Topics in Shear Flow Chapter 10 - The Wall Jet

Topics in Shear Flow
Chapter 10 – The Wall Jet
Donald Coles
Professor of Aeronautics, Emeritus
California Institute of Technology Pasadena, California

Assembled and Edited by

Kenneth S. Coles and Betsy Coles
Copyright 2017 by the heirs of Donald Coles

Except for figures noted as published elsewhere, this work is licensed
under a Creative Commons Attribution-NonCommercial-ShareAlike
4.0 International License
DOI: 10.7907/Z90P0X7D
The current maintainers of this work are Kenneth S. Coles
(kcoles@iup.edu) and Betsy Coles (betsycoles@gmail.com)
Version 1.0 December 2017

Chapter 10
THE WALL JET
A plane jet flowing parallel to an adjacent wall—a wall jet—is a

configuration often encountered in ejector design, in film-cooling ap-
plications, and in boundary-layer control. The radial wall jet is a
variation that is important in problems of heat and mass transfer, as
in heating by a torch or drying by an impinging jet. The situations
of interest are almost always turbulent. The latter flows are sensitive
to residual effects of transition, and the approach to experimental
similarity is awkward because a simple displacement of the origin is
not compatible with the radial geometry.
In CHAPTER 4, the model for the turbulent boundary layer
is a continuously evolving turbulent wake, modified in a definite way
by the insertion of a wall along the plane of symmetry. The no-
slip condition reduces the velocity to zero at the wall and strongly
affects the flow near the original plane of symmetry. In particular, the
presence of the wall radically changes the normal or v 0 fluctuations,
which are now reduced to zero at y = 0. The no-slip condition also
changes the other two fluctuations u0 and w0 in a more complicated
way, as discussed in various places in this monograph. As far as I am
aware, the corresponding model has never been considered seriously
for the wall jet. This model might be expected to lead to something
called the law of the jet and to further development of the concept
of equilibrium flow, and it will be addressed in SECTION 10.3.2.
501
502 CHAPTER 10. THE WALL JET
Figure 10.1: Schematic connection between the laminar

plane free jet and wall jet.
The laminar wall jet can be visualized as a laminar plane jet

with a thin plate inserted on the plane of symmetry, as shown in
FIGURE 10.1. The main mathematical consequence of the loss of
symmetry for the wall-jet flow is a qualitative change in the sim-
ilarity argument, which now leads to an eigenvalue problem. It is
therefore more important than usual to practice technique with the
problem of laminar flow. Part of this technique is an application
in SECTION 10.1.6 of the Mangler transformation, which relates a
plane flow and a radial flow in the manner shown earlier for the free
jet. Many of the other operations carried out in this chapter have al-
ready been encountered in CHAPTER 9 on the free jet, where they
are described in somewhat greater detail and supported by more ex-
tensive arguments.
10.1. LAMINAR PLANE WALL JET 503
10.1 Laminar plane wall jet into fluid at rest
10.1.1 The eigenvalue problem
Priority in solving the problem of the laminar plane wall jet with
similarity is generally assigned to GLAUERT (1956), although an
essentially complete account was published earlier by TETERVIN
(1948). The problem is more subtle than the problem of the plane
free jet, and the subtleties were fully appreciated by Glauert. The mo-
mentum equation in the boundary-layer approximation is the same
as for the free jet;
∂2u

∂uu ∂uv ∂τ
ρ + =µ 2 = . (10.1)
∂x ∂y ∂y ∂y
The boundary conditions are suitably chosen from
ψ = u = v = 0 at y = 0 , u = τ = 0 at y = ∞ . (10.2)
The momentum-integral equation is easily written down by

inspection of equation (10.1);
Z∞
d dJ
ρ u u dy = = −τw , (10.3)
dx dx
0
where τw = µ(∂u/∂y)w and J is the momentum integral previously

defined for the plane free jet by equation (9.8). The fact that J is
no longer a constant, as it was for the free jet, prevents the intro-
duction at the outset of intrinsic scales for mass, length, and time.
The friction at the wall continuously removes momentum from the
wall jet, beginning at the origin of the flow at x = 0, at a rate that
is slow but significant. It will be found for the case of laminar flow
that similarity requires the terms in equation (10.3) to behave like
x−5/4 near the origin (see equations (10.60) and (10.61) below). Thus
the singularity in τw at x = 0 is not integrable. Moreover, since the
integral in equation (10.3) behaves like x−1/4 near x = 0, the initial
momentum flux J in the similarity formulation is infinite. I empha-
size these points because some experimenters have assumed that the
momentum flux J measured at the jet exit or elsewhere in a laminar

