TIJSAT

A Similarity Approach for an Unsteady

Two-dimensional Forced Convective

Heat Transfer Boundary Layer Flow along

a Convergent Channel
M. S. Alam, Department of Mathematics, Jagannath University, Bangladesh

Correspondence: Abstract
M. S. Alam
Department of Mathematics, Jagannath The unsteady flow in a converging channel (“wedge nozzle”) is analyzed in
University, Dhaka-1100, Bangladesh. the present paper. A specific similarity approach is used, which leads to a
E-mail: dralamjnu@gmail.com
model where the temporal and spatial variables are combined into a single
parameter in the flow problem. In the second part of the paper the problem
of convective heat transfer is analyzed. Considering a variable wall
temperature, there are two possibilities identified, leading to similarity
solutions, namely a linear or an inverse linear wall temperature. Only the
second case produces physical solutions. Numerical results are given in
terms of skin friction and Nusselt number variations along the wedge nozzle
and velocity and temperature boundary layer distributions. The obtained
numerical results show that the skin friction increases along the wedge
nozzle for positive values of unsteadiness parameter while it decreases for
negative values of unsteadiness parameter. Results also show sharp
variations in the streamwise direction of the Nusselt number, for larger
values of the unsteadiness parameter.

Keywords: Wedge nozzle, unsteady flow, forced convection, boundary

layer.

1. Introduction
We consider in this paper the unsteady flow in a converging channel (“wedge nozzle”), formed
between two intersecting planes as shown in Fig. 1. This problem, studied originally by Pohlhausen [1]
for the steady case, refers to a pressure gradient driven flow, whose main-stream velocity varies inversely
with the coordinate along the channel flow. A contemporary reference where the wedge nozzle problem is
presented in a more general framework is the textbook by Schlichting and Gersten [2].
With the x coordinate chosen as the distance from the line of intersection of the planes, similarity
solutions of the boundary layer equations may be obtained if the velocity distribution U(x) of the inviscid
outer flow is of power-law type, which is characteristic for so-called wedge flows. One of the flows
belonging to this class is the flow in a convergent channel (sink flow), where U(x) = -a/x. If a > 0 this
velocity distribution describes a flow directed toward the line of intersection of the planes.
In this case, a backward boundary layer flow will develop, according to Goldstein [3]. This problem
was revisited by Magyari [4], who studied the Pohlhausen problem with heat transfer, with a prescribed
power-law variation of the wall temperature.

Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013 1
 TIJSAT

Fig. 1. Flow configuration and coordinates system.

The present paper is the counterpart of Magyari’s study, in unsteady conditions. Unlike in that
paper, where a fine full analytical approach was carried out, our case uses numerical solutions, also in the
framework of similarity solutions.

2. Flow dynamics and similarity analysis
We consider an unsteady two-dimensional hydrodynamic laminar boundary layer flow of a viscous
incompressible fluid in a convergent channel. With the x-axis chosen as explained in the introductory
section and the y-axis taken normal to it, the governing equations are those of continuity and momentum
(see also Mia et al.[5])
∂u ∂v (1)
+ = 0,
∂x ∂y

∂u ∂u ∂u 1 ∂p ∂2u (2)
+ u + v =- + 2 ,
∂t ∂x ∂y ρ ∂x ∂y

with boundary conditions

u = 0, v = 0 at y = 0, (3a)
u = U(x, t) as y → ∞, (3b)
where (u, v) are the velocity components along (x, y) directions, t is the time, p is the pressure, ρ is
the fluid density, υ is the kinematic viscosity, and U(x, t) is the potential velocity.
In a certain analogy with Magyari [4], who took for the steady problem U(x) = U0 L/x, where L is a
reference length defined as that x-distance where the main stream velocity is U0, we consider here the
potential flow velocity generated by a convergent channel as
U(x, t) = U0δ , (4)
-
x

where U0 is taken as positive, according to the introductory discussion and δ is a time dependent
similarity quantity δ = δ(t), which has the unit of a length. It appears that the idea to introduce such an
expression of the flow unsteadiness was proposed by Birkhoff [6] and used in a series of papers by Sattar
and co-workers [7-10].
Since the potential flow velocity is a function of x and t, the inviscid momentum equation outside
the boundary layer is

