Vous êtes sur la page 1sur 8

Chapter 11

Laminar boundary layers


11.1 Boundary layer prole curvature
If we evaluate the boundary layer momentum equation (10.17)
u
t
+ u
u
x
+ v
u
y
=
1

p
x
+

2
u
y
2
,
right at the wall (y = 0 ) so that u = v = 0 (for no-slip and impermeability) , it reduces to
0 =
1

p
x
+

2
u
y
2
, (11.1)
so that the prole curvature

2
u
y
2
at the wall has the same sign as the pressure gradient
p
x
.
Toward the outer edge of the boundary layer, u will generally be increasing asymptotically
up to its free-stream value u

, so that

2
u
y
2
< 0 there. This means that if the pressure gradient
is positive there will be an inexion point in the velocity prole (where

2
u
x
2
changes sign).
11.2 Pohlhausens quartic proles
From (11.1), there is a simple direct link between the curvature of the boundary layer prole at
the wall and the local pressure gradient. To explore the eect of this using the assumed-velocity
method in the Karman integral relation, we need a family of proles with curvature at the wall.
Such a family was developed by Pohlhausen (1921) based on quartic polynomials. The other four
coecients of the quartic were chosen to satisfy the no-slip condition at the wall u(0) = 0 , the
matching condition at the edge of the boundary layer u() = u

, and two smoothness conditions


there too:
u
x
=

2
u
x
2
= 0 . With y/ , the requisite dimensionless quartic is (Goldstein 1938,
p. 158)
u
u

= 2 2
3
+
4
+

6
(1 )
3
, (11.2)
115
116 AERODYNAMICS I COURSE NOTES, 2005
0
0.5
1
1.5
2
0 0.2 0.4 0.6 0.8 1
N
O
R
M
A
L

D
I
S
T
A
N
C
E
,

y
/

TANGENTIAL VELOCITY, u/u

nonzero free-stream
nonzero pressure gradient
Figure 11.1: The two ingredients of Pohlhausens quartic prole: 2 2
3
+
4
which is 1 at
= 1 and (1 )
3
/6 which has -1 curvature at = 0 .
which satises
u = 0 ( = 0) (11.3)

2
u
y
2
=
u

2
( = 0) (11.4)
u = u

( = 1) (11.5)
u
y
=

2
u

2
y
= 0 ( = 1) . (11.6)
The pressure gradient parameter dened here is


2
u

p
x
=

2

x
. (11.7)
Pohlhausens prole consists of two parts, 2 2
3
+
4
which accounts for the nonzero free-
stream velocity and the part multiplied by which accounts for the curvature at the wall. The
two parts are plotted in gure 11.1 and various combinations are shown in gure 11.2.
The various boundary layer parameters are given by Goldstein (1938, p. 159). The thickness
parameters are

=
36
120
(11.8)

=
5328 48 5
2
45360
(11.9)
so that the shape factor is
H =
378( 36)
5
2
+ 48 5328
. (11.10)
Laminar boundary layers 117
0
0.5
1
1.5
2
-0.2 0 0.2 0.4 0.6 0.8 1
N
O
R
M
A
L

D
I
S
T
A
N
C
E
,

y
/

TANGENTIAL VELOCITY, u/u

=12
0
-12
-24
Figure 11.2: Some of Pohlhausens quartic proles, including that for zero pressure gradient
= 0 and the separation prole = 12 .
The local skin friction coecient is
c
f

2

u
u
Re

=
12 +
3Re

. (11.11)
Thus at = 12 , the skin friction is zero, and if < 12 , its negative which corresponds to
reversed ow and boundary layer separation. This is illustrated in gure 11.2 for = 24 .
Whenever < 0 , the pressure gradient is positive and the free-stream is decelerating; this is
called an adverse pressure gradient. Whenever > 0 , the pressure gradient is negative and the
free-stream is acceleration; this is called a favourable pressure gradient.
11.3 Pressure gradient and boundary layer thickness
In Karmans integral relation (10.41)
d
dx
+

u

(2 + H)
du

dx
=
c
f
2
,
neglecting the possibility of boundary layer separation for the moment, the skin friction coecient
will be positive. Basically this tends to make the boundary layer thickness increase with x ,
corresponding the rst term on the left-hand side, but in general this quantity is to be balanced
by both terms on the left-hand side. If the free-stream is decelerating (adverse pressure gradient),
the second term is negative and the boundary layer thickness increases more rapidly with x; if
the free-stream is accelerating (favourable pressure gradient), the second term is positive and
the boundary layer thickness will increase more slowly or may even decrease if the favourable
pressure gradient is large enough.
118 AERODYNAMICS I COURSE NOTES, 2005
11.4 Thwaitess method for laminar boundary layers
In Pohlhausens family of quartic proles,