laboratory flow has some important role to play in similarity formu-
lations of their data. Similar problems with turbulent flow are taken
up in SECTION 10.3.1.
One primitive but popular version of dimensional analysis is to
assume a power-law behavior and to determine the exponents for two
local scales U (x) ∼ xp and L(x) ∼ xq by substitution of a suitable
ansatz in the momentum equation (10.1) and its integral (10.3). This
approach is demonstrated for several different flows by BIRKHOFF
and ZARANTONELLO (1957), for example. These authors did not
anticipate the problem of the wall jet, but did comment on an eigen-
value problem for a different flow, the momentumless wake. The new
feature in the case of the momentumless wake is that the global
constant defined by the momentum integral, the drag, vanishes iden-
tically. The new feature in the case of the wall jet is that the two
equations (10.1) and (10.3) have essentially the same dimensional
structure. In either case, only one condition can be found for the two
exponents p and q unless the problem is attacked at a deeper level.
Glauert. To arrive at a dimensionless ansatz, Glauert assumed
power-law behavior. He took the free jet as a model and a point of
departure. The analysis that follows is faithful in spirit to Glauert’s
presentation, but the notation and certain details have been changed
to suit the style of this monograph. I have also chosen to begin with
plane flow rather than radial flow. Glauert postulated the existence
of a local velocity scale U and a corresponding local length scale ν/U ,
and assumed a solution of the form

ψ/ν (U y/ν)
=f = f (η) , (10.4)
(U x/ν)a (U x/ν)b
where
∂ψ ∂ψ
u= , v=− , (10.5)
∂y ∂x
as usual. Substitution in equation (10.1) yields
(a − b)f 0 f 0 − af f 00 = (U x/ν)1−a−b f 000 , (10.6)
where primes indicate differentiation with respect to η. If f is re-
quired to depend only on η and not separately on x, this equation
supplies one relation between the exponents a and b;
a+b=1 , (10.7)
together with an ordinary differential equation for f ,
f 000 + (1 − b)f f 00 − (1 − 2b)f 0 f 0 = 0 , (10.8)
whose boundary conditions, from equations (10.2), are
f (0) = f 0 (0) = f 0 (∞) = 0 . (10.9)
Glauert’s first major contribution was to establish that there

exists at least one non-trivial similarity solution of equation (10.8),
satisfying the null boundary conditions (10.9), provided that the ex-
ponent b has the eigenvalue 3/4. The analysis begins with an inte-
gration whose purpose is to examine the shearing stress f 00 and to
deal with the absence of symmetry. Replace f f 00 by (f f 0 )0 − f 0 f 0 and
integrate equation (10.8) formally from some arbitrary positive value
of η to η = ∞ to obtain
f 00 + (1 − b)f f 0 + (2 − 3b) g = 0 , (10.10)
where
Z∞
g(η) = f 0 f 0 dη . (10.11)
η
The range of integration is evidently chosen to exploit the fact that

f 0 and f 00 vanish at infinity for both the free jet and the wall jet. In
particular,
f 00 (0) = (3b − 2)g(0) , (10.12)
where
Z∞
g(0) = f 0 f 0 dη . (10.13)
0
A brief digression disposes of the symmetric problem (the free

jet). The boundary conditions in the plane of symmetry are then
f (0) = 0 and f 00 (0) = 0, corresponding to ψ(x, 0) = 0 and τ (x, 0) =
0, with f 0 (0) ∼ uc (x) left unspecified. Since g(0) is a positive con-

stant, it follows from equation (10.12) that the boundary condition
f 00 (0) = 0 can be satisfied only if b = 2/3, a = 1/3, in agreement
with the result obtained more directly in SECTION 9.1.2 above.
Now return to the unsymmetric problem, the wall jet. The
boundary conditions at the wall are f (0) = 0 and f 0 (0) = 0, cor-
responding to ψ(x, 0) = 0 and u(x, 0) = 0, with f 00 (0) ∼ τw (x)
left unspecified. Nothing can be learned from equation (10.12), and
something more is required. Glauert eliminated f 00 by multiplying
equation (10.10) by f 0 and integrating through the thickness of the
wall jet. After some integration by parts and use of the identity
g 0 = −f 0 f 0 and the boundary condition g(∞) = 0, the result is
Z∞
(3 − 4b) f 0 g dη = 0 . (10.14)
0
The integral in equation (10.14) is a positive constant, provided that

