5.3 Inviscid Intability Mechanism of Parallel Flows: 5.3.1 Rayleigh's Equation
5.3 Inviscid Intability Mechanism of Parallel Flows: 5.3.1 Rayleigh's Equation
5.3 Inviscid Intability Mechanism of Parallel Flows: 5.3.1 Rayleigh's Equation
Notes on
1.63 Advanced Environmental Fluid Mechanics
Instructor: C. C. Mei, 2002
[email protected], 1 617 253 2994
November 4, 2002
5.3
We now turn to an older problem of the instability of parallel flow without stratification and
gravity, such as channel flows, jets, wakes and boundary layers.
5.3.1
Rayleighs equation
In a shear flow a necessary conditon for instablity is that there must be a point of inflection
in the velocity profile U (z), i.e,
d2 U
=0
(5.3.1)
dz 2
somewhere in the flow.
Let us begin with an inviscid uniform flow.
~ = U~i,
U
P =0
(5.3.2)
and add an infinitesimal disturbance so that the total velocity and pressure fields are
(u, w, p) = (U + u0 , w0 , p0 )
(5.3.3)
where primes denote disturbances which are functions of (x, z, t). The governing equations
are
u0x + wz0 = 0
(5.3.4)
(u0t + (U + u0 )u0x + w0 (Uz0 + u0z ) = p0x
(wt0 + (U + u0 )vx0 + w0 wz = p0z
(5.3.5)
(5.3.6)
(5.3.7)
!
dU 0
0
0
ut + U ux +
w = p0x
(5.3.8)
dz
(wt0 + Uwx0 ) = p0z
(5.3.9)
2
Eliminating p0 by cross-dierntiation,
u0t
U u0x
dU 0
+
w
dz
dU
zt + U xz
x
dz
= (xt + Uxx )x
dU
f
ikCfz + ikUfz ik
dz
or
where C =
!
= ik k 2 Cf + k 2 Uf
(5.3.10)
(5.3.11)
5.3.2
Uzz
f =0
U C
f (fzz ) k 2 f f
(5.3.13)
Uzz
ff = 0
U C
dz
Z d
Uzz
|f |2 =
dz |fz |2 + k 2 |f |2
U C
0
(5.3.14)
3
Equation (5.3.14) becomes
Z
Z d
dz
|f
|
+
k
|f
|
z
|U C|2
0
(5.3.15)
dz
0
Uzz |f |2
=0
|U C|2
(5.3.16)
if Ci 6= 0. Thus for instability it is necessary that Uzz is positve for some y and negative
elsewhere, i.e., Uzz must vanish somewhere in the flow, i.e., the velocity profile must have an
inflection point inside the flow. This is Rayleighs necessary condition for instability.
5.3.3
Figure 5.3.1: (a) Velocity profile with an inflection point. (b) Vor5ticty rofile. (c) Eect of
exchanging fluid parcels on vorticity
In Figure 5.3.1 we sketch the velocity profile with a inflection point at z = 0 and the
corresponding profile of vorticity = Uz which has a maximum at z = 0. (Note the in the
right-handed system the y axis points into the paper, therefore the flow of a positive vortex
is clockwise.
Consider three layers in the region where U (z) increases with z monotonically, i.e., below
the inflection point, with (z1 ) < (z2 ) < (z3 ). If a fluid parcel descends from level 3 to
level 2, it brings with its vorticity without change, according the vorticity transport law in
an inviscid fluid. An excess vorticity is created at level 2 which tends to replace the fluid
4
on the right in layer 2 by fluid with higher vorticity, and the fluid on the left in layer 2
by fluid with lower voriticity. The net consequence is to force the original excess vortex to
return to layer 3. Similarly if a fluid parcel ascends from level 1 to level 2, it brings with its
vorticity, hence creates an excess vorticity defect with tends to replace the fluid on the left
in layer 2 by fluid with higher vorticity, and the fluid on the right in layer 2 by fluid with
lower vorticity. The net consequence is to force the original defect votex to return to layer 3.
Thus an accidental diplacement of a fluid parcel tends to it original level; the flow is stable.
By a similar reasoning, if U (z) decreases monotonically in z the flow is also stable.
However, if there is a level of vorticity extremum, say level 0 at z = 0, then a fluid element
arriving at this layer is not forced back to its origin. A fluid element on one side of level 0
is equally at home on the opposite side. The flow is unstable.
5.3.4
Z d
Uzz (U Cr
2
2
2
2
dz
|f
|
=
dz
|f
|
+
k
|f
|
z
|U C|2
0
(5.3.17)
Now let UI be the velocity at the point of inflection, we use (5.3.16) so that
Z
dz
(5.3.18)
d
0
Z d
Uzz (U UI ) 2
2
2
2
dz
|f
|
=
dz
|f
|
+
k
|f
|
<0
z
|U C|2
0
(5.3.19)
(5.3.20)
for all z in the flow. For impoications of Fjortofts theorem, see Figure 5.3.2. Jets and wakes
satisfy Rayleigth- Fjortoft criterion, but Poiseuille flows in a pipe and boundary layers on
a flat plate do not. Why then is Poiseuille flow known to be unstable beyond Re = 2100
(Reynolds)?