Welty

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 24

144 Chapter 12 Viscous Flow

12.3 THE BOUNDARY-LAYER CONCEPT


The observation of a decreasing region of inuence of shear stress as the Reynolds number is
increased led Ludwig Prandtl to the boundary-layer concept in 1904. According to Prandtls
hypothesis, the effects of uid friction at high Reynolds numbers are limited to a thin layer
near the boundary of a body, hence the term boundary layer. Further, there is no signicant
pressure change across the boundary layer. This means that the pressure in the boundary
layer is the same as the pressure in the inviscid ow outside the boundary layer. The
signicance of the Prandtl theory lies in the simplication that it allows in the analytical
treatment of viscous ows. The pressure, for example, may be obtained from experiment or
inviscid ow theory. Thus the only unknowns are the velocity components.
The boundary layer on a at plate is shown in Figure 12.5. The thickness of the
boundary layer, d, is arbitrarily taken as the distance away from the surface where the
velocity reaches 99% of the free-stream velocity. The thickness is exaggerated for clarity.

Region of transition

Laminar Turbulent
boundary-layer boundary-layer
region region

n

Streamline

Figure 12.5 Boundary layer


Laminar sublayer
on a at plate. (The thickness
is exaggerated for clarity.)
x

Figure 12.5 illustrates how the thickness of the boundary layer increases with
distance x from the leading edge. At relatively small values of x, ow within the
boundary layer is laminar, and this is designated as the laminar boundary-layer
region. At larger values of x the transition region is shown where uctuations
between laminar and turbulent ows occur within the boundary layer. Finally, for
a certain value of x, and above, the boundary layer will always be turbulent. In the
region in which the boundary layer is turbulent, there exists, as shown, a very thin lm
of uid called the laminar sublayer, wherein ow is still laminar and large velocity
gradients exist.
The criterion for the type of boundary layer present is the magnitude of Reynolds
number, Rex, known as the local Reynolds number, based on the distance x from the leading
edge. The local Reynolds number is dened as
xvr
Rex  (12-4)
m
For ow past a at plate, as shown in Figure 12.5, experimental data indicate that for
(a) Rex < 2  105 the boundary layer is laminar,
(b) 2  105 < Rex < 3  106 the boundary layer may be either laminar or turbulent,
(c) 3  106 < Rex the boundary layer is turbulent
12.4 The Boundary-Layer Equations 145

12.4 THE BOUNDARY-LAYER EQUATIONS


The concept of a relatively thin boundary layer at high Reynolds numbers leads to some
important simplications of the NavierStokes equations. For incompressible, two-
dimensional ow over a at plate, the NavierStokes equations are

@v x @v x @v x @s xx @tyx
r vx vy (12-5)
@t @x @y @x @y

and

@v y @v y @v y @t xy @s yy
r vx vy (12-6)
@t @x @y @x @y

where txy tyx m(@v x /@y @v y /@x), s xx P 2m(@v x /@x) and s yy P 2m(@v y /
@y). The shear stress in a thin boundary layer is closely approximated by m(@v x /@y). This
can be seen by considering the relative magnitudes of @v x /@y and @v y /@x From Figure 12.5,
we may write v x jd /v y jd
O(x/d), where O signies the order of magnitude. Then
   
@v x v x jd @v y v y jd

O
O
@y d @x x

so

@v x /@y x2

O
@v y /@x d
which, for a relatively thin boundary layer, is a large number, and thus @v x /@y 
@v y /@x: The normal stress at a large Reynolds number is closely approximated by
the negative of the pressure as m(@v x /@x)
O(mv 1 /x) O(rv 21 /Rex ); therefore
s xx s yy  P: When these simplications in the stresses are incorporated, the
equations for ow over a at plate become
 
@v x @v x @v x @P @ 2 vx
r vx vy  m 2 (12-7)
@t @x @y @x @y
and
 
@v y @v y @v y @P @2vy
r vx vy  m (12-8)
@t @x @y @y @x2

Furthermore,2 the terms in the second equation are much smaller than those in the rst
equation, and thus @P/@y 0; hence @P/@x dP/dx; which according to Bernoullis
equation is equal to rv 1 dv 1 /dx:
The nal form of equation (12-7) becomes
@v x @v x @v x dv 1 @2vx
vx vy v1 n 2 (12-9)
@t @x @y dx @y

2
The order of magnitude of each term may be considered as above. For example, v x (@v y /@x)
O(v 1 (v 1 /x)
(d/x)) O(v 21 d/x2 ):
146 Chapter 12 Viscous Flow

The above equation, and the continuity equation


@v x @v y
0 (12-10)
@x @y
are known as the boundary-layer equations.

12.5 BLASIUSS SOLUTION FOR THE LAMINAR


BOUNDARY LAYER ON A FLAT PLATE
One very important case in which an analytical solution of the equations of motion has been
achieved is that for the laminar boundary layer on a at plate in steady ow.
For ow parallel to a at surface, v 1 (x) v 1 and dP/dx 0, according to the
Bernoulli equation. The equations to be solved are now the following:
@v x @v x @2vx
vx vy n 2 (12-11a)
@x @y @y

and
@v x @v y
0 (12-11b)
@x @y
with boundary conditions v x v y 0 at y 0, and v x v 1 at y 1.
Blasius3 obtained a solution to the set of equations (12-11) by rst introducing the
stream function, C, as described in Chapter 10, which automatically satises the two-
dimensional continuity equation, equation (12-11b). This set of equations may be reduced
to a single ordinary differential equation by transforming the independent variables x, y, to h
and the dependent variables from C(x, y) to f(h)
where
y v 1 1/2
h(x; y) (12-12)
2 nx
and
C(x; y)
f (h) (12-13)
(nxv )1/2
1

The appropriate terms in equation (12-11a) may be determined from equations (12-12)
and (12-13). The following expressions will result. The reader may wish to verify the
mathematics involved.
@C v 1 0
vx f (h) (12-14)
@y 2
@C 1 nv 1 1/2
vy  (h f 0  f ) (12-15)
@x 2 x
@v x v 1 h 00
 f (12-16)
@x 4x
@v x v 1 v 1 1/2 00
f (12-17)
@y 4 nx
@ 2 v x v 1 v 1 000
f (12-18)
@y2 8 nx

3
H. Blasius, Grenzshichten in Flussigkeiten mit kleiner Reibung, Z. Math. U. Phys. Sci., 1, 1908.
12.5 Blasiuss Solution for the Laminar Boundary Layer on a Flat Plate 147

Substitution of (12-14) through (12-18) into equation (12-11a) and cancellation gives, as
a single ordinary differential equation
f 000 f f 00 0 (12-19)
with the appropriate boundary conditions
f f0 0 at h 0
0
f 2 at h 1
Observe that this differential equation, although ordinary, is nonlinear and that, of
the end conditions on the variable f (h), two are initial values and the third is a boundary
value. This equation was solved rst by Blasius, using a series expansion to express the
function, f (h), at the origin and an asymptotic solution to match the boundary condition
at h 1. Howarth4 later performed essentially the same work but obtained more
accurate results. Table 12.1 presents the signicant numerical results of Howarth. A plot
of these values is included in Figure 12.6.
A simpler way of solving equation (12-19) has been suggested in Goldstein5 who
presented a scheme whereby the boundary conditions on the function f are initial values.

