Acta Mechanica 148, 117-127 (2001)

ACTA MECHANICA
9 Springer-Verlag 2001

Flow o f an O l d r o y d 8 - c o n s t a n t fluid in a
convergent channel
S. Barl~, Istanbul, Turkey
(Received January 13, 2000; revised February 28, 2000)

Summary. The problem considered is the steady flow of an Oldroyd 8-constant fluid in a convergent
channel. Using series expansions in terms of decreasing powers of r given by Strauss [8] for the stream
function and stress components, the governing equations of the problem are reduced to ordinary differential equations. The resulting differential equations have been solved by employing a numerical technique.
It is shown that the streamline patterns are strongly dependent on the non-Newtonian parameters.

1 Introduction
The problem of a steady two-dimentional flow of viscous, incompressible fluid in a channel
bounded by two infinite non-parallel plane walls with a line source or sink at the intersection
of the walls was first discovered by Jeffrey [1], and Hamel [2]. They found that a purely radial
flow can be obtained which satisfies exactly both the equations of motion and the boundary
conditions. Subsequently, many writers worked on special aspects of the problem but the
most comprehensive treatments have been given by Rosenhead [3], and Millsaps and Pohlhausen [4]. An explicit exact solution can be obtained in the event of non-inertial flow (see
Birkhoff and Zarantanello [5]). Moffatt and Duffy [6] have analyzed the breakdown of the
Jeffrey-Hamel similarity transformations.
The flow of the Newtonian fluid between intersecting planes has been extended within the
context of various different non-Newtonian fluid models because of the wide range of engineering applications. For example, it will help provide valuable information on the design of
extrusion dies of industrial importance. Theoretical research on steady flow of this type was
initiated by Langlois and Rivlin [7] who carried out perturbation analyses about slow flow of
a Newtonian fluid through a wedge and a cone. They found that the stresses developed by the
viscoelastic properties of the fluid were incompatible with radial flow and vortices were predicted in the perturbation solutions. Strauss [8] was the first to present the solution for the
steady, two-dimensional, and inertial flow of an incompressible Maxwell fluid between intersecting planes by using a series expansion in terms of decreasing powers of r. In his subsequent study [9], he considered the stability of the same flow problem. Han and Drexler [10]
have studied non-inertial converging flow, using a modified second-order fluid model which
assumes that all three material functions depend on the second invariant of the rate of deformation. They also made a comparison the theoretically predicted velocity profiles and stress
distributions with the experimentally determined ones. Yoo and Han [11] carried out experiments on the converging slow flow of a polymer between intersecting planes and tried to

118

S. Ban~

explain the data in terms of the second-grade fluid model. They found that the theoretical
analysis corroborates qualitatively experimentally determined stress distributions.
In the case of most non-Newtonian fluids a purely radial flow is not possible if inertial
terms are to be retained in the equations of motion. Kaloni and Kamel [12] have shown that
there can not be a purely radial flow of Cosserat fluids in convergent channels. Later, Hull
[13] studied the non-inertial flow of a general linear viscoelastic fluid in this geometry. He
showed that radial flow is obtained for a wedge of 90 ~ and no others. The similar results are
valid for the Rivlin-Ericksen fluids (e.g., [14, 15]).
Mansutti and Rajagopal [14] studied the non-inertial flow of a shear thinning fluid
between intersecting planes. They showed that sharp and pronounced boundary layers
develop adjacent to the solid boundaries, even at zero Reynolds number. Recently, Bhatnagar
et al. [15] have extended the analysis of Strauss [8] to an Oldroyd-B fluid which is characterized by a viscosity and two material constants with units of time. In their work, the effects of
much higher values of the Reynolds number than Strauss' work [8] and the elastic parameter
c~# 0 on the streamline patterns have been discussed.
In this paper, the flow of an Oldroyd 8-constant fluid in a convergent channel is examined
using series expansions proposed by Strauss [8] for stream function and stress components.
The signiflciant contribution of a new non-Newtonian parameter, i.e. 7, which does not
appear in the previous studies, to the flow patterns (solutions) is carefully delineated. Our
results are similar to those of Strauss [8], and Bhatnagar et al. [15] but differ in detail. Also, it
is as expected possible to establish relations with their works.

