Journal of Non -Newtonian FItlid Mechanics, 15 (1984) 199-226 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
199
DECAY OF ...
Journal of Non -Newtonian FItlid Mechanics, 15 (1984) 199-226 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
199
DECAY OF SHEAR LAYERS AND VORTEX SHEETS
R.S. RIVLIN Lehigh Universiv,
Bethlehem, PA (U.S.A.)
(Received November 8, 1983)
Summary
An incompressible fluid is contained in the domain between two stationary infinite parallel rigid plates. It is assumed that for shear flows, the shear stress in an element of the fluid depends linearly on the history of the velocity gradient in that element. It is supposed that initially two steady shear layers exist in the fluid and are symmetrically disposed with respect to the mid-plane. The time-dependent velocity field which results from the removal of the forces maintaining this steady flow is calculated in the cases when the fluid is Newtonian and when it is Maxwellian. The limiting cases when the shear layers reside in an unbounded space of the fluid and when they further become vortex sheets are discussed.
1. Introduction
In a previous paper [l] we considered the time-dependent rectilinear flow of an incompressible viscoelastic fluid in a domain bounded by stationary infinite rigid parallel plates. It was assumed that prior to time t = 0 the fluid is at rest and at time t = 0 is subjected to a longitudinal pressure gradient which is subsequently held constant. The pressure gradient was assumed to depend in an arbitrary manner on distance from the mid-plane and an expression for the Laplace transform of the resulting velocity was obtained. It was assumed that the fluid is such that the relation between the shear stress and the history of the velocity gradient in an element of the fluid is linear. In [l] this result was applied to the problem of run-up of plane Poiseuille flow, i.e. the case when the pressure gradient is independent of position, in the particular cases when the fluid is Newtonian, or Maxwellian. The
200 dependence of the velocity on position and time was obtained by inverting its Laplace transform. This result was then used to calculate the time-dependent velocity field which results when the initial conditions are those of steady plane Poiseuille flow and the pressure gradient supporting it is removed. In the present paper, the expression for the Laplace transform of the velocity field obtained in [l] is used to calculate the velocity field which results when the pressure gradient is such that the steady velocity ultimately attained at time t = oo consists of two shear layers, symmetrically disposed with respect to the mid-plane, in which the velocity gradients are equal in magnitude but opposite in sign, but are constant in each shear layer. Results are obtained for both Newtonian and Maxwellian fluids. These results are used to calculate the time-dependent velocity fields which result when the shear layers are present initially and at time t = 0 the forces maintaining them are removed. The results for a Newtonian fluid are given in $5 and those for a Maxwellian fluid in 58. In the case when the fluid is Newtonian the velocity field spreads diffusively throughout the fluid (Fig. 2), decaying to zero at infinite time. When the fluid is Maxwellian the velocity field may be interpreted as the superposition of eight waves which originate at time t = 0 at the boundaries of the shear layers and are reflected back and forth from the rigid boundaries. Two of these eight waves originate from each boundary of the shear layers. One of these travels in the positive and the other in the negative direction (Fig. 4). The special cases when the initial shear layers exist in an unbounded space of Newtonian or Maxwellian fluid are discussed in $6 and 59 respectively. The limiting cases in which these shear layers are vortex sheets are discussed in 57 and 510 respectively. From the results in 59 and 510 for a Maxwellian fluid the velocity field associated with the decay of a single shear layer or vortex sheet in an unbounded space of Maxwellian fluid can be read off easily. This is done explicitly in $10 for the case of a single vortex sheet. From this result the time-dependent velocity field associated with the decay of an arbitrary initial longitudinal velocity distribution could be calculated by simple integration. 2. A basic result We consider an incompressible fluid to be contained between two fixed infinite parallel plates situated at x = f h in some rectangular Cartesian coordinate system x. We suppose that the fluid is initially at rest and that it is subjected at time t = 0 to a pressure gradient P(x) in a direction parallel to the plates, which is independent of t for t > 0. It is easy to show that a
201 possible resulting flow field is one in which the velocity is longitudinal at each point. Let u(x, t) be this velocity and let E(x, S) be its Laplace transform. We suppose that the shear stress o(t) in the fluid is related to the velocity gradient history u’(x, T) (- 00 -C 7 < t) by * a(t)
=Jt
f(t-CO
+‘(x,
r)dr,
(2.1)
where the prime denotes differentiation with respect to x. It was shown in [l] that if P(x) is an even function of x and we introduce the notation
s= (g$‘z>
(2.2)
then
G? 4 --L(cosh
[x[o(/3)
ssf(s)
- i’P(
sinh @dP
p) sinh @d/3 + tanh [hl’P(
- sinh lx
xP( p) cash #dP
J0
The corresponding velocity version integral, thus:
u(x, t) = &/Y_t’mes5(x,
j3) cash SpdS]
.
(2.3)
field can be obtained
s)ds,
Y 100
by means
of the in-
(24
where y is a positive constant such that U(x, S) is analytic for Re s > y. We now suppose that the fluid, instead of being initially at rest, is in a steady state of longitudinal flow with velocity u(x, cc). At time t = 0 we remove the pressure gradient which supports this flow, i.e. we apply an additional pressure gradient -P(x). The resulting velocity field U(X, t) is then given by 24(x, t) = u(x,
co) - u(x,
t).
(2.5)
* We shall interpret the relation (2.1) as u=
‘+f(t-+‘(x,7)dq Lt f-r0 I --m
r>O.
This allows us to include as special cases Newtonian exhibit instantaneous elasticity.
fluids and other fluids which do not
202 3. Uniform shear layers In this section we consider run-up in the case when the steady flow which is attained at time t = cc is one in which there are two regions of uniform (non-zero) velocity gradient, separated from each other and from the boundaries by regions in which the velocity gradient is zero. We suppose that the regions of non-zero velocity gradient are symmatrically disposed with respect to the mid-plane and occupy the domains x = [a, b] and x = [--a, -b], where h > b B a > 0. Let - Q, and K,, be the velocity gradients in these regions. Then,
u(x, co) =
K,,(b-X)
b>,X>U,
K&--U)
IXI
%b 0
-aax>,
+ b)
h a 1x12 b.
We may express the corresponding u’(x, cc)=
(3.1)
-b,
velocity
gradient
field in the form
-K,,{H(x-u)--(x-b)+ff(x+u)-H(x+b)},
(3 *2)
where H( ) is the Heaviside unit step function. The equation of motion for steady shearing flow of the fluid fluid is u”(X, X)0)= 9(x)//A, where p denotes
(3.3)
the steady-state
viscosity
of the fluid. Then, from (3.2) and
(3.3), P(X)=pK,,{ti(X-a)-6(x-b)+S(x+u)-S(x+b)},
where S( ) is the Dirac delta function. (2.3) we obtain