2 Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013
 TIJSAT

1 ∂p ∂U ∂U U0 dδ U02 δ2 . (5)
= + U =- -
ρ ∂x ∂t ∂x x dt x3
In order to obtain similarity solutions of the problem, we now introduce the transformation
y U0δ
ψ=- υU0δ f (η), η = (6)
x υ

where ψ is the stream function that satisfies the continuity equation (1). Thus, the velocity
components are obtained as
∂ψ Uδ ∂ψ U υδ (7)
u= = - 0 f ', v = - = - 0 ηf '
∂y x ∂x x

Now using (5) and (7) into (2), we obtain

f ''' - (f')2 + 1 + A (2 - 2 f ' - ηf '') = 0 . (8)
The boundary conditions are transformed as
f = 0, f ' = 0 at η = 0 (9a)
f ' = 1, as η → ∞, (9b)
where prime denotes differentiation with respect to the variable η and
x 2
dδ X dΔ .
2
(10)
A= = 2
2U0δ2 dt 2Δ dτ
The following dimensionless spatial and temporal variables
δ x U (11)
Δ= , X = , τ = 0 t
L L L
have been introduced, where L is a characteristic length.
We remark that the parameter A contains a combination of the spatial and temporal coordinates, as the
single parameter left in the problem.
Further, the wall shear stress is given by
∂u μU0δ U0δ (12)
τw = - μ = f '' (0).
∂y y=0 x2
υ

Here the negative sign is due to the fact that the flow is in the opposite direction to the positive x-axis.
Therefore the local skin-friction coefficient is obtained as
τw (13)
Cfx = = 2(Reδ) -1/2 f '' (0),
1 ρU2 (x,t)
2

where Reδ = U0δ /υ is a local Reynolds number. This non-conventional (modified) Reynolds number may
be related to the usual local Reynolds number Rex = U0 x/υ through the relationship Reδ = (δ /x) Rex.

Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013 3
 TIJSAT

3. Heat transfer
The energy equation is
∂T ∂T ∂T ∂2T (14)
+ u +v = α 2 ,
∂t ∂x ∂y ∂y
when buoyancy, as well as other effects, such as viscous dissipation are neglected, as in Magyari [4].
Usual notations are used for T (fluid temperature) and α (thermal diffusivity). The boundary conditions
are
T = Tw (x, t) at y = 0, (15a)
T → T∞ as y → ∞. (15b)
The wall temperature is taken as
Tw (x, t) = T∞ + (T0 - T∞) h (x, t), (16)
where h(x,t) is a dimensionless function to be found by requiring similarity solutions. Besides the
similarity variable η defined in (6), we introduce
T(x, t) = T∞ + (T0 - T∞) h (x, t) θ (η ). (17)
Energy equation (14) becomes
1 Uδ η dδ U δ ∂h ∂h (18)
h 20 θ" - hθ' + 0 f'θ- θ=0
Pr x 2δ dt x ∂x ∂t
Imposing the proportionality of the coefficients of the first and third term in (18), two possible cases are
found, which are discussed below:
(a) h = C1 x, where C1 = C1 (t) has the dimension of a (length)-1. In this case (18) becomes
1 x2 1 dδ 2 dC1 (19)
θ"+ f ' θ - ηθ' + θ = 0,
Pr 2 U0δ δ dt C1 dt
and it follows that C 1 = δ / L 2 in order to get similarity solutions and to satisfy the appropriate
dimensionality of C1 . Finally, we get
1 (20)
θ" + f ' θ - A (ηθ ' + 2θ) = 0
Pr

and the temperature distribution is given by

T(x,t) = T∞ + (T0 - T∞) δx2 θ (η) = T∞ + (T0 - T∞) + Δ Xθ (η ), (21)
L
where the quantity A is defined in (10).
(b) h = C2 / x, where C2 = C2 (t) has the dimension of a length. In this case (20) becomes
1 x2 1 dδ 2 dC2 (22)

θ " + f 'θ - ηθ' + θ = 0,

Pr 2U0δ δ dt C2 dt
and it follows the natural choice C2 = δ in order to get similarity solutions and to satisfy the appropriate
dimensionality of C2. Finally, we get
1 (23)

θ" - f 'θ + A (ηθ' + 2θ) = 0

Pr

4 Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013
 TIJSAT

and the temperature distribution is given by

(T0 - T∞)δ (T - T ) Δ (24)
T (x, t) = T∞ + θ (η) = T∞ + 0 ∞ θ (η).
x X
Equations (20) and (23) may be written in the unified form
1 (25)
θ" + s [ f 'θ - A (ηθ' + 2θ)] = 0,
Pr
where s = 1 stands for case (a), i.e. linear wall temperature variation and s = -1 signifies inverse linear
wall temperature variation as described by case (b). Equation (25) must be solved along the boundary
conditions:
θ(0) = 1, θ(∞) = 0 (26)
Furthermore, the wall heat flux is given by
∂T h(x, t) U0δ (27)
qw = -k = -k (T0 - T∞) θ' (0).
∂y y=0 x υ
Therefore the local Nusselt number is (Nux = xqw / k(Tw - T∞)) becomes, on using (16),
Nux = - (Reδ)1/2 θ ' (0), (28)
irrespective of the choice of the function h.