2 Formulation of the problem

Consider the steady, two-dinentional, incompressible, laminar flow of the Oldroyd 8-constant
fluid through a converging channel bounded by two nonparallel planes is schematically illustrated in Fig. 1.
The Cauchy stress tensor T for the Oldroyd 8-constant fluid is related to the motion in the following manner (e.g., [16], [17])
T : - p I + S,
S+AI~-+~-

(2.1)
(S.A,+AI.S)+--

(trS) A x + - -

[tr(S.Ax)]I

(
DA1,
Az
)
: # A1 -+-/~2- ~ - t/~4A12 -[- ~- [tr (A12)] I ,

0~--O~

Fig. 1. Definition sketch for a convergent channel

(2.2)

Flow of an Oldroyd 8-constant fluid in a convergent channel

119

where p is the pressure, I is the identity tensor, S is the extra stress tensor, and
#, A1, A2, A3, A4, As, A6, A7 are the material constants. A1 is the first Rivlin-Ericksen tensor and
l)/Dt the contravariant convected derivative defined as follows (see Giesekus [18]):
A1 = V v z_ V v T

(2..3)

and
~S

Dt

OS

Ot

t- v . V S -

S- Vv-

V v T- S ,

(2.4)

where v being the velocity vector, V is the gradient operator, the superscript T denotes a
transpose operation.
It should be noted that this model includes the classical linearly viscous Navier-Stokes
fluid as a special case for A1 = A2 = Aa = A4 = A5 = A6 = A7 = 0.
In addition to Eqs. (2.1) and (2.2), the field equations consist of the equations of motion
and the continuity equation. In the case of steady flow, the former equations in the absence of
body forces takes the form
t)(v. Vv) = V . T ,

(2.5)

where t) is the (constant) density. The continuity equation is

(2.6)

t r A 1 = 0.
We shall assume a velocity field in a plane polar coordinate system (r, O) of the from
v(r, 0) = u(r, 0) er -~- v(r, O) eo,

(2.7)

where u and v denote the velocity components in the directions of r and 0, respectively.
We shall now write the field equations in terms of a set of dimensionless variables and, for
this purpose, we shall choose A1, > and Q as characteristic units. If f is used to denote the
dimentionless form of a quantity f, it follows that (see [15]):

r--

~V~I'

~ =

u,

A~

v =

v,

A1

--

p = # ]9,

S ij

A1
#

(2.8)

S ij,

where Q is the volume flux per unit distance normal to the plane of flow and has dimensions
of L2T -1. We suppose that the flow is driven steadily with flux Q. Thus, Eqs. (2.1), (2.2),
(2.5), and (2.6) in nondimensional form become:
= -fI +

S,

(2.9)

S _l_~ _S~~_
D e3 (S- ~kI @ J~kI 9S) -[- 2 (tr S) A1 + ~56 [tr (S. A1)]I =
i l + s2 ~~)AI + r
Re(V.

Vv)

e7
2 + ~-

[tr (A12)] I,

(2.1o)
(2.11)

= V-T,

(2.12)

tr 2~kl = 0,
where
Re = ~Q,
#

A2
s

~ ~11 ~

Aa
s

= ~

A4
s

= ~11'

A5
s

~ --'~I ~

A6
s

= --.~i ~

A7
s

= --,)~i

(2.13)

120

S. Ban~

where the -A1 satisfies a b o v e dimensionless equations o b t a i n e d from Eq. (2.3) by replacing A ,
by A1- H e n c e f o r t h for convenience, we shall d r o p the bars that a p p e a r over the various quantities.
W e now turn our a t t e n t i o n to the e q u a t i o n s of m o t i o n (2.11). A f t e r the pressure is elimin a t e d by cross-differentiating Eqs. (2.11), one obtains the following g o v e r n i n g equation:

q-v

r 002

--

arcOSJ

r cOrcqO{ r ( S ~ - See)} -Or

f o r (r2S'e)

q r 082

(2.14)

W e shall express stream function and extra stress c o m p o n e n t s as a series e x p a n s i o n of the

following f o r m (see [8[)

>(r, 8) = ~ r ,Fn

(2.15)

n=0

S,.~.(r, 8) = ~

a~(8))
rn

n=0

s~O(r, 8) = ~
~

b~(8)

n=0

Tn

sOO(r,O)= ~

c~(O)

