition is a

0045-7930/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compfluid.2009.04.009
1896 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902

linear function of the distance from the leading edge measured @ q=@t þ r  ðqVÞ ¼ 0; ð2Þ
from the position of the dye. They presented a solution using a ðk þ 2l þ jÞrðr  VÞ  ðl þ jÞr  r  V þ jr  t  rp
power-series expansion of the velocity and microrotation func- _
þ qf ¼ qV; ð3Þ
tions taking the vortex viscosity as a parameter. The numerical
solution of the steady, laminar, incompressible and two-dimen- ða þ b þ cÞrðr  tÞ  cðr  r  tÞ þ jr  V  2jt þ ql
sional flow of a micropolar fluid between two porous coaxial disks ¼ qjt_ : ð4Þ
was examined by Anwar Kamal et al. [13]. They used the similarity
transformation similar to that of Berman [1]. The finite difference In these equations V is the fluid velocity vector, t the microrotation,
equations were solved using SOR method. The results were further q the density, p the pressure, f and l are body force and body couple
refined by Richardson’s extrapolation method. per unit mass, respectively, j the microinertia, k; l; a; b; c; j, the
In the present work the numerical solutions of the steady, lam- material constants (or viscosity coefficients) and the dot signifies
inar and incompressible flow of a micropolar fluid through parallel material derivative.
and uniformly porous walls of different permeability are investi- For the two-dimensional problem under consideration, there
gated. The main flow is superimposed by the large injection at will be only two non-zero velocity components, the streamwise
the two porous walls. The transformed ODEs are discretized using velocity u in the x-direction and the normal velocity v in the y-
the central finite differences. The formulation of the problem is direction both depending on x and y. We may, therefore, express
given in Section 2. Section 3 contains the details of numerical algo- the velocity and microrotation vectors in the form
rithm, which we used to solve the problem. Finally, the results are V ¼ ðuðx; yÞ; v ðx; yÞ; 0Þ;
discussed and presented in tabular as well as graphical form in Sec- ð5Þ
t ¼ ð0; 0; /ðx; yÞÞ:
tion 4. A comparison of some results of micropolar and Newtonian
fluids is also given for the validation of our results. Using (5) the equations of motion (2)–(4) for our problem take the
form

2. Problem formulation @u=@x þ @ v =@y ¼ 0; ð6Þ

 1=q@ p=@x ¼ u@u=@x þ v @u=@y
Consider a steady, laminar and incompressible flow of a micro-
þ ðl þ jÞ=qð@ 2 v =@y@x  @ 2 u=@y2 Þ  j=q@/=@y; ð7Þ
polar fluid through a channel having porous walls of different per-
meability. The flow is driven by a constant pressure gradient with
 1=q@ p=@y ¼ u@ v =@x þ v @ v =@y
uniform entrance velocity U and is superimposed by the injection þ ðl þ jÞ=qð@ 2 u=@y@x  @ 2 v =@x2 Þ þ j=q@/=@x; ð8Þ
at the two walls. The effects of body forces and body couples are 2
qjðu@/=@x þ v @/=@yÞ ¼ cð@ /=@x þ @ /=@y Þ
2 2 2
neglected. We further assume that the flow is fully developed. Un-
þ jð@ v =@x  @u=@yÞ  2j/: ð9Þ
der these assumptions, the problem becomes two-dimensional and
interest lies in the investigations of flow characteristics at any The boundary conditions for the problem under consideration
cross section of the channel. may be expressed as
The velocity components u and v are taken to be in the direc-
tions of x- and y-axes, respectively, as shown in Fig. 1. The lower uðx; dÞ ¼ 0; uðx; dÞ ¼ 0;
wall is located at y = d and the upper at y = d so that the channel v ðx; dÞ ¼ V 1 ; v ðx; dÞ ¼ V 2 ; ð10Þ
width is 2d. Fluid is injected with constant velocity V1 at the lower /ðx; dÞ ¼ 0; /ðx; dÞ ¼ 0:
wall and with constant velocity V2 at the upper wall, where
|V2| > |V1|. In order to study the effect of different permeability of where V1 and V2 are injection velocities at the lower and upper
both the porous walls due to different injection velocities, we de- walls, respectively. It may be noted that V1 will be positive while
fine the permeability parameter as V2 will be negative. In order to obtain the velocity field and the
microrotation field for our problem, we have to solve (6)–(9) subject
A ¼ 1  V 1 =V 2 : ð1Þ to the appropriate boundary conditions (10). For this purpose, we
This parameter gives relative injection velocities at the porous use the following similarity transformations, which are proposed
walls. in view of those suggested by Berman [1] to convert the governing
The general equations governing the motion of the micropolar partial differential Equations (PDEs) (6)–(9) into ODEs.
fluid as given by Eringen [10] may be expressed as: Wðx; yÞ ¼ ðUd=A  V 2 xÞf ðgÞ; ð11Þ
Uðx; yÞ ¼ ðU=A  V 2 x=dÞgðgÞ=d; ð12Þ
where U is a constant velocity as described above in the problem
y-axis statement and g = y/d.
The velocity components in terms of the stream function W are