4. Analytical solution for steady case
In steady flow (A = 0, i.e. K = 0) the momentum equation (8) becomes
f' ' '+ 1 -f '2 = 0 (29)
which, subjected to (9), has the analytic solution
η 2 (30)
f ' (η ) = -2 + 3 tanh2 ξ, where ξ = + tanh
2 3
obtained by Pohlhausen [1]. From (30) one has f '' (0) = 4/3.

5. Numerical solutions
The boundary value problems (8-9) and (25-26) have been solved using comparatively the routines
bvp4c and dsolve from Matlab [11] and Maple [12] packages, respectively.

5.1 Flow dynamics

In a first step, we tested the numerical method for K = 0 and we obtained f"(0) = 1.1547005 which
is identical to the value obtained analytically: f '' (0) = 4/3 see equation (30).
Consequently, all curves in Fig. 2, which give the skin-friction group Cfx (Reδ)1/2 /2 = f "(0), start
from the common point (0, 4/3). The trends are evident: the skin friction increases along the wedge
nozzle for positive K and decreases for negative K.
In Fig. 3 there are plotted some samples of velocity profiles for K = -0.5. There is a small effect of
X on these velocity distributions in the boundary layer. Fig. 4 is also for velocity profiles, this time
represented at X = 1. Included also for reference in this figure is the curve corresponding to steady

Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013 5
 TIJSAT

3
2.8
K = -1, -0.5, 0, 0.5, 1, 5
2.6
2.4
2.2
2
1.8

f'' (0)
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0 0.2 0.4 0.6 0.8 1
X

Fig. 2. Streamwise skin-friction variation for several values of the unsteadiness parameter K.

0.9

0.8

0.7 X = 0 (steady), 0.2, 0.5, 0.8, 1

0.6

0.5
f'

0.4

0.3

0.2

0.1

0
0 2 4 6
η

Fig. 3. Dimensionless velocity profiles for different values of X at K = -0.5.

0.9

0.8

0.7

0.6 X = -1, -05, 0 (steady), 0.5, 1.5

0.5
f'

0.4

0.3

0.2

0.1

0
0 2 4 6
η
Fig. 4. Dimensionless velocity profiles for different values of K at X = 1.

6 Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013
 TIJSAT

conditions (for K = 0, i.e. A = 0). The influence of the unsteadiness parameter on the velocity
profiles is evident in Fig. 4. We point out that no backward flows were detected in this analysis.

5.2. Heat transfer

Results for the heat transfer are obtained by solving (25-26), and f ' (η) is known from the previous
step. Besides the parameter A from the momentum equation, there is also the Prandtl number Pr to be
prescribed for the heat transfer problem. We have two options: linear or inverse linear wall temperature
distributions. From the physical side, the linear case is easier to be understood than the inverse linear one.
However, the second one is also encountered in the literature.
Equation (25) cannot be solved analytically, but further insight is obtained by the following procedure:
multiplying (25) by θ' , integrating once and using the boundary conditions (26) yields
� f ' θθ dη = s θ'2 + θ2 + A � ηθ'2 dη
∞ ∞
(33)

η 2Pr η

It is evident that the last term in the RHS is positive, while LHS is negative, as can easily be
demonstrated. This leads to the conclusion that when s = 1 and A > 0 (i.e. K > 0) the solution of (25) does
not exist.
Moreover, the numerical runs, performed with both Maple and Matlab solvers, prove that the case s
= 1 (linear wall temperature) and K < 0 cannot be achieved, from the physical point of view, giving
unrealistic temperature profiles.
Consequently, the following results are reported for the case of inverse linear wall temperature
distribution (s = -1). Some samples of the streamwise variation of the Nusselt number are shown in Fig. 5,
for two typical values of Pr = 0.7 and Pr = 7. Our numerical runs prove that, except for very small values
of K (in absolute value), sharp variations of the Nusselt number, as given by [-θ'(0)], occur. These jumps
appear in several locations along the channel length, and these locations are more numerous as the
unsteadiness parameter K increases (in absolute value). This is the reason to interrupt the curves in Figs. 5
and 6. Otherwise, for both Pr = 0.7 and Pr = 7, the overall trend of the Nusselt number is to decrease in the
streamwise direction for K < 0, and conversely for positive K (not shown in figures).
Finally in Fig. 6 we observe that the temperature profiles are found to decrease with the increasing
values of the Prandtl number.