(2.16)

gn

n=0

Here, we take into account the first five terms of the series e x p a n s i o n (2.15).
By defining a stream function ~b(r, 8), such t h a t
u --

1 0r

0r

v --

r 08 '

(2.17)

Or

the c o n t i n u i t y e q u a t i o n is satisfied automatically. N o w is the time to express the functions

a~(8), b.,~(8), and c~(8) in terms of the r
functions in the series e x p a n s i o n (2.15). W e do
this by substituting Eqs. (2.16) and (2.17) into Eqs. (2.10). E q u a t i n g p o w e r s of r -~ up to
n = 6, and thus
a~ = b~ = ~.,~= 0,

(n = 0, 1)

a2 = - 2 r

b2 = r

a3 = - 4 r

b3 = r

- 3r

a4 = - 6 r
b4 = r

(2.18)

c2 = 2 r
,

(2.19)

ca = 4r

(2.20)

+ 2flr ''2 + r](r ''2 + 4r

-- 8r

(2.21)

q- 4flr

(2.22)

c4 = 6~2' + 8fl9o '2 + r/(gao"2 + 4r

a5 = - - 8 r

q- 4fl(r162

b5 = r

- 15r

c5 = 8r

-- Sfl(4r162

r162

q- f l ( - 3 r 1 6 2
- r162

- 2r162

q- f l ( - 1 6 r 1 6 2
- r162

' - 3r162

q- 5r162

q- 2r](8r162

-c 16@1 I2 -~ 24r162 ' - 16r162

- 24r

+ 27](8r

-k 6r162

a6 = -- 10~)4' q- 2fl(2r '2 -- 2r162

b6 = r

(2.23)
q- r 1 6 2

- r162

- 3r162

(2.25)

q- r 1 6 2

~ , ,V2 ~_ r .Dr0
~ / . V0 ,2 -- r162 '/ -}- r
-- 6 V0
-- 6'~Jdl@l'Zq- r

(2.24)

-}- 2r162

q- 6r162 q- 24g;oI'r '2 q- 8r162

- 27(4r '2 + r ''2) ~bo'/,

(2.26)
q- 2r , ,r , ) + r1(9r

'/) -- 2T(4~/)0 ,3 @ r162

-- 7r

(2.27)

Jr 6r
(2.28)

Flow of an Oldroyd 8-constant fluid in a convergent channel

e6 = 10r

+ 2fl(9r 2 24r 'a + 14r

"4- 24r162

-- 16r162

-- 6r162

q- 26r162

121

- 10r162

q- r "2 q- 2r162

- 5r162

+ r/(9@12 q- 16~)1'2

q- 2~(4r '2 + r "2) r

(2.29)

where fl, ~, % % and t~ are dimensionless constants which only depend on the material constants of the viscoelastic fluid considered here. They can be represented in the following way:

(2.30)
f]:c4

(2.31)

q - c 7 - - ~ 3 --c6

= ,(~3 + C5 + 1) + fl(C5 + ~6),

(2.32)

% = (~3 + C5) (~ + ~) -- , ,

(2.33)

= - ( ~ + ~ ) (, + 9) + 3~ - 9 ( ~ + ~ ) .

(2.34)

Next, inserting a~'s, brJs, and c~'8 (up to n = 6) from Eqs. (2.18)-(2.29) into Eq. (2.16) and
substituting these expressions for S "T, co~'~and S ~176
into Eq. (2.14), a very long and tedious calculation yields the equations at various orders of r -~. Here, we carry out our analysis up to
order r~ = 4. Equating the coefficients o f t ~ r -1, r -2, 7" a, and r 4 to zero, we get the following
differential equations, respectively,
r IV @ 4r

+ 2Re r162

= 0,

(2.35)

r Iv + (10 + 3Rer

+ (2Re r

+ (9 + Re {3r

r Iv + (20 + 4Re r

+ (2Re r

+ (64 + Re {16r

= - 4fl(r ~v + 4r

+ Re (r

- 2r162 - 3 ~ 1 ' ~ " ) ,

r IV q- (34 + 5Re r
= fl(--45r162

% " + (2Re r

-- 16r

, -- 50r162

q- 4r162

(52 + 6Re r
= Re ( - 8 r 1 6 2

+ r162

q- 4r162

- 4r

+ 2r162

+ fl(-384r162

- 24r162162
+ 10r

G~" + (2Re r

- 24~oIVr '2 - 4r

- 3r
' -- 5r162

q- 2r162

q- r162

- 144r162 '2 - 16r162

' - 6~o'r

r
-1- r162

r
- 5r162

+ 12r "3 - 120r162

2r162

+ 3~z"%

+ 8r162

- 36r162 - 50r162

"~ - .5~r162 Iv + r162 v) + 22/(144r162 ~2 + 4r "3 + 6 4 r 1 6 2

'2 q- 3r162

(2.38)