u ¼ @ W=@y ¼ ðU=A  V 2 x=dÞf 0 ðgÞ and v ¼ @ W=@x ¼ V 2 f ðgÞ:

ð13Þ
V2 V2 V2 V2
The above proposed velocity field identically satisfies the conti-
nuity Eq. (6) and, therefore, represents possible fluid motion. Since
the pressure term is eliminated from the governing equations, as
0 x-axis we shall see later, the number of unknowns is reduced by one
and we do not need to consider the continuity equation any more.
Using (11)–(13) in (7)–(9), it is found that
2 2
V1 V1 V1 V1  ððl þ jÞ=d Þf 000 þ jg 0 =d þ ð1=ðU=A  V 2 x=dÞÞ@ p=@x
00
Fig. 1. Geometry of the porous channel.
¼ qV 2 =dðf 02  ff Þ; ð14Þ
 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902 1897

2 2 0
ððl þ jÞV 2 =d Þf 00  jV 2 g=d  @ p=@y ¼ qV 22 ff =d; ð15Þ tions for the derivatives. This results into the following difference
00 2 00 2 0 0 equations
cg þ jd f þ 2jd g ¼ qjdV 2 ðf g  fg Þ: ð16Þ
2
ð2 þ Rhfn Þpn1  ð4  2Rh pn Þpn þ ð2  Rhfn Þpnþ1
The prime denotes differentiation with respect to g.
2
Boundary conditions (10) in view of (13) are reduced to the fol- ¼ hc1 ðg n1  g nþ1 Þ þ 2Bh ; ð25Þ
lowing dimensionless form. 2 2
ð2  c3 hfn Þg n1  ð4 þ 4c2 h þ 2c3 h pn Þg n þ ð2 þ c3 hfn Þg nþ1
3
f 0 ð1Þ ¼ 0; f 0 ð1Þ ¼ 0; ¼ c2 hðpnþ1  pn1 Þ; ð26Þ
7
f ð1Þ ¼ 1  A; f ð1Þ ¼ 1; 5 ð17Þ
where h represents the grid length, fn  f ðgn Þ; pn  pðgn Þ and
gð1Þ ¼ 0; gð1Þ ¼ 0: g n  gðgn Þ.
After eliminating the pressure term from Eqs. (14) and (15) we The algebraic system of equations represented by the above
obtain equations is solved iteratively by SOR method, Hildebrand [17],
subject to the appropriate boundary conditions (24).
f ðiv Þ  c1 g 00 þ Rðf 0 f 00  ff Þ ¼ 0:
000
ð18Þ In order to accelerate the iterative procedure and to improve
the accuracy of the solution, the solution procedure is mainly
Integration of Eq. (18) with respect to g yields
based on the algorithm described in [18]. A brief description of
00
f 000  c1 g 0 þ Rðf 02  ff Þ ¼ B; ð19Þ the procedure is given below.
Iterative procedure: For a suitable choice of the values of the grid
where B is a constant of integration.Eq. (16) can be rewritten as
size h and the relaxation parameter x, an iterative procedure is
0
g 00 þ c2 ðf 00 þ 2gÞ ¼ c3 ðf 0 g  fg Þ: ð20Þ started with some initial guess for the values of the constant of
integration B and the solution vectors p, g and f, where kth iteration
In the above equations, R is the parameter based on the injec-
performs the following steps:
tion velocity at the walls, and c1, c2 and c3 are dimensionless micro-
polar material properties characterizing the vortex viscosity, the
1. A new approximation for the solution of the finite difference
spin gradient viscosity and the microinertia density, respectively.
Eqs. (25) and (26), p(k+1) and g(k+1), respectively, is generated
These quantities are defined as
by SOR method [17] subject to the boundary conditions
2
R ¼ qV 2 d=ðl þ jÞ; c1 ¼ j=ðl þ jÞ; c2 ¼ jd =c imposed on p and g in (24).
and c3 ¼ qjV 2 d=c: 2. A new approximation for the solution of the differential Eq.
(21), f (k+1), is computed by Simpson’s rule [15] subject to the
Since R involves V2, therefore R < 0 for injection at the upper wall. first boundary condition given in (24), where p(k+1) is employed
We observe that the Eq. (19) reduces to the corresponding gov- for p in Eq. (25).
erning equation for the Newtonian fluid as given in [4–6] whereas 3. P (k+1), g (k+1) and f (k+1) are compared with p(k), g(k) and f (k),
the Eq. (20) becomes identically zero for vanishing microrotation respectively, to test for the convergence.
and j = 0.This validates our model for the micropolar fluid motion.
The iterative procedure is stopped if the following criteria are
3. Numerical solution satisfied for three consecutive iterations:

The governing equations (19) and (20) being highly non-linear kpðkþ1Þ  pðkÞ kL2 < TOLiter ; ð27Þ
cannot be solved analytically. We use a finite difference based ðkþ1Þ ðkÞ
kf f kL2 < TOLiter ; ð28Þ
numerical algorithm to solve this coupled pair of equations. Since
kg ðkþ1Þ  g ðkÞ kL2 < TOLiter : ð29Þ
Eq. (19) is a third order ODE, following [13,14], we use the method
of reduction of order to get a system of a first order and a second For all the results discussed here, we have taken at least 1012 as
order ODE equivalent to Eq. (19). For this, we use the substitution the value of TOLiter.
0
p = f in the problem determined by Eqs. (19), (20) and (17) and Finding constant of integration: The constant of integration B is
obtain the following equivalent problem. determined by hit and trial by requiring that the computed value
Solve of f at the upper boundary g = 1 matches with the given boundary
f 0 ¼ p; ð21Þ value of f up to a specified tolerance.
0 Improving order of accuracy by extrapolation: To increase the
p00  c1 g 0 þ Rðp2  fp Þ ¼ B; ð22Þ
order of accuracy, the discrete problem is first solved on a basic
00 0 0
g þ c2 ðp þ 2gÞ ¼ c3 ðpg  fg Þ ð23Þ grid, say h 2 [0, H] for H > 0. A sequence of approximate solutions
is then computed on successively refined grids. Let U(hl) denote
subject to the boundary conditions
3 the discrete solution corresponding to the step sizes
pð1Þ ¼ 0; pð1Þ ¼ 0;
7 H
f ð1Þ ¼ 1  A; f ð1Þ ¼ 1; 5 ð24Þ hl ¼ ; l ¼ 1; 2; 3; . . . ;
nl
gð1Þ ¼ 0; gð1Þ ¼ 0:
where U(hl) stands for either of p and f and {nl} is a sequence of inte-
We note that the method of reduction of order not only facilitates gers associated with the step size sequence {hl} to govern the suc-
the determination of numerical solution of Eq. (19) but also gives cessive refinement procedure. There are several choices for the
velocity field as a solution of Eq. (22). sequence {nl} found in literature. We use the Romberg sequence
For the numerical solution of the above problem we first dis- given below:
cretize the domain [1, 1] uniformly with step h. The solution pro-
SR ¼ f1; 2; 4; 8; 16; 32; 64; 128; 256; 512; 1024g:
cedure may be described as follows. Eq. (21) is integrated using
Simpson’s rule Gerald [15] with the formula given in Milne [16]. On the basis of solutions U(hl), higher order approximations to
Eqs. (22) and (23) are discretized at a typical grid point g = gn of the exact solutions can be obtained by the use of Richardson’s
the interval [1, 1] by employing central difference approxima- extrapolation. This process can be carried out using any extrapola-
1898 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902