Table 12.1 Values of f, f 0 , f 00 , and vx /v1 for laminar ow parallel to a at plate (after Howarth)
r
y v1 vx
h f f0 f 00
2 nx v1

0 0 0 1.32824 0
0.2 0.0266 0.2655 1.3260 0.1328
0.4 0.1061 0.5294 1.3096 0.2647
0.6 0.2380 0.7876 1.2664 0.3938
0.8 0.4203 1.0336 1.1867 0.5168
1.0 0.6500 1.2596 1.0670 0.6298
1.2 0.9223 1.4580 0.9124 0.7290
1.4 1.2310 1.6230 0.7360 0.8115
1.6 1.5691 1.7522 0.5565 0.8761
1.8 1.9295 1.8466 0.3924 0.9233
2.0 2.3058 1.9110 0.2570 0.9555
2.2 2.6924 1.9518 0.1558 0.9759
2.4 3.0853 1.9756 0.0875 0.9878
2.6 3.4819 1.9885 0.0454 0.9943
2.8 3.8803 1.9950 0.0217 0.9915
3.0 4.2796 1.9980 0.0096 0.9990
3.2 4.6794 1.9992 0.0039 0.9996
3.4 5.0793 1.9998 0.0015 0.9999
3.6 5.4793 2.0000 0.0005 1.0000
3.8 5.8792 2.0000 0.0002 1.0000
4.0 6.2792 2.0000 0.0000 1.0000
5.0 8.2792 2.0000 0.0000 1.0000

4
L. Howarth, On the solution of the laminar boundary layer equations, Proc. Roy. Soc. London, A164 547
(1938).
5
S. Goldstein, Modern Developments in Fluid Dynamics, Oxford Univ. Press, London, 1938, p. 135.
148 Chapter 12 Viscous Flow

1.0

0.8

nx /v 0.6

0.4
Blasius theory

0.2

0
0 1.0 2.0 3.0
h = 1
y v
2 nx
Figure 12.6 Velocity distribution in the laminar boundary layer over a at plate. Experimental
data by J. Nikuradse (monograph, Zentrale F. wiss. Berichtswesen, Berlin, 1942) for the Reynolds
number range from 1:08  105 to 7:28  105 .

If we dene two new variables in terms of the constant, C, so that


f f /C (12-20)
and
j Ch (12-21)
then the terms in equation (12-19) become
f (h) Cf(j) (12-22)
f 0 C 2 f0 (12-23)
00 3 00
f C f (12-24)
and
f 000 C4 f000 (12-25)
The resulting differential equation in f(j) becomes
f000 ff00 0 (12-26)
and the initial conditions on f are
f0 f0 0 f00 ? at j 0
The other boundary condition may be expressed as follows:
f 0 (h) 2
f0 (j) 2
2 at j 1
C C
An initial condition may be matched to this boundary condition if we let f 00 (h 0) equal
some constant A; then f00 (j 0) A/C3 : The constant A must have a certain value to
12.5 Blasiuss Solution for the Laminar Boundary Layer on a Flat Plate 149

satisfy the original boundary condition on f 0 . As an estimate we let f00 (j 0) 2; giving


A 2C 3 : Thus initial values of f, f0 , and f00 are now specied. The estimate on f00 (0)
requires that
 2/3
0 2 2
f (1) 2 2 (12-27)
C A
Thus equation (12-26) may be solved as an initial-value problem with the answer scaled
according to equation (12-27) to match the boundary condition at h 1.
The signicant results of Blasiuss work are the following:
(a) The boundary thickness, d, is obtained from Table 12.1. When h 2:5, we have
v x /v 1 0:99 thus, designating y d at this point, we have
r r
y v1 d v1
h 2:5
2 nx 2 nx
and thus
r
nx
d5
v1
or
d 5 5
r p (12-28)
x v1x Rex
n

(b) The velocity gradient at the surface is given by equation (12-27):


 r
@v x  v 1 v 1 1/2 00 v1
f (0) 0:332 v (12-29)
@y y0
1
4 nx nx
As the pressure does not contribute to the drag for flow over a flat plate, all the drag is
viscous. The shear stress at the surface may be calculated as

@v x 
t0 m
@y  y0

Substituting equation (12-29) into this expression, we have


r
v1
t0 m 0:332 v 1 (12-30)
nx
The coefcient of skin friction may be determined by employing equation (12-2) as
follows: r
v1
0:332mv 1
t Fd /A nx
Cfx  2 2
rv 1 /2 rv 1 /2 rv 21 /2
r
n
0:664
xv 1
0:664
Cfx p 12-31
Rex
Equation (12-31) is a simple expression for the coefcient of skin friction at a particular
value of x. For this reason the symbol Cfx is used, the x subscript indicating a local coefcient.
150 Chapter 12 Viscous Flow

While it is of interest to know values of Cfx, it is seldom that a local value is useful; most
often one wishes to calculate the total drag resulting from viscous ow over some surface of
nite size. The mean coefcient of skin friction that is helpful in this regard may be
determined quite simply from Cfx according to
Z
rv 21 rv 21
Fd ACfL Cfx dA
2 2 A
or the mean coefcient, designated CfL, is related to Cfx by
Z
1
CfL Cfx dA
A A
For the case solved by Blasius, consider a plate of uniform width W, and length L, for which
Z Z r
1 L 1 L n 1/2
CfL Cfx dx 0:664 x dx
L 0 L 0 v1
r
n
1:328
Lv 1
1:328
CfL p (12-32)
ReL

12.6 FLOW WITH A PRESSURE GRADIENT


In Blasiuss solution for laminar ow over a at plate, the pressure gradient was zero. A
much more common ow situation involves ow with a pressure gradient. The pressure
gradient plays a major role in ow separation, as can be seen with the aid of the boundary-
layer equation (12-7). If we make use of the boundary, conditions at the wall v x v y 0, at
y 0 equation 12-7 becomes

