Вы находитесь на странице: 1из 10

Chapter 10

Boundary layer equations


10.1 The boundary layer over a at plate
Consider plane ow of a slightly viscous incompressible Newtonian uid over a solid surface lying
in the plane y = 0 . The ow will be governed by the conservation of mass condition (2.3) and
the NavierStokes equation (9.15)(9.16).
Following the solution for the impulsively slid plate ( 9.6), its plausible that the eects of
viscosity might be largely limited to a thin layer along the plate, 0 < y < , where the thickness
may depend on x and time. Outside this layer ( y ), we expect the ow to be essentially
the same as that of a perfect uid.
Assume that any changes in the ow in the tangential (x) direction occur over a length scale
L, which might be the length of a at plate, or chord of a wing section.
10.1.1 Scales in the conservation of mass
In the equation for the conservation of mass (2.3), if U is a representative scale for the tangential
speed, and L is a representative tangential length scale, then u/x should be of order U/L (by
application of the mean value theorem, for example). Similarly if V is a typical magnitude of
the normal component of velocity, then (expressing the same idea in mathematical notation)
v
y
= O
_
V

_
(10.1)
where is again the boundary layer thickness scale. If the two terms of the continuity equation
are to balance each other, they must have comparable magnitudes and therefore
O
_
u
x
_
+ O
_
v
y
_
= 0 (10.2)
O
_
U
L
_
+ O
_
V

_
= 0 . (10.3)
It follows that the velocity scales are related to the length scales by
V
U
= O
_

L
_
. (10.4)
105
106 AERODYNAMICS I COURSE NOTES, 2005
There is an obvious geometric interpretation of this relation in terms of the slope of stream-lines
(magnitude V/U ) and the shape of the boundary layer ( thick and L long).
If the boundary layer is thin, so that L, then (10.4) implies V U , which means that
the ow is basically parallel to the solid surface, as expected.
10.1.2 Scales in the conservation of momentum
Now apply the same technique, scale analysis, to the NavierStokes equations. Let t and p
be typical scales for the time and pressure, then (9.15)(9.16) imply:
O
_
U
t
_
+ O
_
U
2
L
_
+ O
_
UV

_
= O
_
p
L
_
+ O
_
U
L
2
_
+ O
_
U

2
_
(10.5)
O
_
V
t
_
+ O
_
UV
L
_
+ O
_
V
2

_
= O
_
p

_
+ O
_
V
L
2
_
+ O
_
V

2
_
. (10.6)
In all cases where L, the rst viscous terms can be neglected in comparison with the
second. Using this and (10.4) to eliminate V in favour of , we have
O
_
U
t
_
+ O
_
U
2
L
_
+ O
_
U
2
L
_
= O
_
p
L
_
+ O
_
U

2
_
(10.7)
O
_
U
Lt
_
+ O
_
U
2

L
2
_
+ O
_
U
2

L
2
_
= O
_
p

_
+ O
_
U
L
_
. (10.8)
10.1.3 The Reynolds number
Now, if we are to have a balance between the viscous term (of magnitude U/
2
) and the inertial
terms (of magnitude U
2
/L) in the tangential momentum balance (10.7), then the boundary
layer thickness scale must satisfy

L
= O
_
_
UL

_
1/2
_
. (10.9)
The dimensionless quantity in parentheses on the right-hand side is the Reynolds number
(based on tangential length scale L):
Re
L

UL


UL

. (10.10)
Since this Reynolds number is typically large in aerodynamical applicationsconsider ight
of a wing section of chord L = 1 m at U = 100 m/s in air of kinematic viscosity 1.5 10
5
m
2
/s ,
for which it is Re
L
7 10
6
the assumption that the boundary layer is thin is supported
by (10.9); this value of Re
L
implies that the boundary layer thickness is of order 4 10
4
times
the tangential length scale L.
10.1.4 Pressure in the boundary layer
If the pressure term in the tangential momentum balance (10.7) is to be comparable with the
inertial terms, the pressure must be of magnitude
p = O(U
2
) , (10.11)
as may have been expected from Bernoullis equation (2.15). Numerical factors like the half in
Bernoullis equation are usually omitted in order relations like (10.11).
Boundary layer equations 107
10.1.5 The normal momentum balance
If this pressure scale is now substituted into the normal momentum balance (10.8), we get
O
_
U
Lt
_
+ O
_
U
2

L
2
_
+ O
_
U
2

L
2
_
= O
_
U
2

_
+ O
_
U
L
_
, (10.12)
so that the ratio of the inertial to the pressure term is
U
2
/L
2
U
2
/
=