tion scheme [19], e.g. polynomial extrapolation (Aitken–Neville Table 2

algorithm) or rational extrapolation (Stoer’s algorithm). We use Dimensionless streamwise velocity f 0 (g) on three grid levels and its extrapolated
values for c1 = 3.2, c2 = 3.3, c3 = 3.4, A = 1.6 and R = 100.
polynomial extrapolation, which can be presented as
g f 0 (g)
U l;1 ¼ Uðhl Þ; l ¼ 1; 2; 3; . . . ; ð30Þ
h = 0.005 h = 0.0025 h = 0.00125 Extrapolated value
U l;m1  U l1;m1
U l;m ¼ U l;m1 þ ; m ¼ 2; 3; . . . ; l; l P 2; ð31Þ 0.8 1.015941 1.015884 1.015851 1.015838
ðnl =nlmþ1 Þ2  1 0.4 2.355390 2.355308 2.355256 2.355236
0.0 2.407308 2.407265 2.407236 2.407224
where l indicates the grid level and m is the number of extrapola- 0.4 1.734890 1.734867 1.734853 1.734847
tion steps. The order of approximation of the solution Ul,m increases 0.8 0.630862 0.630855 0.630850 0.630849
in even multiples of m, i.e. the error of Um,m is proportional to H2m.
Improving initial guess for higher grid levels: An element of mul-
tigrid methods called nested iterations has been used to a limited Table 3
extent to obtain an improved initial guess for the solution U(l), Comparison of the present and literature [4] results for the Newtonian case.
where l > 1, so that rapid convergence of the iterative procedure
R A Type of results f 0 (1) f 0 0 (1) g0
may be achieved. This may be done by subjecting the solution
U(l1) to an appropriate interpolation operator so that interpolated 48.54 1.9461 Lit. numerical 10.197 9.657 0.4866
Lit. series 10.202 9.662 0.4866
values are provided at the new grid points of the level l. A compu- Present 10.196 9.658 0.4862
tationally convenient choice is the linear operator, which gives the 156.44 1.8602 Lit. numerical 10.035 8.635 0.4628
improved initial guess for the solution U(l) as follows: Lit. series 10.036 8.636 0.4629
Present 10.036 8.636 0.4687
ðlÞ ðl1Þ
U 2i ¼ U i ; 206.42 1.6194 Lit. numerical 10.449 6.477 0.3855
Lit. series 10.548 6.539 0.3837
ðlÞ 1 ðl1Þ ðl1Þ
U 2iþ1 ¼ ðU i þ U iþ1 Þ; Present 10.548 6.540 0.3837
2
where 0 6 i 6 nl1 N  1, and l denotes the grid level.
flow fields, we first note that A is determined by the ratio of the
4. Results and discussion injection velocities at the two walls and R is based on the injection
velocity at the upper wall. When R = 0, the problem becomes that
In this section we present our findings in tabular and graphical of the channel flow with impermeable walls, i.e. V1 = 0 = V2. In such
form together with the discussion and their interpretations. As our a case A will be arbitrary in view of the relation V1 = (1  A)V2 (by
objective is to develop a better understanding of the effects of definition of A) and have no effect on the flow. When the magni-
micropolar structure of fluids and permeability of two walls on tude of R is increased for a fixed value of A, it corresponds to the
the flow characteristics, we choose to present shear and couple situation in which the injection velocities at the upper and lower
stresses at the two walls, and the velocity and microrotation fields walls are increased keeping their ratio constant. On the other hand
across the channel for a range of values of the Parameter R, perme- if R is fixed and A is increased from 1 to 2, it represents the situa-
ability parameter A and a few cases of material properties of micro- tion in which, for a given injection velocity at the upper wall, the
polar fluids. All the cases of material properties considered in the injection velocity at the lower wall is increased from zero to its
present work are shown in Table 1. These values of c1, c2 and c3 value at the upper wall. For R < 0, the case A = 1 corresponds to
are taken arbitrarily to investigate their influence on the flow as the problem of channel flow with impermeable lower wall and
chosen customarily in the literature by [11,13,20,21]. permeable upper wall, 1 < A < 2 corresponds to the non-zero and
In order to establish the validity of our numerical computations unequal injection velocities at the lower and upper walls and
and to improve the order of accuracy of the solutions, numerical A = 2 represents the problem of symmetric channel flow with equal
results are computed for three grid sizes h = 0.005, 0.0025 and injection velocities at both the walls.
0.00125 and then extrapolated using Richardson’s extrapolation. First of all we present the effect of R on the shear and couple
The comparison of numerical values of streamwise velocity for stresses at the channel walls. The numerical values of f 0 (1) and
the three grid sizes and its extrapolated values is given in Table f 00 (1),and g 0 (1) and g 0 (1), which are proportional to the shear
2. Excellent comparison validates our numerical computation and and couple stresses at the lower and upper walls, respectively,
the use of extrapolation to improve the accuracy of the results. An- are given in Table 4 for typical values of the parameters c1, c2
other source of validity of our results is Table 3 in which present and c3 and A. We observe that the shear and couple stresses ex-
results have been compared with the literature results [4]. Shres- erted by the lower wall are larger in magnitude than those exerted
tha and Terrill [4] computed values of shear stresses at the walls by the upper wall and are in opposite direction. By increasing the
and the position of viscous layer across the channel using numer- magnitude of R, the magnitude of the shear and couple stresses
ical technique and the series method. An excellent comparison of at the lower wall increases while it decreases at the upper wall
our numerical results for the case of Newtonian fluid flow with by increasing the magnitude of R. This behavior is due to the fact
the literature results may be seen in Table 3 for three sets of values that the injection velocity at the upper wall is larger than that at
of R and A. the lower wall. The fluid is pushed from the upper wall towards
In order to facilitate understanding of the influence of the per-
meability parameter A defined in Eq. (1) and the Parameter R on Table 4
Shear and couples stresses at the lower and upper walls for c1 = 3.2, c2 = 3.3 and
c3 = 3.4, A = 1.6 and various values of R.
Table 1
Values of dimensionless parameters c1, c2 and c3 for the three cases. R f 0 0 (1) f 0 0 (1) g 0 0 (1) g 0 0 (1)
0 8.3398 8.2464 3.8650 3.5444
Case c1 c2 c3
10 9.8874 6.8704 4.1454 3.2634
1 2.5 2.7 2.9 50 10.5103 6.4413 4.3025 3.1394
2 7.5 7.7 7.9 100 10.5403 6.3787 4.3282 3.1199
3 11.5 11.7 11.9 200 10.5398 6.3474 4.3404 3.1102
 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902 1899