the velocity f 0 is non-negative everywhere, and the equation can
therefore be satisfied only for the exponents
3 1
b= , a= . (10.15)
4 4
This value for b requires, from equation (10.12),
1
f 00 (0) =
g(0) . (10.16)
4
It reduces the differential equation (10.8) to
4f 000 + f f 00 + 2f 0 f 0 = 0 (10.17)
and also provides the necessary invariant, which can have different
forms;
Z∞ Z∞ Z∞ Z∞
0 0 0 0
f g dη = f f f dη dη = − f g 0 dη =
0 0 η 0
Z∞
= f f 0 f 0 dη = constant . (10.18)
0
Of these, the two dominant forms in physical variables are the second
and the fourth;
Z∞ Z∞ Z∞
ρ u uu dy dy = ρ ψuu dy = F = constant . (10.19)
0 y 0
Like Glauert, I have some difficulty in assigning a physical meaning to

the quantity F . His best effort produced the phrase “flux of exterior
momentum flux.”
Having established the structure of his problem, Glauert re-
peated his derivation from the beginning in physical variables for
readers who do not object to a strong element of deus ex machina.
Note, as did Glauert, that this second derivation does not require
the assumption of similarity or of power-law behavior. First, write
an incomplete integral corresponding to equation (10.3) in the form
Z∞
∂
ρ uu dy − ρ u v + τ = 0 . (10.20)
∂x
y
Denote the integral by W , say;

Z∞
W =ρ u u dy , (10.21)
y
and observe that good things happen if the equation

∂W
− ρuv + τ = 0 (10.22)
∂x
is multiplied by the streamwise velocity u and if it is noticed that
−ρuu = ∂W/∂y from equation (10.21). Thus
∂W ∂W
u +v + τu = 0 . (10.23)
∂x ∂y
Add to this the continuity equation multiplied by W to obtain
∂uW ∂vW
+ + τu = 0 . (10.24)
∂x ∂y
Finally, integrate over the thickness of the layer and use the boundary
conditions v(0) = 0, W (∞) = 0. The result is
Z∞ Z∞
d
uW dy + τ u dy = 0 . (10.25)
dx
0 0
A last crucial step can be carried out provided that the flow is lam-
inar, with τ = µ ∂u/∂y. Then the second term in equation (10.25)
drops out;
Z∞ Z∞ 2
∂u /2
τ u dy = µ dy = 0 , (10.26)
∂y
0 0
since u is zero at both limits. For laminar flow, this procedure has
reproduced the conserved quantity (10.19);
Z∞ Z∞ Z∞
uW dy = ρ u uu dy dy = F = constant . (10.27)
0 0 y
For turbulent flow, neither equation (10.26) nor equation (10.27) is

valid.
(Interpret this process in terms of work done on fluid? Minimize
the integral of τ u? W is the momentum flux outboard of a particular
point in the flow. Equation (10.23), written as
DW
+ τu = 0 , (10.28)
Dt
suggests that the rate of change of this quantity following a stream-
line is given by the rate that work is done by the shearing stress
(this needs work). Look at the difference between F and the con-
served quantity J for the free jet. Interpret as divergence. Look at
energy. Comment on vorticity as variable, with no symmetry and
zero integral. See the Rayleigh problem in the introduction.)
Intrinsic scales. Given the existence of the integral invariant
F , it is now a simple matter to work out intrinsic scales for the
laminar wall jet. The dimensional statements
ML2 M M
[F ] = , [ρ] = , [µ] = (10.29)
T3 L3 LT
imply, in their alternative role as definitions,
ρ4 ν 9 ρν 3 ρ2 ν 5
M= , L= , T= , (10.30)
F3 F F2
with
L F
U= = 2 (10.31)
T ρν
and, as for the free jet,
UL
=1 . (10.32)
ν
The relation (10.31) provides a posteriori justification for Glauert’s
original ansatz (10.4), because U is now precisely defined. In fact,
substitution for U yields immediately
" 1/4 #
ρ 1/4 F
ψ=f y . (10.33)
F νx ρν 3 x3
Another brief calculation shows that this expression is equivalent to
ψ y
=f . (10.34)
UL3/4 x1/4 L1/4 x3/4
Tetervin. Tetervin’s earlier approach to the same problem

was handicapped by a dreadful notation and by failure to introduce
a stream function until the last possible moment. What follows is a
radical paraphrase of his argument. In effect, he assumed similarity
in terms of local scales for velocity U and layer thickness L;
ψ y
=f = f (η) , (10.35)
UL L
where U (x) and L(x) have to be determined. Substitution in the

momentum equation (10.1) gives, just as in the case of the laminar
free jet (see SECTION 9.1.2),
L dU L 00 L2 dU 0 0
f 000 + ff − f f =0 . (10.36)
ν dx ν dx
Substitution in the momentum-integral equation (10.3) gives

Z∞
L dU 2 L
f 0 f 0 dη = −f 00 (0) . (10.37)
νU dx
0
Tetervin eventually normalized the integral to unity;