1.8

1.6

Pr = 7
1.4

1.2 K = -0.1
Nu

K = -0.05
1 K = -0.01

0.8

0.6
Pr = 0.7

0.4
0 0.2 0.4 0.6 0.8 1
X

Fig. 5. Streamwise Nusselt number variation, inverse linear wall temperature distribution.

Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013 7
 TIJSAT

0.9

0.8 K = 0 (X=0)
K = 0.5
0.7 K = 1

0.6 Pr = 0.7
K = 0.5
θ
0.5

0.4

0.3

0.2 Pr = 7
K = -0.5
0.1

0
0 2 4 6 8
η

Fig. 6. Dimensionless temperature profiles for several values of Pr and K.

6. Concluding remarks
In this paper, local similarity solutions for the heat transfer flow along a convergent channel have
been obtained only when the wall temperature is a linear or inverse linear function of x. No solutions exist
for the temperature in the first case. Extensive numerical computations have been carried out for the
second case, using two solid solvers, namely bvp4c and dsolve from Matlab and Maple packages,
respectively. Both of them gave the result of sharp variations in the streamwise direction of the Nusselt
number, for larger values of the unsteadiness parameter, but no anomalies were detected in the
temperature profiles.

7. Acknowledgement
The author wish to thank the anonymous referee for his valuable comments for further improvement of
the paper.

8. References
[1] Pohlhausen, K., Zur naherungsweisen Integration der Differentialgleichung der laminaren

Grenzschicht. J. Appl. Math. Mech. (ZAMM), Vol. 1, pp. 252-268, 1921.

[2] Schlichting, H., Gersten, K., Boundary layer theory, 8th Edition, Springer-Verlag Berlin Heidelberg,

p. 174, 2000.
[3] Goldstein, S., On backward boundary layers and flow in converging passages. J. Fluid Mech.

Vol. 21, pp. 33-45, 1965.

[4] Magyari, E., Backward boundary layer heat transfer in a converging channel. Fluid Dyn. Res.

Vol. 39, pp. 493-504, 2007.

[5] Mia, M. M. Sattar, M. A., Alam, M. S., A local similarity solution for unsteady two-dimensional

hydrodynamic boundary layer flow in a convergent channel, Int. J. Energy and Technology,

Vol. 3(7), pp. 1-4, 2011b.

[6] Birkhoff, G., Hydrodynamics. A study in Logic, Fact, and Similitude. Second edition, revised and

enlarged, 2nd ed, Princeton University Press, Princeton, 1960.

8 Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013
 TIJSAT

[7] Sattar, M.A., Hossain, M.M., Unsteady hydromagnetic free convection flow with Hall current and

mass transfer along an accelerated porous plate with time dependent temperature and concentration.

Can. J. Phys. Vol. 70, pp. 369-374, 1992.

[8] Sattar, M.A., Unsteady hydromagnetic free convection flow with Hall current mass transfer and

variable suction through a porous medium near an infinite vertical porous plate with constant heat

flux.. Int. J. Energy Research, Vol. 18, pp. 771-775, 1994.

[9] Alam, M. S., Sattar, M. A., Rahman, M. M., Postelnicu, A., Local similarity solutions of ajn

unsteady hydromagnetic convection flow of a micropolar fluid along a continuously moving

permeable plate, Int. J. Heat and Technology, Vol. 28(2), pp. 95-105, 2010.
[10] Sattar, M. A., A local similarity transformation for the unsteady two-dimensional hydrodynamic

boundary layer equations of a flow past a wedge, Int. J. Appl. Math. and Mech., Vol. 7(1),

pp. 15-28, 2011a.

[11] Matlab 6.00 Mathworks, Inc., 2000.
[12] Maple 8.00 Waterloo Maple, Inc., 2002.

Thammasat International Journal of Science and Technology, Vol. 18, No. 1, January-March 2013 9