, , and H are all functions of the local pressure


gradient, as is the product of the skin friction coecient and a Reynolds number based on
boundary layer thickness. Of course, this is just a trivial consequence of the fact that the family
contains a single parameter which can be related to this pressure gradient. However, it turns out
that if the shape factor H is plotted against


2

du

dx
(11.12)
for several proles obtained from exact laminar solutions of the boundary layer equations, then
H() seems to be a universal function (Thwaites 1987, p. 63). Notice that Thwaitess pres-
sure gradient parameter (11.12) diers from Pohlhausens (11.7) only in using the momentum
thickness instead of the plain boundary layer thickness (which is always somewhat arbitrary).
Further, the dimensionless skin friction parameter
l
Re

c
f
2


u

u
y

y=0
; (11.13)
also seems to be a universal function of for laminar boundary layers.
To apply Thwaitess ideas to Karmans integral relation (10.41)
d
dx
+

u

(2 + H)
du

dx
=
c
f
2
,
rst substitute l for c
f
from (11.13) and multiply through by Re

d
dx
+

2

(2 + H)
du

dx
= l , (11.14)
and then replace the free-stream velocity gradient with (11.12)
u

d
dx
+ (2 + H) = l , (11.15)
and rearrange
u

d
_

2
_
dx
= 2{l (2 + H)} . (11.16)
So far this is equivalent to Karmans integral relation; the novel part of Thwaitess method
is assuming that the right-hand side is a universal function of (which is known, at least
approximately) .
11.4.1 F() 0.45 6
A simple choice that performs well is (Thwaites 1987, p. 63; Moran 2003, p. 207)
2{l (2 + H)} 0.45 6. (11.17)
Laminar boundary layers 119
Substituting this into (11.16) gives
u

d
_

2
_
dx
= 0.45 6
2
du

dx
(11.18)
u

d
_

2
_
dx
+ 6
2
du

dx
= 0.45 (11.19)
u
6

d
_

2
_
dx
+ 6
2
u
5

du

dx
= 0.45u
5

(11.20)
d
_

2
u
6

_
dx
= 0.45u
5

. (11.21)
For known u

(x) , this can be integrated downstream (analytically or numerically) to get


(x) , then (x) can be calculated from (11.12). Finally, the other parameters of interest, such
as the shape factor H and in particular the skin friction parameter l are obtained from approxi-
mations to their universal functions of .
11.4.2 Correlations for shape factor and skin friction
Thwaites (1987, p. 64) provides tables of H() and l() , but more convenient are the formulae
quoted by Moran (2003, p. 207):
H()
_
2.088 +
0.0731
0.14+
, 0.1 < < 0
2.61 3.75 + 5.24
2
, 0 < < 0.1
(11.22)
l()
_
0.22 + 1.402 +
0.018
0.107+
, 0.1 < < 0
0.22 + 1.57 1.8
2
, 0 < < 0.1 .
(11.23)
Notice that the skin friction parameter l() vanishes near = 0.090, and recall that l = 0
corresponds to boundary layer separation.
11.4.3 Example: zero pressure gradient
As a simple example, lets apply Thwaitess method to a at plate with zero pressure gradient;
assume the boundary layer begins with zero thickness at the leading edge ((0) = 0 ). This
example is taken from Moran (2003, pp. 205206). The zero pressure gradient corresponds to
the boundary layer on a thin at plate at zero incidence.
In the ordinary dierential equation
d(
2
u
6

)
dx
= 0.45u
5

, (11.24)
u

is constant here so
d(
2
)
dx
=
0.45
u

(11.25)
which can be integrated to give
=
_
0.45x
u

, (11.26)
the constant of integration being zero if the boundary layer thickness begins at zero at the leading
edge (this is the case for a sharp leading edge on a at plate).
120 AERODYNAMICS I COURSE NOTES, 2005
This can be written nondimensionally as

x
=
0.671

Re
x
(11.27)
or
Re

= 0.671Re
1/2
x
. (11.28)
From the correlations of 11.4.2, noting that = 0 throughout for zero pressure gradient, we
have H 2.61 and l 0.22 ; the latter means the skin friction coecient is
c
f

0.44
Re

=
0.656

Re
x
. (11.29)
This turns out to be very close to the exact answer, 0.664Re
1/2
x
, which can be obtained by a
similarity transformation (Abbott and von Doenho 1959, pp. 8789, e.g.).
11.4.4 Example: laminar separation from a circular cylinder
The full aerodynamics of viscous ow over a circular cylinder is very complicated, involving
large-scale unsteadiness in the form of vortex shedding (Goldstein 1938, pp. 417 .; Anderson
2001, pp. 256262); however, Thwaitess method can be used to show that the laminar boundary
layer will fail to reach the rear stagnation point.
The complex velocity for the ow without circulation with free-stream speed q

over a circular
cylinder of radius a is (see 3.6.2)
w = q

_
1
a
2
z
2
_
, (11.30)
so that the azimuthal velocity on the surface |z| = a is
v