Table 5 1.2
Shear and couple stresses at the lower and upper walls for c1 = 1.7, c2 = 1.8 and
1
c3 = 1.9, R = 20 and various values of A.
0.8

f(η )
A f 00 (1) f 00 (1) g 0 (1) g 0 (1)
0.6 R=0
1.0 12.9490 3.1575 1.8460 1.2118 R = - 10
0.4 R = - 50
1.2 12.0356 4.1926 2.1296 1.5017
1.4 11.2159 5.4288 2.3628 1.8111 0.2 R = - 300
1.6 10.6757 6.8757 2.5513 2.1373
0
1.8 10.4203 8.5304 2.7029 2.4766
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
2.0 10.3851 10.3851 2.8256 2.8256 -0.2

-0.4 η

-0.6
the lower wall. This results into smaller value of streamwise veloc-
-0.8
ity gradient at the upper wall and its larger value at the lower wall.
The influence of permeability parameter A on shear and couple Fig. 3. Normal velocity profiles for A = 1.6, c1 = 3.2, c2 = 3.3 and c3 = 3.4 and various
stresses is shown in Table 5 for a fixed value of R and material con- values of R.
stants c1, c2 and c3. As A is increased, the shear stress at the lower
wall decreases while it increases at the upper wall. For this case, 0.5
the shear stress at the lower wall is maximum while it is minimum
0.4
in magnitude at the upper wall. This is the most asymmetric case.

g(η)
0.3
We note from Table 5 that as A is increased from 1 to 2 the shear R=0
stress at the lower wall decreases from its maximum value to a 0.2 R = - 10
R = - 50
minimum value while at the upper wall its magnitude increases 0.1
R = - 300
from its minimum value to the maximum value so that for A = 2, 0
the shear stress at both the walls becomes equal in magnitude -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
-0.1
reflecting the symmetry of the problem. The couple stresses in-
-0.2 η
crease at both the walls by increasing A. A comparison of Tables
4 and 5 show that the Parameter R and permeability parameter A -0.3
have opposite effects on the shear stresses at the two walls. -0.4
The influence of the micropolar parameters c1, c2 and c3 on the -0.5
shear and couple stresses is given in Table 6. It is concluded that
the shear stresses decrease whereas the couple stresses increase Fig. 4. Microrotation profiles for A = 1.6, c1 = 3.2, c2 = 3.3 and c3 = 3.4 and various
at the walls by increasing the values of micropolar parameters c1, values of R.

c2 and c3.
Now in order to investigate the effect of dimensionless con-
stants c1, c2 and c3, R and A on the primary flow fields, we give 3.5
graphical presentation of the streamwise and normal velocity pro-
3
files, and the microrotation across the channel. The influence of the
parameter R on the streamwise velocity is presented in Fig. 2 for 2.5
fixed values of A and c1, c2 and c3.
2
f ' (η )