2
L
2
= Re
1
L
(10.13)
and the ratio of the viscous to the pressure term is (again)
U/L
U
2
/
=

UL
= Re
1
L
. (10.14)
So, in the normal momentum balance, both the inertial and viscous terms are much smaller than
the pressure term when the Reynolds number is large.
Again, from the tangential momentum balance (10.7), the timescale must be t = O(L/U)
(i.e. the time for the uid to ow along the boundary layer length of interest if its moving at
the characteristic speed). Then the ratio of the transient to the pressure term in the normal
momentum balance is (again)
U
2
/L
2
U
2
/
= Re
1
L
. (10.15)
This means that there are no terms in the normal momentum equation capable of balancing the
pressure term, and therefore that the pressure term itself must be small. Thus, approximately,
p
y
= 0 ; (10.16)
i.e. the pressure is uniform through a viscous boundary layer on a at surface.
This is an extremely useful result. It means that the pressure at any point in the boundary
layer is the same as that at the same longitudinal position x outside the boundary layer, and
is therefore the same as that which would prevail near the surface if the uid were perfect.
Assuming we can solve the ow problem for the perfect uid in the given geometry, the pressure
term in the tangential momentum equation is then a known force term.
10.1.6 The boundary layer momentum equation
With the neglect of the rst viscous term, the tangential momentum equation reduces to
u
t
+ u
u
x
+ v
u
y
=
1

p
x
+

2
u
y
2
, (10.17)
where p = p(x, t) and the pressure term is to be understood as known following the prelim-
inary perfect uid analysis. This equation, the boundary layer momentum equation, is to be
supplemented with boundary conditions.
Since the equation is second-order in y , two boundary conditions are required in the normal
direction; these are usually that the uid adhere to the solid surface and that the boundary layer
ow smoothly merge with the external inviscid ow outside the boundary layer
u = 0 (y = 0) (10.18)
u u

(y ) , (10.19)
108 AERODYNAMICS I COURSE NOTES, 2005
where u

is the tangential component of velocity at x that would be obtained by a perfect uid


in the same geometry.
Note that the tangential velocity and pressure obtained as outer conditions on the boundary
layer analysis (y ) are those at y = 0 in the perfect uid analysis. This is because the
boundary layer is very thin, on any meaningful scale in the perfect uid ow. One way to
make this clearer is to use y in the perfect uid analysis and introduce a new stretched normal
coordinate = y/ in the boundary layer analysis, as was done in 9.6.
Since the equation is rst-order in x, only one boundary condition is required in the tan-
gential direction. Usually information is prescribed about the upstream state and then the
equations integrated downstream to study the longitudinal development of the boundary layer.
This situation is quite dierent to the full NavierStokes equations which are second-order in all
directions and require boundary conditions both upstream and downstream. This simplication
is an essential feature of the boundary layer approximation.
10.1.7 Pressure and external tangential velocity
Since the ow outside the viscous boundary layer is presumed to be the same as that of a perfect
uid, the pressure and velocity there are related by Bernoullis equation (2.15), so, since q = u
for y = 0 in the perfect uid (as the surface is impermeable),
1

p
x
= u

x
. (10.20)
This is often more convenient. With it, the boundary layer momentum equation is
u
t
+ u
u
x
+ v
u
y
= u

x
+

2
u
y
2
. (10.21)
10.1.8 Application to curved surfaces
Although the analysis thus far has been for a plane surface, provided the radius of curvature of
the surface is much less than the boundary layer thickness, it applies well to curved surfaces. In
this case, x is to be interpreted as the tangential coordinate and y the normal. This means that
acceptable results can be obtained for typical wing sections by using the present model.
10.2 Momentum integral equation
Although the boundary layer momentum equation (10.21) can be solved analytically in special
congurations using various mathematical techniques, or numerically for almost any geometry,
much of the most useful information about the boundary layer (such as the skin friction) can
be approximated from the momentum integral equation. We consider this for steady boundary
layers.
To obtain the momentum integrate equation, integrate (10.21), without the time derivative
term, normal to the surface from 0 to (by which we mean to a distance far outside the
boundary layer on the scale of the boundary layer thickness, but still very near the surface on
the scale of the perfect uid ow) :
_

0
u
u
x
dy +
_

0
v
u
y
dy =
_

0
u

x
dy +
_

0

2
u
y
2
dy . (10.22)
Boundary layer equations 109
We want to eliminate v from this, which we can do, using the conservation of mass: inte-
grate (2.3) from 0 to y :
_
y
0
_
u
x
+
v
y
_
dy