- 4r

- 3r

TM +

(2.36)

-- 4r162

+ 6r

= 0,

(2.37)

+ (576 + Re {96r

- 42r

- 2r

@ (225 + Re {45r

-4- Re (--14r162 q- r162 -- 3r162 r -- 4r

- r

+ 6r162 m2

(2.a9)

The adherence boundary conditions of the problem as follows:

u ( r , i ~ / = 0,

v ( r , ~) = 0

(2.40)

which by virtue of Eqs. (2.15) and (2.17) implies that

~) = o,

(,~ -- o, 1, 2, ...)

(2.41)

5) = o,

(n = 1, 2, 3, ...).

(2.42/

122

S. Ban~

In order to solve 0o we need to specify two additional boundary conditions. For this reason, the volumetric flow rate through the channel is used:
~-d~

~o

= 0(~, + ~ ) - 0 ( ~ , - ~ )

: -1.

(2.43)

Here the minus sign denotes the flow of a convergent channel. Assuming the flow is symmetric about 0 = 0, then
]
0(r, + a ) = :F ~

(2.44)

and using the series expansion (2.15), we have

00(

1
= :F ~ .

(2.45)

0o(0) is the Jeffrey-Hamel solution for the corresponding flow of a Newtonian fluid which
can be expressed exactly in terms of elliptical functions (see [19]).
Eq. (2.36) is a linear homogeneous ordinary differential equation subject to boundary conditions (2.41) and (2.42) and its solutions is
01(0) _= o.

(2.46)

On substituting Eq. (2.46) into Eqs. (2.37) and (2.38), the term having Re factor in Eq. (2.37)
vanishes and the terms on the right-hand side of Eq. (2.38) do. We thus find that
~3(0) - o.

(2.47)

By virtue of Eqs. (2.46) and (2.47), the stream function reduces to

0

o) -- 00(0) +

+ ..

(2.4s)

and the extra stress components are:

s,T(

,ol =

...,

(2.491

sT~ (~, o) = ~b2(o)

- + Tb4(o)
4 - + vo(o)
~6 ...,

(2.50)

s~176 o) = c~o_~)+ c4(0)

~4 + c6(0)
~6 ....

(2.51)

We now pay special attention to the case in which inertial effects are neglected. For such
flows, the right-hand side of Eq. (2.37) is equal to zero due to Eq. (2.35). Then it becomes a
linear homogeneous ordinary differential equation and gives a zero solution under the boundary conditions (2.41) and (2.42), i.e. 02(0) -= 0. Thus, the resulting equations are now in the
following form:
OoTM + 400" = O,

(2.52)

04 TM + 5204" + 57604 = 2fl(--7200"00 '2 + 600 f'3 -- 1200'00"06" -- 1200IV00 '2)

--27 (14400"00 '2 406 "a + 6400'0~00" + 600"00 "2 + 400IV00'2 + 300IV00"2) 9

(2.53)

Flow of an Oldroyd 8-constant fluid in a convergent channel

123

Eq. (2.52) which satisfies boundary conditions (2.41) and (2.45) is solved by the following
simple analytical expression
20 cos 2c~ - sin 20
~0 (0) = 2 sin 2c~ - 4c~ cos 2c~'

(2.54)

Using the solution given above for ~0(0) in Eq. (2.53), it is simplified to yield
~)4 IV -I-

52~4't -F 576~4

1 6 ( 3/3 + 87 + (16-y
-- - 3/3) cos4a'~/

20

(2.55)

The general solution of Eq. (2.55) satisfying the boundary conditions (2.41) and (2.42) is
~4 (0) = (
3/3 + 82/+ (167 - 3/3) cos 4c~
~ (cos 2c~ - cos 20) 2 sin 20.
\ 6 ( 3 + 2 cos 4c~)(sin 25 - 25 cos2a)3 J