Table 6 1.5
The effect of micropolar material parameters c1, c2 and c3 on shear and couple stresses Case 1
1 Case 2
for R = 40 and A = 1.8.
Case 3
Case c1 c2 c3 f 00 (1) f 00 (1) g 0 (1) g 0 (1) 0.5

1 2.5 2.7 2.9 10.1616 8.1987 3.7194 3.2576 0

2 7.5 7.7 7.9 9.1423 7.5622 7.0022 5.4490 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
3 11.5 11.7 11.9 8.2423 7.0772 7.8420 5.8687 η

Fig. 5. Streamwise velocity profiles for R = 40, A = 1.8 and three cases of c1, c2 and c3.

2.5 1.2
1
2 0.8
f( η)

Case 1
0.6
Case 2
1.5
f ' (η )

Case 3 0.4

R=0 0.2
1
R = - 10 0
R = - 50 -1 -0.8 -0.6 -0.4 -0.2 -0.2 0 0.2 0.4 0.6 0.8 1
0.5 R = - 300 η
-0.4
0 -0.6
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
η -0.8
-1
Fig. 2. Streamwise velocity profiles for A = 1.6, c1 = 3.2, c2 = 3.3 and c3 = 3.4 and
various values of R. Fig. 6. Normal velocity profiles for R = 40, A = 1.8 and three cases of c1, c2 and c3.
1900 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902

1.5 injection velocity at the lower wall to be 60% of that at the upper
wall, the profile becomes asymmetric pushed towards the lower
1 wall. As the magnitude of R is increased, the profile becomes more

g(η)
Case 1
Case 2 asymmetric. The point of maximum velocity is shifted towards the
Case 3 0.5 lower wall and the streamwise velocity on the lower wall side
increases while it decreases on the upper wall side. For large value
0 of R, the effect of varying R tends to become insignificant.
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Fig. 3 presents the profiles of normal velocity component across
-0.5 η the channel to reflect the influence of R on its behavior for typical
values of A and dimensionless constants c1, c2 and c3.
-1 The normal velocity takes its dimensionless value 1 at the
lower wall and increases to1 at the upper wall with a point of
-1.5 inflection somewhere near the centre of the channel where it
changes its concavity. As the magnitude of R is increased, the pro-
Fig. 7. Microrotation profiles for R = 40, A = 1.8 and three cases of c1, c2 and c3. files reflect significant increase in the normal velocity within the
channel away from the walls for 50 6 R 6 0. As R is further in-
creased in magnitude, its effect becomes less influential and the
f'(η)

3.5 gaps between the profiles appear to approach zero as R ! 1. This
A = 1.0
3
is due to the constraints of normal velocity being 1 at the lower
A = 1.2
A = 1.4 wall and 1 at the upper wall, which do not allow the normal veloc-
A = 1.6 2.5 ity to increase indefinitely as the magnitude of R is increased.
A = 1.8 The behavior of the microrotation across the channel for fixed
A = 2.0 2
values of A, c1, c2 and c3 and for different values of R is shown graph-
1.5 ically in Fig. 4. The shear stresses at the two walls tend to rotate the
1
fluid in opposite directions because of which the microrotation has
opposite signs near the walls. When R = 0, the effect of shear stres-
0.5 ses propagates at equal rate from the walls across the channel
0
resulting into zero microrotation at the mid point of the channel.
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 As the magnitude of R is increased, the microrotation increases,
η and the point of zero microrotation moves towards the lower wall.
Figs. 5–7 show streamwise and normal velocity and microrota-
Fig. 8. Streamwise velocity profiles for R = 20, c1 = 1.7, c2 = 1.8 and c3 = 1.9 and
various values of A.
tion profiles for three sets of values of c1, c2 and c3 (given in Table
1) when A and R are fixed. The values of c1, c2 and c3 have signifi-
cant effect on microrotation as compared to their effect on stream-
For R = 0, the profile is symmetric and parabolic with respect to wise and normal velocity profiles. The maximum value of
the centre of the channel, the well known couette flow with zero streamwise velocity and microrotation increases as we increase
velocity at the walls and maximum velocity at the centre of the the values of c1, c2 and c3.
channel. When the non-zero injection velocity V2 is imposed at The effect of A on the velocity and microrotation profiles is
the upper wall (i.e. R < 0), for A = 1.6 which corresponds to the shown in Figs. 8–10 for fixed values of R and c1, c2 and c3. The