@ 2 v x  dP
m 2 (12-33)
@y y0 dx
which relates the curvature of the velocity profile at the surface to the pressure gradient.
Figure 12.7 illustrates the variation in v x, @v x /@y, and @ 2 v x /@y2 across the boundary layer
for the case of a zero-pressure gradient.

y y y

vx vx 2 vx +
y y2
Figure 12.7 Variation in velocity and velocity derivatives across the laminar boundary layer
when dP/dx 0.
12.6 Flow with a Pressure Gradient 151

When dP/dx 0, the second derivative of the velocity at the wall must also be zero;
hence the velocity prole is linear near the wall. Further, out in the boundary layer, the
velocity gradient becomes smaller and gradually approaches zero. The decrease in
the velocity gradient means that the second derivative of the velocity must be negative.
The derivative @ 2 v x /@y2 is shown as being zero at the wall, negative within the boundary
layer, and approaching zero at the outer edge of the boundary layer. It is important to note
that the second derivative must approach zero from the negative side as y ! d. For values of
dP/dx 6 0, the variation in v x and its derivatives is shown in Figure 12.8.

y y y

dP
dx
dP
dx

dP dP
dx dP dx
dP
dx dx
vx vx 2 vx +
y y2
Figure 12.8 Variation in v x and its derivatives across the boundary layer for various pressure
gradients.

A negative pressure gradient is seen to produce a velocity variation somewhat similar to


that of the zero-pressure-gradient case. A positive value of dP/dx, however, requires a
positive value of @ 2 v x /@y2 at the wall. As this derivative must approach zero from the
negative side, at some point within the boundary layer the second derivative must equal zero.
A zero second derivative, it will be recalled, is associated with an inection point. The
inection point is shown in the velocity prole of Figure 12.8. We may now turn our
attention to the subject of ow separation.
In order for ow separation to occur, the velocity in the layer of uid adjacent to the
wall must be zero or negative, as shown in Figure 12.9. This type of velocity prole is seen

Separated
d region
Figure 12.9 Velocity
proles in separated-
Separation point ow region.
152 Chapter 12 Viscous Flow

to require a point of inection. As the only type of boundary-layer ow that has an


inection point is ow with a positive pressure gradient, it may be concluded that a
positive pressure gradient is necessary for ow separation. For this reason a positive
pressure gradient is called an adverse pressure gradient. Flow can remain unseparated
with an adverse pressure gradient, thus dP/dx > 0 is a necessary but not a sufcient
condition for separation. In contrast a negative pressure gradient, in the absence of sharp
corners, cannot cause ow separation. Therefore, a negative pressure gradient is called a
favorable pressure gradient.
The presence of a pressure gradient also affects the magnitude of the skin friction
coefcient, as can be inferred from Figure 12.8. The velocity gradient at the wall increases as
the pressure gradient becomes more favorable.

RMA
12.7 VON KA N MOMENTUM INTEGRAL ANALYSIS
The Blasius solution is obviously quite restrictive in application, applying only to the case of
a laminar boundary layer over a at surface. Any situation of practical interest more
complex than this involves analytical procedures that have, to the present time, proved
inferior to experiment. An approximate method providing information for systems involv-
ing other types of ow and having other geometries will now be considered.
Consider the control volume in Figure 12.10. The control volume to be analyzed is
of unit depth and is bounded in the xy plane by the x axis, here drawn tangent to the
surface at point 0; the y axis, the edge of the boundary layer, and a line parallel to the y
axis a distance Dx away. We shall consider the case of two-dimensional, incompressible
steady ow.

vy
Control volume
y
vx

Stream lines

x
d 0
x

Figure 12.10 Control volume for integral analysis of the boundary layer.

A momentum analysis of the dened control volume involves the application of the
x-directional scalar form of the momentum theorem
ZZ ZZZ
@
Fx :
v x r(v n) dA v x r dV (5-5a)
c:s: @t c:v:

A term-by-term analysis of the present problem yields the following:


 
Pj  Pjx
Fx Pdjx  PdjxDx Pjx xDx (djxDx  djx )  t0 Dx
2
where d represents the boundary-layer thickness, and both forces are assumed
negligible. The above terms represent the x-directional pressure forces on the
12.7 Von Karman Momentum Integral Analysis 153

left, right, and top sides of the control volume, and the frictional force on the bottom,
respectively.
The surface integral term becomes
ZZ Z  Z 
d  d 
v x r(v: n)dA rv 2x dy  rv 2x dy v 1 m_ top
c:s: 0 xDx 0 x

and the accumulation term is


ZZZ
@
v x r dV 0
@t c:v:

as this is a steady-ow situation.


An application of the integral equation for conservation of mass will give
ZZ ZZZ
@
r(v: n) dA r dV 0 (4-1)
c:s: @t c:v:

ZZ Z  Z 
d  d 
r(v: n) dA 
rv x dy  rv x dy m_ top
c:s: 0 xDx 0 x

ZZZ
@
r dV 0
@t c:v:

and the mass-ow rate into the top of the control volume, m_ top , may be evaluated as
Z d  Z d 
 
m_ top 
rv x dy  rv x dy (12-34)
0 xDx 0 x

The momentum expression, including equation (12-34), now becomes


 
PjxDx  Pjx
(PdjxDx  Pdjx ) Pjx (djxDx  djx )  t 0 Dx
2
Z d  Z d  Z d  Z d 
   

rv x dyxDx 
2 
rv x dy v 1
2 
rv x dy  rv x dy
0 0 x 0 xDx 0 x

Rearranging this expression and dividing through by Dx, we get

      
PjxDx  Pjx PjxDx  Pjx djxDx  djx Pdjx  Pdjx
 djxDx
Dx 2 Dx Dx
Rd 2 Rd 2 ! Rd Rd !
rv x dyj xDx  rv x dyjx rv x dyjxDx  rv x dyj x
0 0
 v1 0 0
t0
Dx Dx

Taking the limit as Dx!0 we obtain


Z d Z d
dP d d
d t0 rv 2x dy  v 1 rv x dy (12-35)
dx dx 0 dx 0
154 Chapter 12 Viscous Flow

The boundary-layer concept assumes inviscid ow outside the boundary layer, for
which we may write Bernoullis equation

dP dv 1
rv 1 0
dx dx
which may be rearranged to the form
d dP d d
(dv 2 )  v 1 (dv 1 ) (12-36)
r dx dx 1 dx