(2.56)

3 Results and discussion

We solved the governing differential equations as a system of first-order differential equations
for Re ~ 0, using the shooting method. For given values of parameters, the conditions
~ I ' ( - a ) and ~1~(-c~) are roughly estimated and differential equations are processed by
using the fourth-order Runge-Kutta procedure. The mathematical problem is to find the correct values of ~"(-c~) and ~'(-a) which yield the known values of ~ ( + a ) and ~'(+c~)
at terminal point. Since for Re = 0 the analytic solution provides exact initial values for
~ " ( - a ) and % ' ( - a ) then successive numerical solution can be generated as Re is increased.
The systematic way used here for finding values of the missing initial conditions is equivalent
to a modified Newton's method, for finding the roots of equations in several variables. The
accuracy of missing initial conditions at 0 - -c~ which yield the known values at the terminal
point (0 = +c~) is 10 5 at least.
The predictions based on the foregoing analysis are displayed graphically in Figs. 2 - 6 .
The curves in these figures are obtained for the constant stream function of the form:
~2(0)
~(r,0) = ~0(0) + W
~ - ~ ~4(0)
r~

(3.1)

The streamline patterns are indicative of the nature of the secondary flow near the corner,
unlike the case of Newtonian fluid.
The same problem as that investigated in the present paper has been solved previously
by Strauss [8] for the Maxwell fluid, and Bhatnagar et al. [15] for the Oldroyd-B fluid. In
the special cases corresponding to Maxwell fluid (e2 = 7 = 0) and Oldroyd-B fluid
(52 r 0, 7 = 0), there is as expected an overlap between their governing equations and ours
(see Eqs. (2.35)-(2.39)). Note that some terms, which are merely relevant to ~1, on the
right-hand sides of the Eqs. (3.6) and (3.13) in [15] were written wrong, but these do not
cause any change in the streamline patterns (solutions). The results presented here are in
complete agreement with those given by the present autors [8], [15], for the specific values of
the Reynolds number Re and the elastic parameter 52 for which they have given results. This
gives us confidence regarding the numerical work.
Figure 2a and b depicts the streamline patterns for the non-inertial flows of the Maxwell
and Oldroyd-B fluid, respectively. We observe from this figure that the effect of 52 is to diminish
the size of the circulating cells. For the Oldroyd-8 constant fluid, the effect of 7 is insignificant
on the streamline patterns, compared with Oldroyd-B fluid, so it is not presented here.

124

S. Ban~

9:-~!i. . . . . . .
0.8

0.6

!j

/
0.8
0.6

Y
0.4 t.....

0.4

0.2

0.2

-0.6

-0.3

0.3

......

0.6

-0.6

-0.3

0.3

0.6

Fig. 2. Contour plots of the streamline r~)for Re = 0: a = 60o and "/= 0; a c2 = 0, b c2 = 0.75

0.8

0.8
0.6
0.4

Y 0.4
0'6 1

0.2

0.2
-0.6

-0.3

0.3

0.6

-0.6

-0.3

0.3

0.6

08

08!

0.6

0.6

Y
0.4

0.4

0.2

0.2

-0.6

-0.3

0.3

0.6

-0.6

-0.3

0.3

0.6

Fig. 3. Contour plots of the streamline ~b for e2 = ")' = 0; a R e = 1 1 . 4 2 , c~=30 ~

a = 3 0 0 , c R e = 1 0 0 , a = 3 0 0 , t i R e = 5 0 , c~=60 o

b Re=19.67,

In the case o f inertial flow, Figs. 3 to 5 provide the streamline patterns in the flow d o m a i n
0 < r < 1 for Maxwell, Oldroyd-B, a n d Oldroyd-8 c o n s t a n t fluid, respectively. O u r m a i n purpose is to delineate the effect of the p a r a m e t e r 7, which does n o t a p p e a r in the previous studies, o n the flow patterns. In Fig. 5, we have plotted the streamlines relevant to the Oldroyd-8
c o n s t a n t fluid with the i n t e n t i o n of investigating the c o n t r i b u t i o n of this new p a r a m e t e r to the
flow field. F r o m Fig. 5 it is evident that the p a r a m e t e r 7 does affect the streamlines of the seco n d a r y flow n e a r the c o r n e r in a significant way.