1.5

A = 1.0
A = 1.2
1
A = 1.4
f (η)

A = 1.6
A = 1.8
A = 2.0
0.5

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

- 0.5

-1

-1.5

Fig. 9. Normal velocity profiles for R = 20, c1 = 1.7, c2 = 1.8, c3 = 1.9 and various values of A.
 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902 1901

0.4

0.3
A = 1.0

g (η)
A = 1.2
A = 1.4
0.2
A = 1.6
A = 1.8
A = 2.0 0.1

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1

η
-0.1

-0.2

-0.3

-0.4

Fig. 10. Microrotation profiles for R = 20, c1 = 1.7, c2 = 1.8, c3 = 1.9 and various values of A.

streamwise velocity increases by increasing A across a major part A. The maximum value of microrotation increases while the mini-
of the channel on the upper wall side and the profile tends to be- mum value decreases as A is increased so that they reach their
come symmetric as A is increased from 1 to 2. The maximum value respective extremes at A = 2. A comparison of the profiles of
of streamwise velocity increases by increasing A. The normal veloc- streamwise velocity for Newtonian and micropolar fluids shown
ity profiles given in Fig. 9 are helpful in finding the position of vis- in Fig. 11 validates our micropolar model. Table 7 compares wall
cous layer, which is developed due to the injection at the two shear stresses and the position of viscous layer for Newtonian fluid
walls. The position of viscous layer is a point where f(g) = 0. We with those for the three cases of micropolar fluids considered in
observe from Fig. 9 that the position of viscous or shear layer the present work. Comparison of Newtonian results with the Case
moves towards the centre of the channel as we increase A. For 1, results of micropolar fluids, shows that for smaller values of c1, c2
A = 2, this viscous layer coincides with the centre of the channel and c3 there is no significant effect of micropolar structure on the
and the problem reduces to the symmetrical injection case, dis- shear stresses. However, as c1, c2 and c3 are increased (Case 2
cussed by Berman [1] for Newtonian fluids. A comparison of Figs. and Case 3), the wall shear stresses decrease significantly indicat-
3 and 9 shows that the position of viscous layer is more sensitive ing that microrotation decreases the viscous forces and increases
to the permeability parameter A than to the parameter R. From the couple stresses at the walls. This fact has also been observed
microrotation profiles shown in Fig. 10, it can be observed that by [20] for the flow and heat transfer of micropolar fluids between
the position of the point where the curves change their concavity two porous disks.
goes on shifting towards the centre of the channel by increasing
5. Conclusions
3
In the present work we have considered the numerical solution
of the problem of two-dimensional steady, laminar and incom-
2.5 pressible flow of a micropolar fluid in a channel with parallel por-
ous walls of different permeability. The problem of channel flow
with one permeable and the other impermeable wall and that of
2

Table 7
1.5 Comparison of the Newtonian and micropolar results.
f '(η)

R A Type of results f 00 (1) f 00 (1) g0

1 Newtonian 10.1989 9.6582 0.4862

Newtonian 48.54 1.9461 Micropolar case 1 10.0393 9.5108 0.4862
Micropolar Micropolar case 2 09.3471 8.9018 0.4862
Micropolar case 3 08.7662 8.4072 0.4862
0.5
Newtonian 10.0367 8.6356 0.4686
156.44 1.8602 Micropolar case 1 09.9799 8.5916 0.4686
Micropolar case 2 09.7178 8.4118 0.4686
0 Micropolar case 3 09.4710 8.2561 0.4686
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Newtonian 10.5493 6.5396 0.3837
η Micropolar case 1 10.4903 6.5124 0.3837
206.42 1.6194 Micropolar case 2 10.1696 6.4092 0.3837
Fig. 11. Comparison of streamwise velocity profiles of Newtonian and micropolar Micropolar case 3 09.8296 6.3236 0.3837
fluids for R = 10 and A = 1.5.
1902 M. Ashraf et al. / Computers & Fluids 38 (2009) 1895–1902