Notice that the left-hand sides of equations (12-35) and (12-36) are similar. We may thus
relate the right-hand sides and, with proper rearrangement, get the result
 Z d Z d
t0 d d
v1 (v 1  v x ) dy v x (v 1  v x ) dy (12-37)
r dx 0 dx 0

Equation (12-37) is the von Karman momentum integral expression, named in honor of
Theodore von Karman who rst developed it.
Equation (12-37) is a general expression whose solution requires a knowledge of the
velocity, v x , as a function of distance from the surface, y. The accuracy of the nal result will
depend on how closely the assumed velocity prole approaches the real one.
As an example of the application of equation (12-37), let us consider the case
of laminar ow over a at plate, a situation for which an exact answer is known. In
this case the free-stream velocity is constant, therefore (d/dx)v 1 0 and equation
(12-36) simplies to
Z d
t0 d
v x (v 1  v x ) dy (12-38)
r dx 0

An early solution to equation (12-38) was achieved by Pohlhausen, who assumed for
the velocity prole a cubic function
v x a by cy2 dy3 (12-39)
The constants a, b, c, and d may be evaluated if we know certain boundary conditions that
must be satised in the boundary layer. These are

(1) vx 0 at y 0
(2) vx v1 at y d
@v x
(3) 0 at y d
@y
and

@2vx
(4) 0 at y 0
@y2
Boundary condition (4) results from equation (12-33), which states that the second
derivative at the wall is equal to the pressure gradient. As the pressure is constant in this case,
@ 2 v x /@y2 0. Solving for a, b, c, and d from these conditions, we get

3 v1
a0 b v1 c0 d
2d 2d3
12.8 Description of Turbulence 155

which, when substituted in equation (12-39), give the form of the velocity prole

vx 3 y 1 y3
 (12-40)
v1 2 d 2 d

Upon substitution, equation (12-38) becomes


Z   
3v v 1 d d
3 y 1 y3 3 y 1 y3
v 21  1 dy
2 d1 dx 0 2d 2 d 2d 2 d

or, after integrating


3 v1 39 d 2
v (v d)
2 d 280 dx 1
As the free-stream velocity is constant, a simple ordinary differential equation in d results
140 v dx
d dd
13 v 1
This, upon integration, yields
d 4:64
p (12-41)
x Rex
The local skin-friction coefficient, Cfx , is given by
t0 2v 3 v 1 0:646
Cfx  1 2
2 p (12-42)
rv
2 1
v 1 2 d Rex
Integration of the local skin-friction coefficient between x 0 and x L as in equation
(12-32) yields
1:292
CfL p (12-43)
ReL
Comparing equations (12-41), (12-42), and (12-43) with exact results obtained by Blasius
for the same situation, equations (12-28), (12-31), and (12-32), we observe a difference of
about 7% in d and 3% in Cf . This difference could, of course, have been smaller had the
assumed velocity profile been a more accurate representation of the actual profile.
This comparison has shown the utility of the momentum integral method for the
solution of the boundary layer and indicates a procedure that may be used with reasonable
accuracy to obtain values for boundary-layer thickness and the coefcient of skin friction
where an exact analysis is not feasible. The momentum integral method may also be used to
determine the shear stress from the velocity prole.

12.8 DESCRIPTION OF TURBULENCE


Turbulent ow is the most frequently encountered type of viscous ow, yet the analytical
treatment of turbulent ow is not nearly well developed as that of laminar ow. In this
section, we examine the phenomenon of turbulence, particularly with respect to the
mechanism of turbulent contributions to momentum transfer.
In a turbulent ow the uid and ow variables vary with time. The instantaneous
velocity vector, for example, will differ from the average velocity vector in both magnitude
and direction. Figure 12.11 illustrates the type of time dependence experienced by the
156 Chapter 12 Viscous Flow

v v

_ _
v(x, y, z) v(x, y, z, t)

t t
(a) (b)
Figure 12.11 Time dependence of velocity in a turbulent ow: (a) steady mean ow;
(b) unsteady mean ow.

axial component of the velocity for turbulent ow in a tube. While the velocity in Figure
12.11a is seen to be steady in its mean value, small random uctuations in velocity occur
about the mean value. Accordingly, we may express the uid and ow variables in terms of a
mean value and a uctuating value. For example, the x-directional velocity is expressed as

v x v x (x; y; z) v 0x (x; y; z; t) (12-44)


Here vx (x; y; z) represents the time-averaged velocity at the point (x, y, z)
Z
1 t1
vx v x (x; y; z; t) dt (12-45)
t1 0
where t1 is a time that is very long in comparison with the duration of any fluctuation.
The mean value of v 0x (x; y; z; t) is zero, as expressed by
Z
0 1 t1 0
vx v (x; y; z; t) dt 0 (12-46)
t1 0 x
Hereafter, Q will Rbe used to designate the time average of the general property, Q,
t
according to Q 1/t1 01 Q(x; y; z; t)dt. While the mean value of the turbulent uctuations
is zero, these uctuations contribute to the mean value of certain ow quantities. For
example, the mean kinetic energy per unit volume is

1
KE r(v x v x0 )2 (v y v y0 )2 (v z v z0 )2 
2
The average of a sum is the sum of the averages; hence the kinetic energy becomes

1 n o
KE r v 2x 2v x v x0 v x02 v 2y 2v y v y0 v y02 v 2z 2v z v z0 v z02
2
or, since v x v x0 v x v x0 0;
1
KE r v 2x v 2y v 2z v x02 v y02 v z02 (12-47)
2
A fraction of the total kinetic energy of a turbulent flow is seen to be associated with the
magnitude of the turbulent fluctuations. It can be shown that the rms (root mean square)
value of the fluctuations, (v x02 v y02 v z02 )1/2 is a significant quantity. The level or
intensity of turbulence is defined as
q

v x02 v y02 v z02 /3
I (12-48)
v1
12.9 Turbulent Shearing Stresses 157

where v 1 is the mean velocity of the flow. The intensity of turbulence is a measure of
the kinetic energy of the turbulence and is an important parameter in flow simulation. In
model testing, simulation of turbulent flows requires not only duplication of Reynolds
number but also duplication of the turbulent kinetic energy. Thus, the measurement of
turbulence is seen to be a necessity in many applications.
The general discussion so far has indicated the uctuating nature of turbulence. The
random nature of turbulence lends itself to statistical analysis. We shall now turn our
attention to the effect of the turbulent uctuations on momentum transfer.