=
_
e
i
w(ae
i
)
_
= 2q

sin (11.31)
(here, for the moment, is the polar coordinate).
Take the longitudinal boundary layer coordinate x to begin at the forward stagnation point
(z = a ) and to follow the upper surface; i.e. x = a() . It is convenient to nondimensionalize
this as x/a. Then, relative to this, the external tangential speed for the boundary layer is
u

() = 2q

sin ( ) 2q

sin . (11.32)
Thus over the forward part of the upper surface (0 < <
1
2
) the pressure gradient is favourable
whereas it becomes unfavourable thereafter. Notice that x and follow the curve of the boundary,
and similarly the proles and thicknesses are measured normal to the boundary; this curvature
does not introduce any new terms into the equations provided the radius of curvature is much
greater than the boundary layer thickness.
In terms of , Thwaitess form of the momentum integral equation (11.21) is (hereafter is
the momentum thickness again)
d(
2
u
6

)
d
= 0.45au
5

(11.33)
d(
2
sin
6
)
d
=
0.45a
2q

sin
5
(11.34)
Laminar boundary layers 121
1.2 1 0.8 0.6 0.4 0.2 0 0.2 0.4
Figure 11.3: Growth of the laminar boundary layer (momentum) thickness circular cylinder in a
uniform stream (moving from left-to-right), as predicted by (11.38) for Re
a
= 100 . The model
assumes, wrongly, that the external ow is given by the complex velocity w = 1 z
2
.
which can be integrated to give

2
sin
6
=
0.45a
2q

_
cos
2
3
cos
3
+
1
5
cos
5
+ const.
_
(11.35)

2
=
0.45a
2q

cos
2
3
cos
3
+
1
5
cos
5
+ const.
sin
6

. (11.36)
The integration constant must be
8
15
if is to be nite at the forward stagnation point = 0 ,
so

2
=
0.45a
2q

cos
2
3
cos
3
+
1
5
cos
5
+
8
15
sin
6

(11.37)

2
a
2
=
0.015
Re
a
_
8 15 cos + 10 cos
3
3 cos
5

sin
6

_
, (11.38)
where
Re
a

q

(11.39)
is the Reynolds number based on the cylinders radius. The predicted momentum thickness is
displayed around the cylinder in gure 11.3.
Given this solution for () , Thwaitess pressure gradient parameter (11.12) can be computed
as
() = 0.03
_
8 15 cos + 10 cos
3
3 cos
5

sin
6

_
cos , (11.40)
122 AERODYNAMICS I COURSE NOTES, 2005
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0 30 60 90
D
I
M
E
N
S
I
O
N
L
E
S
S

F
A
C
T
O
R
ANGLE DOWNSTREAM FROM FORWARD STAGNATION POINT (degrees)
H
l
Figure 11.4: Variation of shape factor and skin friction factor along laminar boundary layer on
a circular cylinder (assuming attached ow).
and then the variation along the boundary layer of the shape factor H and skin friction factor
l computed; the results are plotted in gure 11.4. It can be seen that the skin friction factor l
vanishes near 103

. At this point the boundary layer would separate.


That the boundary layer separates means that the outer supposedly inviscid region is changed.
This invalidates the procedure used, since our assumption of a perfect uid solution holding
everywhere except in a thin layer near the boundary is untrue. This means that we do not
obtain a realistic prediction for the laminar boundary layer which does occur on the forward
part of a circular cylinder, neither for the location of the separation point. What we do obtain
is a very good argument that the ow picture cannot consist of the basic perfect uid ow
plus thin laminar boundary layers. In fact, the displacement of the exterior irrotational ow by
the separated region pushes the points of minimum pressure forward from 90

, and the actual


laminar separation points forward too. The failure to obtain a quantitative result should not be
attributed to Thwaitess method, but rather to the assumption about the surrounding inviscid
ow; any boundary layer method requires an accurate description of the prevailing longitudinal
pressure gradient as input.
Another question to be addressed in this and similar ow congurations is whether and
where the laminar boundary layers undergo transition to turbulence. This is outside the scope
of this course; see Abbott and von Doenho (1959, pp. 105107), Thwaites (1987, p. 64), Kuethe
and Chow (1998, ch. 17), Anderson (2001, p. 721), Bertin (2002, p. 128), and Moran (2003,
pp. 220222) for discussions.
In a sense only a negative result has been obtained here, but it does highlight the importance
of adverse pressure gradients in aerodynamics. Notice that in the circular cylinder problem,
where the minimum pressure was supposed to occur at 90

, the laminar boundary layer did not


proceed far past this into the adverse region without separating.

Vous aimerez peut-être aussi