the symmetric channel flow with both permeable or both imper- [2] Cox SM. Analysis of steady flow in a channel with one porous wall, or with
accelerating walls. SIAM J Appl Math 1991;51(2):429–38.
meable walls occur as special cases of the present problem corre-
[3] Brady JF. Flow development in a porous channel and tube. Phys Fluids
sponding to the parametric values A = 1 and R < 0, A = 2 and R < 0 1984;27:1061–7.
and R = 0 and A arbitrary, respectively. The objective of the present [4] Shrestha GM, Terrill RM. Laminar flow with large injection through parallel
study is to investigate the effects of the parameter R and perme- and uniformly porous walls of different permeability. Quart J Mech Appl Math
1968;21(4):413–32.
ability parameter A, determined by the injection velocities at the [5] Terrill RM, Shrestha GM. Laminar flow through parallel and uniformly porous
two walls, and the material constants on the flow variables. The walls of different permeability. ZAMP 1965;16:470–82.
results indicate that the parameters R and A have a strong influence [6] Terrill RM. Laminar flow in a uniformly porous channel with large injection.
Aeronaut Q 1965;15:299–310.
on the velocity and microrotation profiles, shear and couple stres- [7] Waston EB, Banks WHH, Zaturska MB, Drazen PG. On transition to chaos in
ses at the walls, and the position of the viscous/shear layer. The two-dimensional channel flow symmetrically driven by accelerating walls. J
velocity and microrotation profiles change from the most asym- Fluid Mech 1990;212:451–85.
[8] Robinson WA. The existence of multiple solutions for the laminar flow in a
metric shape to the symmetric shape across the channel as R is in- uniformly porous channel with suction at both walls. J Eng Math
creased from some negative value to zero or the permeability 1976;10:23–40.
parameter A is increased from 1 to 2. The shear stress at a wall [9] Eringen AC. Simple microfluids. Int J Eng Sci 1964;2:205–17.
[10] Eringen AC. Theory of micropolar fluids. J Math Mech 1966;16:1–8.
depends strongly on the injection velocity at it determined by [11] Guram GS, Anwar M. Micropolar flow due to a rotating disc with suction and
the choice of the values of R and A. Larger the injection velocity injection. ZAMM 1981;61:589–605.
at a wall relative to the other is, smaller will be the shear stress [12] Kelson NA, Desseaux A. Flow of a micropolar fluid bounded by stretching
sheet. ANZIAM J 2000;42(E):535–60.
at it than that at the other. The position of viscous layer has been
[13] Kamal MA, Ashraf Muhammad, Syed KS. Numerical solution of steady viscous
found to be more sensitive to the permeability parameter A than flow of a micropolar fluid driven by injection between two porous disks. Appl
to the parameter R. The material constants c1, c2 and c3 have pro- Math Comput 2006;179:1–10.
found effect on microrotation as compared to their effect on [14] Chamkha AJ. Effects of heat generation/absorption and thermophoresis on
hydromagnetic flow with heat and mass transfer over a flat surface. Int J
streamwise and normal velocity profiles. For the fluids with larger Numer Methods Heat Fluid Flow 2000;10(4):432–49.
values of these material constants, the effect of micropolar struc- [15] Gerald CF. Applied numerical analysis. Reading, MA: Addison–Wesley
ture cannot be ignored. The micropolar fluids reduce magnitude Publishing Company; 1974.
[16] Milne WE. Numerical solutions of differential equations. New York: John Wiley
of shear stress and increase that of couple stress at the walls as and Sons; 1953.
compared to the Newtonian fluids. [17] Hildebrand FB. Introduction to numerical analysis. Tata McGraw Hill
Publishing Company, Ltd.; 1978.
[18] Syed KS, Tupholme GE, Wood AS. Iterative solution of fluid flow in finned
Acknowledgements tubes. In: Taylor C, Cross JT, editors. Proceedings of the 10th international
conference on numerical methods in laminar and turbulent flow 429–
440. Swansea, UK: Pineridge Press; 1997. p. 21–5.
The authors are thankful to Dr. Michel Deville and the referees
[19] Deuflhard P. Order and step-size control in extrapolation methods. Numer
for their valuable suggestions. Math 1983;41:399–422.
[20] Takhar HS, Bhargava R, Agraval RS, Balaji AVS. Finite element solution of
micropolar flow and heat transfer between two porous disks. Int J Eng Sci
References 2000;38:1907–22.
[21] Chang Cheng Long. Numerical simulation of micropolar fluid flow along a flat
[1] Berman AS. Laminar flow in channels with porous walls. J Appl Phys plate with wall conduction and buoyancy effects. J Phys D Appl Phys
1953;24:1232–5. 2006;39:1132–40.