12.9 TURBULENT SHEARING STRESSES


In Chapter 7, the random molecular motion of the molecules was shown to result in a net
momentum transfer between two adjacent layers of uid. If the (molecular) random motions
give rise to momentum transfer, it seems reasonable to expect that large-scale uctuations,
such as those present in a turbulent ow, will also result in a net transfer of momentum.
Using an approach similar to that of Section 7.3, let us consider the transfer of momentum in
the turbulent ow illustrated in Figure 12.12.

y
vx =
vx ( y) vy' vx'

y

x

x
Figure 12.12 Turbulent motion at the surface of a control volume.

The relation between the macroscopic momentum ux due to the turbulent uctuations
and the shear stress may be seen from the control-volume expression for linear momentum
ZZ ZZZ
@
F vr(v: n) dA vr dV (5-4)
c:s: @t c:v:

The flux of x-directional momentum across the top of the control surface is
ZZ ZZ
vr(v: n) dA v 0y r(v x v 0x ) dA (12-49)
top top

If the mean value of the momentum flux over a period of time is evaluated for the case of
steady mean flow, the time derivative in equation (5-4) is zero; thus
ZZ ZZ 0 ZZ
Fx 0 0
v y r(v x v x )dA %
0
v y rv x dA rv 0y v 0x dA (12-50)

The presence of the turbulent fluctuations is seen to contribute a mean x directional


momentum flux of rv x0 v y0 s per unit area. Although the turbulent fluctuations are functions
158 Chapter 12 Viscous Flow

of position and time, their analytical description has not been achieved, even for the
simplest case. The close analogy between the molecular exchange of momentum in
laminar flow and the macroscopic exchange of momentum in turbulent flow suggests that
the term rv 0x v 0y be regarded as a shear stress. Transposing this term to the left-hand side of
equation (5-4) and incorporating it with the shear stress due to molecular momentum
transfer, we see that the total shear stress becomes
d vx
t yx m  rv x0 v y0 (12-51)
dy
The turbulent contribution to the shear stress is called the Reynolds stress. In turbulent
flows it is found that the magnitude of the Reynolds stress is much greater than the
molecular contribution except near the walls.
An important difference between the molecular and turbulent contributions to the shear
stress is to be noted. Whereas the molecular contribution is expressed in terms of a property
of the uid and a derivative of the mean ow, the turbulent contribution is expressed solely
in terms of the uctuating properties of the ow. Further, these ow properties are not
expressible in analytical terms. While Reynolds stresses exist for multidimensional
ows,6 the difculties in analytically predicting even the one-dimensional case have proved
insurmountable without the aid of experimental data. The reason for these difculties may
be seen by examining the number of equations and the number of unknowns involved. In the
incompressible turbulent boundary layer, for example, there are two pertinent equations,
momentum and continuity, and four unknowns, v x , v y , v x0 , and v y0 .
An early attempt to formulate a theory of turbulent shear stress was made by
Boussinesq.7 By analogy with the form of Newtons viscosity relation, Boussinesq
introduced the concept relating the turbulent shear stress to the shear strain rate. The shear
stress in laminar ow is tyx m(dv x /dy); thus by analogy, the Reynolds stress becomes
d vx
(tyx )turb At
dy
where At is the eddy viscosity. Subsequent renements have led to the introduction of the
eddy diffusivity of momentum, eM  At /r, and thus
d vx
(tyx )turb reM (12-52)
dy
The difficulties in analytical treatment still exist, however, as the eddy diffusivity, eM , is a
property of the flow and not of the fluid. By analogy with the kinematic viscosity in a
laminar flow, it may be observed that the units of the eddy diffusivity are L2/t.

12.10 THE MIXING-LENGTH HYPOTHESIS


A general similarity between the mechanism of transfer of momentum in turbulent ow
and that in laminar ow permits an analog to be made for turbulent shear stress. The analog
to the mean free path in molecular momentum exchange for the turbulent case is the
mixing length proposed by Prandtl8 in 1925. Consider the simple turbulent ow shown in
Figure 12.13.

6
The existence of the Reynolds stresses may also be shown by taking the time average of the NavierStokes
equations.
7
J. Boussinesq, Mem. Pre. par div. Sav., XXIII, (1877).
8
L. Prandtl, ZAMM, 5, 136 (1925).
12.10 The Mixing-Length Hypothesis 159

L
y

Figure 12.13 The Prandtl mixing


vx length.

The velocity uctuation v x0 is hypothesized as being due to the y-directional motion of a


lump of uid through a distance L. In undergoing translation the lump of uid retains the
mean velocity from its point of origin. Upon reaching a destination, a distance L from the
point of origin, the lump of uid will differ in mean velocity from that of the adjacent uid by
an amount v x jyL  v x jy . If the lump of uid originated at y L, the velocity difference
would be v x jyL  v x jy . The instantaneous value of v 0x jy is then v x jyL  v x jy , the sign of L,
of course, depending on the point of origin with respect to y. Further, the mixing length,
although nite, is assumed to be small enough to permit the velocity difference to be written
as and thus
dv x
v x jyL  v x jy L
dy
and thus
dv x
v 0x L (12-52)
dy
The concept of the mixing length is somewhat akin to that of the mean free path of a
gas molecule. The important differences are its magnitude and dependence upon ow
properties rather than uid properties. With an expression for v 0x at hand, an expression for v 0y
is necessary to determine the turbulent shear stress, rv x0 v y0 .
Prandtl assumed that v 0x must be proportional to v y0 . If v x0 and v y0 were completely
independent, then the time average of their product would be zero. Both the continuity
equation and experimental data show that there is some degree of proportionality between v x0
and v y0 . Using the fact that v y0
v x0 , Prandtl expressed the time average, v x0 v y0 , as
 
 
2 d v x  d v x
0 0
v x v y (constant)L  (12-53)
dy  dy
The constant represents the unknown proportionality between v x0 and v y0 as well as their
correlation in taking the time average. The minus sign and the absolute value were
introduced to make the quantity v x0 v y0 agree with experimental observations. The constant
in (12-53), which is unknown, may be incorporated into the mixing length, which is also
unknown, giving
 
 
2 d v x  d v x
0 0
v x v y L  (12-54)
dy  dy
Comparison with Boussinesqs expression for the eddy diffusivity yields
 
d v x 
eM L2   (12-55)
dy 
160 Chapter 12 Viscous Flow

At rst glance it appears that little has been gained in going from the eddy viscosity to
the mixing length. There is an advantage, however, in that assumptions regarding the nature
and variation of the mixing length may be made on an easier basis than assumptions
concerning the eddy viscosity.