Flow of an Oldroyd 8-constant fluid in a convergent channel

0.8 i

'.

0.6

125

0.8
'".

..

0.6

Y
0.4

'"

'"

0.4
0.2

0.2
-0.6

-0.3

0.3

0.6

-0.6

-0.3

0.3

0.6

....'

'

Itit l l

0.6

,..'

0.6

'

0.4
0.2

0.4

ii

0.2
-0.6

-0.3

0.3

0.6

-0.6

-0.3

0.3

0.6

Fig. 4. Contour of the streamline ~ for ~,=0; a Re=11.42, a = 3 0 0 , c2=0.5, b

a = 3 0 ~ c2=0.75, c R e = 1 0 0 , a = 3 0 ~ e2=0.5, fl R e = 5 0 , c~=60 ~ e2=0.75

Re=19.67,

As the p a r a m e t e r 7 is increased from 0 to 0.01, while keeping Re fixed at 11.42 and

c2 = 0.5, flow has four-cell structure like Oldroyd-B fluid but the size of circulating cells
adjacent to the stationary plates being bigger and the others smaller (see Figs. 4a and 5a). In
Fig. 5b, we p o r t r a y the flow field when Re -- 19.67, c~ = 300 , c2 = 0.75, and 7 = 0.01. In
this case, it is concluded that the shape of circulating cells is m a r k e d l y different from those
in Figs. 3b and 4b. On comparing Fig. 5c with Figs. 3c and 4c (or Fig. 5d with Figs. 3d and
4d), we arrive at the conclusion that the streamlines of the secondary flow near the corner
corresponding to 7 r 0 changes from two-cell structure into four-cell structure.
Finally we shall discuss reliability of the solutions near the apex of the wedge. The differential contitutive equations used in this p a p e r are not limited to small, slowly changing deform a t i o n rates as is the Rivlin-Ericksen fluids, a subclass o f the fluids of the differential type.
N o t e that the flow in a converging channel is considered to be a rapid m o t i o n and the gradients of velocity become very large near the corner. However, Strauss' series expansion is not
a p p r o p r i a t e to a perturbation for r < 1. This is why the solutions based u p o n his approximation cannot be reliable when r < 1. Of course, this also depends on the nature of the functions
~b~(0), that is, if ~ ( 0 ) are not identically zero for large n, the solution cannot be trusted as
being meaningful for r < 1. In addition to, as a result of singularity at r = 0, the solutions are
expected to be valid only in converging (or diverging) nozzles rather than in between two
intersecting planes having a source (or sink) at r = 0. These are as expected; however it should
be mentioned that the expansions given in Eqs. (2.15) and (2.16) result in asymptotic diverging

126

S. Bar1@

0.8

0.8

0.6

,"

0.6

0.4

0.4

0.2

0.2
-0.6

-0.3

0.3

0.6

-0.6

-0.3

0.3

0.6

0.3

0.6

.'

",

~ ~

0.8

0.8~

0.6

i.

0.6

Y
0.4

0.4

0.2

0.2
-0.6

-0.3

0.3

0.6

-0.6

-0.3

0
X

Fig. 5. Contour plots of the streamline ~b a Re = 11.42, c~ = 30o , c2 = 0.5, q, = 0.01, b Re = 19.67,
c~=30 ~ c2=0.75, 7=0.01, c Re 100, ce=30 ~ c2=0.5, ~=0.018, d Re=50, c~=60 ~
e2 = 0.75, ~ = 0.1

1.5

.y."

1.3
Y 09.11.13.1~"5

1.1

/ ~ /

[ /l / / " "

0.9
-0.8

-0.4

0.4

018

-0.8

-0.4

0.4

0.8

X
a

Fig. 6. Contour plots of the streamline ga when c~= 30~ e2

b Re = 19.67

7 = 0 for 0.8 < r < 1.6; a Re = 11.42,

series (see [13]). It is evident that for solutions reliable near the corner, series expansions in
terms of ascending powers of r are required. But, as shown in [20], there is a significant difficulty in doing this due to the effect of the upstream history of deformation.
On the other hand, for r > 1, since the effect of successive terms are less significant, the
solution is probably quite reliable as r~ increases. This gives us adequate information in the
flow domain r < 1 from the tendencies suggested by the results at r _> 1. For instance, the
streamlines depicted in Fig. 6a indicate the presence of two-cells in Fig. 3a, whereas Fig. 6b
suggests a four-cell structure in Fig. 3b. Of course, the streamlines of the secondary flow illustrated in Figs. 2 to 5 for r < 1 may not have the precise structure.