= 0 (10.23)
_
y
0
u
x
(x, y

) dy

+ v = 0 , (10.24)
(noting that v = 0 at y = 0 for impermeability) so that
_

0
v
u
y
dy =
_

0
__
y
0
u
x
(x, y

) dy

_
u
y
dy (10.25)
which can be integrated by parts to give

_

0
__
y
0
u
x
(x, y

) dy

_
u
y
dy =
___
y
0
u
x
(x, y

) dy

_
u
_

0
+
_

0
u
u
x
dy (10.26)
= u

_

0
u
x
dy +
_

0
u
u
x
dy . (10.27)
For the viscous term,

_

0

2
u
y
2
dy =
_
u
y
_

0
=
u
y

y=0
(10.28)
since u/y 0 for y . The right-hand side can be written as
w
/ , where
w
stands
for the shear stress component
xy
at the wall (w for wall). In general
xy
=
_
v
x
+
u
y
_
, but
v vanishes along the impermeable solid surface so that on y = 0 we have
xy
= u/y , as
asserted.
Thus, the integrated momentum equation is
_

0
_
2u
u
x
u

u
x
u

x
_
dy =

. (10.29)
Now,

x
{u(u u

)} = u
u
x
u
u

x
+ u
u
x
u

u
x
(10.30)
= 2u
u
x
u

u
x
u
u

x
(10.31)
so the momentum integral equation can be rewritten
_

0
_

x
[u(u u

)] + (u u

)
u

x
_
dy =

(10.32)

x
_

0
u(u u

) dy +
u

x
_

0
(u u

) dy =

(10.33)

x
_
u
2

_

0
u
u

_
1
u
u

_
dy
_
+ u

x
_

0
_
1
u
u

_
dy =

w

. (10.34)
110 AERODYNAMICS I COURSE NOTES, 2005
10.3 Local boundary layer parameters
10.3.1 The displacement and momentum thicknesses
Lets dene (Abbott and von Doenho 1959, pp. 89, 92; Moran 2003, pp. 200, 202)

_

0
_
1
u
u

_
dy (10.35)

_

0
u
u

_
1
u
u

_
dy . (10.36)
We call

the displacement thickness and the momentum thickness. In terms of them, the
momentum integral equation can be written as

x
(u
2

) +

x
=

w

. (10.37)
Both

and depend on x , and both parameterize the boundary layer tangential velocity
prole u(y) , at that value of x.
The physical signicance of the displacement thickness can be understood as follows. If
the uid were perfect, the tangential velocity throughout the boundary right down to the solid
surface would be u

, and the volume ow (per unit span) along the boundary layer would be
_

0
u

dy ; however, in the boundary layer the actual value is


_

0
u dy so that the dierence is
_

0
(u

u) dy = u

(10.38)
(using the fact that the integrand is negligible for y > ). Thus, the retardation of the ow in
the boundary layer causes a defect in the volume ow rate the same as the subtraction of a layer
of thickness

.
For almost all boundary layer proles u(y) , the tangential velocity remains in 0 < u <
u

; therefore, the momentum thickness will be less than the displacement thickness, since the
integrand of contains the additional weighting factor
u
u
. The ratio of the two thicknesses
H

, (10.39)
called the shape factor (Abbott and von Doenho 1959, p. 92; Moran 2003, p. 204), is therefore
greater than one. This is illustrated in the examples of 10.3.3.
10.3.2 The skin friction coecient
Dene a dimensionless version of the shear stress at the wall as
c
f


w
1
2
u
2

; (10.40)
c
f
is called the (local) skin-friction coecient (Abbott and von Doenho 1959, p. 101; Moran
2003, p. 199).
Expressing the wall shear stress in terms of c
f
, the momentum integral equation is

x
+

u

(2 + H)
u

x
=
c
f
2
. (10.41)
Boundary layer equations 111
This is the most common and useful form of the K arm an integral relation (Abbott and von
Doenho 1959, p. 92; Moran 2003, p. 204). It provides one equation linking the three unknown
local boundary layer parameters,