12.11 VELOCITY DISTRIBUTION FROM THE MIXING-LENGTH THEORY


One of the important contributions of the mixing-length theory is its use in correlating
velocity proles at large Reynolds numbers. Consider a turbulent ow as illustrated in
Figure 12.13. In the neighborhood of the wall the mixing length is assumed to vary
directly with y, and thus L Ky, where K remains a dimensionless constant to be
determined via experiment. The shear stress is assumed to be entirely due to turbul-
ence and to remain constant over the region of interest. The velocity v x is assumed to
increase in the y direction, and thus d v x /dy jd v x /dyj. Using these assumptions, we may
write the turbulent shear stress as
 
d vx 2
t yx rK y
2 2
t0 (a constant)
dy
or
p
d vx t0 /r

dy Ky
p
The quantity t 0 /r is observed to have units of velocity. Integration of the above equation
yields
p
t0 /r
vx In y C (12-56)
K
where C is a constant of integration. This constant may be evaluated by setting v x
v x max at y h, whereby

v x max  v x 1h yi
p  ln (12-57)
t0 /r K h

The constant K was evaluated by Prandtl9 and Nikuradse10 from data on turbulent
ow in tubes and found to have a value of 0.4. The agreement of experimental data for
turbulent ow in smooth tubes with equation (12-57) is quite good, as can be seen from
Figure 12.14.
The empirical nature of the preceding discussion cannot be overlooked. Several
assumptions regarding the ow are known to be incorrect for ow in tubes, namely
that the shear stress is not constant and that the geometry was treated from a two-
dimensional viewpoint rather than an axisymmetric viewpoint. In view of these obvious
difculties, it is remarkable that equation (13-15) describes the velocity prole so
well.

9
L. Prandtl, Proc. Intern. Congr. Appl. Mech., 2nd Congr., Zurich (1927), 62.
10
J. Nikuradse, VDI-Forschungsheft, 356, 1932.
12.12 The Universal Velocity Distribution 161

12

11
Data at Re = 106
10

8
(vx max vx) 7
t0 /r
6

5
Equation (12-57)
4

1
Figure 12.14 Comparison of
0
0 0.2 0.4 0.6 0.8 1.0 data for ow in smooth tube
y/h with equation (12-57).

12.12 THE UNIVERSAL VELOCITY DISTRIBUTION


For turbulent ow in smooth tubes, equation (12-57) p may be taken as a basis for a more
general development. Recalling that the term
p t0 /r has the units of velocity, we may
introduce a dimensionless velocity v x / t0 /r. Dening
vx
v  p (12-58)
t 0 /r
we may write equation (12-56) as
1
v ln y C (12-59)
K
The left-hand side of (12-59) is, of course, dimensionless; therefore the right-hand side of
this equation must also be dimensionless. A pseudo-Reynolds number is found useful in
this regard. Defining
p
t0 /r
y  y (12-60)
n
we find that equation (12-59) becomes
1 ny 1
v ln p C (ln y ln b) (12-61)
K t0 /r K
where the constant b is dimensionless.
Equation (12-61) indicates that for ow in smooth tubes v f (y ) or
p

vx y t 0 /r
v  p f ln (12-62)
t0 /r n
The range of validity of equation (13-19) may be observed from a plot (see Figur 12.15)
of v versus ln y+, using the data of Nikuradse and Reichardt.
162 Chapter 12 Viscous Flow

25
Laminar Buffer Turbulant
layer layer layer

20

Equation
15 (12-63)
v+
Equation (12-65)
10

Equation (12-64) Nikuradse


5
Reichardt

0
1 2 5 10 20 50 100 200 500 1000
y+
Figure 12.15 Velocity correlation for ow in circular smooth tubes at high Reynolds number
(H. Reichardt, NACA TM1047, 1943).

Three distinct regions are apparent: a turbulent core, a buffer layer, and a laminar
sublayer. The velocity is correlated as follows:
for turbulent core, y  30
v 5:5 2:5 ln y (12-63)
for the buffer layer, 30  y  5
v 3:05 5 ln y (12-64)
for the laminar sublayer, 5 > y > 0
v y (12-65)
Equations (12-63) through (12-65) dene the universal velocity distribution. Because
of the empirical nature of these equations, there are, of course, inconsistencies. The velocity
gradient, for example, at the center of the tube predicted by (12-63) is not zero. In spite of
this and other inconsistencies, these equations are extremely useful for describing ow in
smooth tubes.
In rough tubes, the scale of the roughness e is found to affect the ow in the turbulent
core, but not in the laminar sublayer. The constant b in equation (13-19) becomes ln b
p
3:4  ln (e t0 /r)/n for rough tubes. As the wall shear stress appears in the revised
expression for ln b, it is important to note that wall roughness affects the magnitude of the
shear stress in a turbulent ow.

12.13 FURTHER EMPIRICAL RELATIONS FOR TURBULENT FLOW


Two important experimental results that are helpful in studying turbulent ows are the
power-law relation for velocity proles and a turbulent-ow shear-stress relation due to
Blasius. Both of these relations are valid for ow adjacent to smooth surfaces.
12.14 The Turbulent Boundary Layer on a Flat Plate 163

For ow in smooth circular tubes, it is found that over much of the cross section the
velocity prole may be correlated by
vx  y 1/n
(12-66)
v x max R
where R is the radius of the tube and n is a slowly varying function of Reynolds number. The
exponent n is found to vary from a value of 6 at Re 4000 to 10 at Re 3,200,000. At Rey-
nolds numbers of 105 the value of n is 7. This leads to the frequently used one-seventh-power
law, v x /v x max (y/R)1/7 . The power-law profile has also been found to represent the velocity
distribution in boundary layers. For boundary layers of thickness d, the power law is written
vx y1/n
(12-67)
v x max d
The power-law prole has two obvious difculties: the velocity gradients at the wall
and those at d are incorrect. This expression indicates that the velocity gradient at the wall is
innite and that the velocity gradient at d is nonzero.
In spite of these inconsistencies, the power law is extremely useful in connection with
the von Karman-integral relation, as we shall see in Section 12.14.
Another useful relation is Blasiuss correlation for shear stress. For pipe-ow Reynolds
numbers up to 105 and at-plate Reynolds numbers up to 107, the wall shear stress in a
turbulent ow is given by
 1/4
n
t0 0:0225rv x max
2
(12-68)
v x max ymax
where ymax R in pipes and ymax d for flat surfaces.