Flow of an Oldroyd 8-constant fluid in a convergent channel

127

References
[1] Jeffrey, G. B.: The two-dimensional steady motion of a viscous fluid. Phil. Mag. 29, 455-465 (1915).
[2] Hamel, G.: Spiralf6rmige Bewegungen z/iher Flfissigkeiten. Jahr. Dentsch. Math. Ver. 25, 34-60
(1917).
[3] Rosenhead, L.: The steady two-dimensional radial flow of a viscous fluid between two inclined walls.
Proc. Roy. Soc. Lond. A-175, 436-467 (1940).
[4] Millsaps, K., Pohlhausen, K.: Thermal distributions in Jeffrey-Hamel flows between non-parallel
plane walls. J. Aero. Sci. 20, 187-196 (1953).
[5] Birkhoff, G. D., Zarantanello, E. H.: Jets, Wakes and Cavities, p. 272. New York: Academic Press
1957.
[6] Moffatt, H. K., Duffy, B. R.: Local similarity solutions and their limitations. J. Fluid Mech. 96,
299-313 (1980).
[7] Langlois, W. E., Rivlin, R. S.: Steady flow of slightly viscoelastic fluids. Technical Report No. DA4725/3, Divison of Applied Mathematics, Brown University, Providence (1959).
[8] Strauss, K.: Die Str6mung einer einfachen viscoelastischen Flfissigkeit in einem konvergenten Kanal.
Teil I: Die stationare Str6mung. Acta Mech. 20, 233-246 (1974).
[9] Strauss, K.: Die Str6mung einer einfachen viscoelastischen Flfissigkeit in einem konvergenten Kanal.
Teil II: Die Stabilit/it der Str6mung. Acta Mech. 21, 141 152 (1975).
[10] Han, C. D., Drexler, L. H.: Studies of converging flows of viscoelastic polymeric melts. III.
Stress and velocity distributions in the entrance region of a tapered slit die. J. Appl. Polym. Sci.
17, 2369 2393 (1973).
[11] Yoo, H. J., Hart, C. D.: Stress distribution of polymers in extrusion through a converging die. J.
Rheol. 25, 115-137 (1981).
[12] Kaloni, P. N., Kamel, M. T.: A note on the Hamel flow of Cosserat fluids. Zamp 31, 293-296
(1980).
[13] Hull, A. M.: An exact solution for the slow flow of a general linear viscoelastic fluid through a slit.
J. Non-Newt. Fluid Mech. 8, 327-336 (1981).
[14] Mansutti, D., Rajagopal, K. R.: Flow of a shear thinning fluid between intersecting planes. Int. J.
Non-Linear Mech. 26, 769-775 (1991).
[15] Bhatnagar, R. K., Rajagopal, K. R., Gupta, G.: Flow of an Oldroyd-B fluid between intersecting
planes. J. Non-Newt. Fluid Mech. 46, 49-67 (1993).
[16] Bird, R. B., Armstrong, R. C., Hassager, O.: Dynamics of polymeric liquids, vol. 1. Fluid Mech.,
p. 352. New York: Wiley 1987.
[17] Huilgol, R. R.: Continuum mechanics of viscoelastic liquids. New York: Wiley 1975, p. 191.
[18] Giesekus, H.: Zur Formulierung der Randbedingungen in Str6mungen viskoelastischer Flfissigkeiten mit Injektion und Absaugung an den W/inden. Rheol. Acta 9, 474-487 (1970).
[19] Strauss, K.: Stability and overstability of the plane flow of a simple viscoelastic fluid in a converging
channel. In: Theoretical rheology (Hutton, J. F., Pearson, J. R. A., Walters, K., eds.), p. 56. New
York: Wiley 1985.
[20] Hull, A. M.: Die-entry flow of polymers. Ph.D. Thesis, London University 1981.
Author's address: Serdar Barl~, Faculty of Mechanical Engineering, Istanbul Technical University,

Gfimfi~suyu 80191, Istanbul, Turkey