, , and c
f
, to the known external speed variation u

(x) .
To proceed further, another two relations between the boundary layer thicknesses and skin
friction are required. Two popular approaches for deriving these are:
1. the assumption of a velocity prole;
2. the use of (possibly empirical) correlations.
Local skin friction in a turbulent boundary layer
Although the above derivation of the Karman integral relation was in the context of laminar
ow, the relation holds as well for turbulent boundary layers. In the laminar case, the local skin
friction can be obtained from an assumed velocity prole simply by evaluating its derivative at
the wall; in the turbulent case its more dicult, as turbulent boundary layers contain a number
of dierent sublayers and a prole function accurately describing the variation of the tangential
velocity over most of the boundary layer thickness might be of no use at all in determining the skin
friction coecient. An example is the much-used one-seventh prole, discussed below in 10.3.3.
Basically, if one takes a velocity prole approach to turbulent boundary layer momentum integral
analysis, a separate correlation is required for the local skin friction (Abbott and von Doenho
1959, p. 97).
10.3.3 Example boundary layer prole parameters
Example: linear prole
Consider the linear prole:
u
u

=
_
y

, 0 < y <
1 , < y .
(10.42)
The displacement and momentum thicknesses are

=
_

0
_
1
u
u

_
dy =
_
1
0
(1 ) d =

2
(10.43)
=
_

0
u
u

_
1
u
u

_
dy =
_
1
0
(1 ) d =

6
(10.44)
and the shape factor is therefore H = 3 . The integrands are plotted in gure 10.1.
The local skin friction coecient of the linear prole is
c
f
=

w
1
2
u
2

u
y

y=0
1
2
u
2

1
2
u
2

=
2
Re

, (10.45)
where
Re

(10.46)
is a Reynolds number based on the boundary layer thickness.
112 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

displacement
momentum
Figure 10.1: Linear boundary layer velocity prole: the displacement thickness is the area to the
right of the solid curve (the velocity prole), the momentum thickness to the right of the dashed
curve.
Example: error function prole
Consider the error function prole, which arose in the impulsively started sliding plate problem:
u
u

= erf
y

. (10.47)
(Note that we use erf instead of erfc as here we take the plate to be xed and the far uid to
be moving at u

.)
The displacement thickness can be carried out analytically (Gautschi 1965, pp. 299300):

=
_

0
(1 erf ) d =
1

.
= 0.56419 , (10.48)
though the (dimensionless) momentum thickness (/ ) is more easily integrated numerically:
octave> deltastar = quad ("1-erf(x)", 0, inf)
deltastar = 0.56419
octave> theta = quad ("erf(x)*(1-erf(x))", 0, inf)
theta = 0.23369
octave> H = deltastar / theta
H = 2.4142
The thickness integrands are illustrated in gure 10.2.
Boundary layer equations 113
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

displacement
momentum
Figure 10.2: Boundary layer velocity prole u = u

erf y/ : the displacement thickness is the


area to the right of the solid curve (the velocity prole), momentum thickness to the right of
dashed curve.
To calculate the skin friction, note:
d erf
d
=
2e

(10.49)

y
_
u

erf
y

y=0
=
2u

; (10.50)
thus
c
f
=
4/

Re

. (10.51)
Just as for the linear prole, the local skin friction coecient is inversely proportional to Re

.
In each case it would also have been proportional to the Reynolds number based on the local
boundary layer momentum thickness
Re

(10.52)
and this latter result applies to many laminar boundary layers (Moran 2003, p. 206); it is exploited
in Thwaitess method for solving the momentum integral equation (ibid.).
Example: one-seventh prole
A popular simple prole used to describe turbulent boundary layers is
u
u

=
_
_
y

_
1/7
, y <
1 , y > .
(10.53)
114 AERODYNAMICS I COURSE NOTES, 2005
0
0.5
1
1.5
2
0 0.2 0.4 0.6 0.8 1 1.2
N
O
R
M
A
L

D
I
S
T
A
N
C
E
,

y
/

TANGENTIAL VELOCITY, u/u

displacement
momentum
Figure 10.3: Boundary layer velocity prole u = u

(y/)
1/7
: the displacement thickness is the
area to the right of the solid curve (the velocity prole), momentum thickness to the right of
dashed curve.
The displacement and momentum thicknesses are

=
_
1
0
_
1
1/7
_
d =
1
8
(10.54)

=
_
1
0

1/7
_
1
1/7
_
d =
7
72
(10.55)
which means the shape function is H =
9
7
. The prole, illustrated in gure 10.3, is much
fuller than the two previous, which are more typical of laminar proles. This is apparent in the
much smaller shape factor. This fullness is caused by the eectiveness of the turbulent eddies
in transferring momentum across the boundary layer. A consequence, and another apparent
feature, is the steeper relative gradient at the wall; however, while this is qualitatively realistic,
the one-seventh prole is useless for estimating the local skin friction, since it implies an innite
velocity gradient at the wall.

Вам также может понравиться