12.14 THE TURBULENT BOUNDARY LAYER ON A FLAT PLATE


The variation in boundary-layer thickness for turbulent ow over a smooth at plate may be
obtained from the von Karman momentum integral. The manner of approximation involved
in a turbulent analysis differs from that used previously. In a laminar ow, a simple
polynomial was assumed to represent the velocity prole. In a turbulent ow, we have seen
that the velocity prole depends upon the wall shear stress and that no single function
adequately represents the velocity prole over the entire region. The procedure we shall
follow in using the von Karman integral relation in a turbulent ow is to utilize a simple
prole for the integration with the Blasius correlation for the shear stress. For a zero pressure
gradient the von Karman integral relation is
Z
t0 d d
v x (v 1  v x ) dy (12-38)
r dx 0
Employing the one-seventh-power law for v x and the Blasius relation, equation (12-68)
for t0, we see that equation (12-38) becomes
  Z

n 1/4 d d 2 y1/7 y2/7


2
0:0225v 1 v  dy (12-69)
v1d dx 0 1 d d
where the free-stream velocity, v 1 , is written in place of v x max . Performing the indicated
integration and differentiation, we obtain
 
n 1/4 7 dd
0:0225 (12-70)
v1d 72 dx
164 Chapter 12 Viscous Flow

which becomes, upon integration


 1/4
n
x 3:45 d5/4 C (12-71)
v1
If the boundary layer is assumed to be turbulent from the leading edge, x 0 (a poor
assumption), the above equation may be rearranged to give
d 0:376
(12-72)
x Re1/5
x

The local skin-friction coefcient may be computed from the Blasius relation for shear
stress, equation (12-67), to give
0:0576
Cfx (12-73)
Re1/5 x

Several things are to be noted about these expressions. First, they are limited to values of
Rex < 107 , by virtue of the Blasius relation. Second, they apply only to smooth at plates.
Last, a major assumption has been made in assuming the boundary layer to be turbulent from
the leading edge. The boundary layer is known to be laminar initially and to undergo
transition to turbulent ow at a value of Rex of about 2  105 . We shall retain the assumption
of a completely turbulent boundary layer for the simplicity it affords; it is recognized,
however, that this assumption introduces some error in the case of a boundary layer that is
not completely turbulent.
A comparison of a laminar and a turbulent boundary layer can be made from Blasiuss
laminar-ow solution and equations (12-28), (12-72), and (12-73). At the same Reynolds
number, the turbulent boundary layer is observed to be thicker, and is associated with a
larger skin friction coefcient. While it would appear then that a laminar boundary layer is
more desirable, the reverse is generally true. In most cases of engineering interest, a
turbulent boundary layer is desired because it resists separation better than a laminar
boundary layer. The velocity proles in laminar and turbulent boundary layers are compared
qualitatively in Figure 12.16.

4 dT
= 3.9
dL

Turbulent
3
y/d laminar

2 dT

Laminar dL Figure 12.16 Comparison of


0 velocity proles in laminar and
0 0.2 0.4 0.6 0.8 1.0 turbulent boundary layers. The
vx /v Reynolds number is 500,000.
12.16 Closure 165

It can be seen that the turbulent boundary layer has a greater mean velocity, hence both
greater momentum and energy than the laminar boundary layer. The greater momentum and
energy permit the turbulent boundary layer to remain unseparated for a greater distance in
the presence of an adverse pressure gradient than would be the case for a laminar boundary
layer.
Consider a at plate with transition from laminar ow to turbulent ow occurring on
the plate. If transition from laminar ow to turbulent ow is assumed to occur abruptly
(for computational purposes), a problem arises in how to join the laminar boundary layer to
the turbulent layer at the point of transition. The prevailing procedure is to equate the
momentum thicknesses, equation (12-44), at the transition point. That is, at the start of
the turbulent portion of the boundary layer, the momentum thickness, u, is equal to the
momentum thickness at the end of the laminar portion of the boundary layer.
The general approach for turbulent boundary layers with a pressure gradient involves
the use of the von Karman momentum integral as given in equation (12-46). Numerical
integration is required.

12.15 FACTORS AFFECTING THE TRANSITION


FROM LAMINAR TO TURBULENT FLOW
The velocity proles and momentum-transfer mechanisms have been examined for both
laminar and turbulent ow regimes and found to be quite different. Laminar ow has also
been seen to undergo transition to turbulent ow at certain Reynolds numbers.
So far the occurrence of transition has been expressed in terms of the Reynolds number
alone, while a variety of factors other than Re actually inuence transition. The Reynolds
number remains, however, the principal parameter for predicting transition.
Table 12.2 indicates the inuence of some of these factors on the transition Reynolds
number.

Table 12.2 Factors affecting the Reynolds number of transition from laminar to turbulent ow
Factor Inuence

Pressure gradient Favorable pressure gradient retards transition; unfavorable pressure


gradient hastens it
Free-stream turbulence Free-stream turbulence decreases transition Reynolds number
Roughness No effect in pipes; decreases transition in external ow
Suction Suction greatly increases transition Re
Wall curvatures Convex curvature increases transition Re. Concave curvature
decreases it
Wall temperature Cool walls increase transition Re. Hot walls decrease it

12.16 CLOSURE
Viscous ow has been examined in this chapter for both internal and external geometries.
Two approaches were employed for analyzing laminar boundary layer owsexact
analysis using boundary layer equations and approximate integral methods. For turbulent
boundary layer analysis along a plane surface, an integral method was employed.
Concepts of skin friction and drag coefcients were introduced and quantitative
relationships were developed for both internal and external ows.
166 Chapter 12 Viscous Flow

Approaches to modeling turbulent ows were introduced, culminating in expressions for


the universal velocity distribution. This approach considers turbulent ows to be described
in three parts: the laminar sublayer, the transition or buffer layer, and the turbulent core.
The concepts developed in this chapter will be used to develop important expressions
for momentum ow application in the next chapter. Similar applications will be developed
in the sections related to both heat and mass transfer in later sections of this text.

PROBLEMS
12.1 If Reynoldss experiment was performed with a 38-mm- 12.8 Plot a curve of drag vs. velocity for a 1.65-in.-diameter
ID pipe, what ow velocity would occur at transition? sphere in air between velocities of 50 fps and 400 fps.
12.2 Modern subsonic aircraft have been rened to such an 12.9 For what wind velocities will a 12.7-mm-diameter cable
extent that 75% of the parasite drag (portion of total aircraft drag be in the unsteady wake region of Figure 12.2?
not directly associated with producing lift) can be attributed to 12.10 Estimate the drag force on a 3-ft-long radio antenna with
friction along the external surfaces. For a typical subsonic jet, the an average diameter of 0.2 in. at a speed of 60 mph.
parasite drag coefcient based on wing area is 0.011. Determine
12.11 A 2007 Toyota Prius has a drag coefcient of 0.26 at
the friction drag on such an aircraft
road speeds, using a reference area of 2.33 m2. Determine the
a. at 500 mph at 35 000 ft; horsepower required to overcome drag at a velocity of 30 m/s.
b. at 200 mph at sea level. Compare this gure with the case of head and tail winds of 6 m/s.
The wing area is 2400 ft2. 12.12 The lift coefcient is dened as CL  (lift force)/
12.3 Consider the ow of air at 30 m/s along a at plate. At 1rv 2 A . If the lift coefcient for the auto in the previous problem
2 x r
what distance from the leading edge will transition occur? is 0.21, determine the lift force at a road speed of 100 mph.
12.4 Find a velocity prole for the laminar boundary layer of 12.13 The auto in Problems and showed a sensitivity to yaw
the form angle. At a yaw angle of 208, the lift coefcient increased to 1.0.
vx What is the lift force at 100 mph for this case?
c1 c2 y c3 y2 c4 y3
v xd 12.14 What diameter circular plate would have the same drag
when the pressure gradient is not zero. as the auto of Problem 12.11?
12.15 Estimate the normal force on a circular sign 8 ft in
12.5 Evaluate and compare with the exact solution d, Cfx, and
CfL for the laminar boundary layer over a at plate, using the diameter during a hurricane wind (120 mph).
velocity prole 12.16 A 1998 Lexus LS400 has a drag coefcient of 0.28 and a
reference area of 2.4 m2. Determine the horsepower required to
v x a sin by:
overcome drag when driving at 70 mph at sea level.
12.6 There is uid evaporating from a surface at which a. on a hot summer day T 100 F.
v x jy0 0, but v x jy0 6 0. Derive the von Karman momentum
b. on a cold winter day T 0 F
relation.
12.17 A baseball has a circumference of 9 14 inches and a
12.7 The drag coefcient for a smooth sphere is shown below.
weight of 5 14 ounces. At 95 mph determine
Determine the speed at the critical Reynolds number for a
42-mm-diameter sphere in air. a. the Reynolds number.
b. the drag force.
0.5 c. the type of ow (see the illustration for Problem ).
12.18 Golfball dimples cause the drag drop (see Figure 12.4
0.4
and the illustration for Problem 12.7) to occur at a lower
0.3 Reynolds number. The table below gives the drag coefcient
CD for a rough sphere as a function of the Reynolds number. Plot the
0.2 drag for1.65-in.-diameter sphere as a function of velocity. Show
several comparison points for a smooth sphere.
0.1

0 Re  104 7.5 10 15 20 25
103 104 105 106 107 CD 0.48 0.38 0.22 0.12 0.10
Re = Dv/n
Problems 167

12.19 The lift coefcient on a rotating sphere is very approxi- 12.27 Using a sine prole the laminar ow and a one-seventh-
mately given by power law for turbulent ow, make a dimensionless plot of the
    momentum and kinetic energy proles in the boundary layer a
RV RV
CL 0:24 0:05 over the range of 1:5 > > 0:2: Reynolds number of 105.
V V
12.28 Estimate the friction drag on a wing by considering the
Here R is the sphere radius and V is rotation rate of the sphere. following idealization. Consider the wing to be a rectangular at
For the baseball in Problem 12.17, determine the rotation rate for plate, 7 ft by 40 ft, with a smooth surface. The wing is ying at
a baseball thrown at 110 mph to have the lift equal the weight. 140 mph at 5000 ft. Determine the drag, assuming
How many rotations would such a ball make in 60 ft 6 in.?
a. a laminar boundary layer;
12.20 If the vertical velocity at the wall is not zero such as
b. a turbulent boundary layer.
would be the case with suction or blowing, what modications
occur to equation (12-33)? 12.29 Compare the boundary-layer thicknesses and local skin-
friction coefcients of a laminar boundary layer and a turbulent
12.21 If the turbulence intensity is 10%, what fraction of the
boundary layer on a smooth at plate at a Reynolds number of
total kinetic energy of the ow is due to the turbulence?
106. Assume both boundary layers to originate at the leading
12.22 In a house, water ows through a copper tube with a 0.75- edge of the at plate.
in.ID, at a ow rate of 2 gpm. Determine the Reynolds number for
12.30 Use the 1/7 power-law prole and compute the drag
a. hot water (T 120 F): force and boundary layer thickness on a plate 20 ft long and 10 ft
b. cold water (T 45 F): wide (for one side) if it is immersed in a ow of water of 20 ft/s
12.23 Plot the boundary-layer thickness along a at plate for velocity. Assume turbulent ow to exist over the entire length of
the ow of air at 30 m/s assuming the plate. What would the drag be if laminar ow could be
maintained over the entire surface?
a. laminar ow;
b. turbulent ow. 12.31 The turbulent shear stress in a two-dimensional ow is
Indicate the probable transition point. given by

12.24 For the fully developed ow of water in a smooth 0.15-m @v x


(t yx )turb reM rv x v y
pipe at a rate of 0:006 m3 /s, determine the thickness of @y
a. the laminar sublayer; Expanding v x0 and v y0 in a Taylor series in x and y near the wall and
b. the buffer layer; with the aid of the continuity equation
c. the turbulent core. @v x0 @v y0
0
12.25 Using Blasius correlation for shear stress (equation 12- @x @y
68), develop an expression for the local skin-friction coefcient in
show that, near the wall, eM
y3 higher order terms in y. How
pipes. In pipes, the average velocity is used for the friction coef-
does this compare with the mixing-length theory?
cient and the Reynolds number. Use the one-seventh-power law.
12.32 Evaluate the velocity derivative, @v x /@y, for the power-
12.26 For a thin a plate 6 in. wide and 3 ft long, estimate the
law velocity prole at y 0 and y R:
friction force in air at a velocity of 40 fps, assuming
12.33 Using the Blasius shear-stress relation (12-68) and the
a. turbulent ow;
power-law velocity prole, determine the boundary-layer thick-
b. laminar ow. ness on a at plate as a function of the Reynolds number and the
The ow is parallel to the 6-in. dimension. exponent n.

You might also like