Z∞
f 0 f 0 dη = g(0) = 1 , (10.38)
0
so that
L dU 2 L
= −f 00 (0) . (10.39)
νU dx
Only two of the three constant coefficients involving U and L in
equations (10.36) and (10.39) are independent, and these two are
not sufficient to determine U (x) and L(x) explicitly. Neither is the
device of the moving observer useful for resolving the question of
exponents. Tetervin, like Glauert, found another way.
When ν is eliminated between equations (10.36) and (10.39),
and the variables depending on x and on η are separated, the result
is
L dU/dx −f 000 + f 00 (0)f f 00
= 000 = −k , (10.40)
U dL/dx 2f − f 00 (0)(f f 00 − f 0 f 0 )
where k must be a positive constant because x and η are arbitrary
and dU/dx < 0, dL/dx > 0. This expression strongly suggests that
power laws are appropriate for U (x) and L(x), and guarantees in any
case that
U Lk = constant . (10.41)
Tetervin noted in passing that the boundary condition f 00 (0) = 0 in
equation (10.40) implies k = 1/2 and thus U 2 L = constant, so that
the case of the plane free jet is accounted for. The present interest
is in the case of lost symmetry with its eigenvalue k. This eigenvalue
appears along with f 00 (0) in the differential equation obtained from
the second part of equation (10.40);

000 1−k 00 00 k
f + f (0)f f + f 00 (0)f 0 f 0 = 0 . (10.42)
2k − 1 2k − 1
At this point, Tetervin’s argument becomes opaque. The essence

of his procedure, suitably revised to leave open the question of nor-
malization, is to multiply equation (10.42) by f and integrate over
the thickness of the wall jet. After the usual integration by parts and
use of the boundary conditions, the result is
Z∞
3k − 2 00
f (0) f f 0 f 0 dη = 0 . (10.43)
2k − 1
0
Both the integral and the factor f 00 (0) are necessarily positive, so
that the desired invariant emerges from this equation together with
the eigenvalue
2
k= . (10.44)
3
Equation (10.42) becomes
f 000
+ f f 00 + 2 f 0 f 0 = 0 . (10.45)
f 00 (0)
Equation (10.41) becomes
U 3 L2 = constant , (10.46)
and it follows from this result and equation (10.39) that
U ∼ x−1/2 , L ∼ x3/4 . (10.47)
Tetervin integrated equation (10.45) numerically for the par-

ticular initial conditions f (0) = f 0 (0) = 0 and f 00 (0) = 1/4. His
conversion of a two-point boundary-value problem to an initial-value
problem was successful, although he may not have been aware of the
reason, which involves a property first pointed out for the Blasius
equation by TÖPFER (1912). The argument is easily extended by
inspection to equation (10.45), which also has no pressure-gradient
term. If f (η) is a solution, so is φ(η) = αf (αη), where α is any
constant. It follows that f 00 (0) can be chosen arbitrarily, with φ(∞)
adjusted later to any desired value by a proper choice of α (see SEC-
TION X).
10.1.2 Similarity
The affine transformation. Discovery of the integral invariant F

allows the problem of the laminar plane wall jet to be treated by
the method of the affine transformation. Let a stream function ψ
be introduced in the usual way to satisfy the continuity equation.
Rewrite equation (10.1) as
∂ψ ∂ 2 ψ ∂ψ ∂ 2 ψ ∂3ψ

ρ − = µ (10.48)
∂y ∂x∂y ∂x ∂y 2 ∂y 3
and apply the affine transformation
x = ab
x ,
y = bb
y ,
ψ = cψb , (10.49)
ρ = db
ρ ,
µ = eb
µ ,
F = f Fb .
This is the same group as equations (9.28) for the plane free jet,
except that F replaces J. The result is
!
c2 d ∂ ψb ∂ 2 ψb ∂ ψb ∂ 2 ψb ce ∂ 3 ψb
ρ − = µ
b 3 . (10.50)
ab2 ∂b
y ∂bx∂b y ∂bx ∂b y2 b3 ∂by
b
Invariance of equation (10.48) thus requires

bcd
=1 , (10.51)
ae
just as in the case of the plane free jet. Transformation of equation
(10.19),
Z∞
∂ψ ∂ψ
ρ ψ dy = F , (10.52)
∂y ∂y
0
yields
Z∞
c3 d ∂ ψb ∂ ψb
ρb ψb db
y = f Fb , (10.53)
b ∂by ∂b y
0
and requires for invariance
c3 d
=1 . (10.54)
bf
As usual, I take the primary variables to be ψ and y. When equations
(10.51) and (10.54) are revised to isolate for c and b, the result is
c4 d2 b4 d2 f
=1 , =1 . (10.55)
aef a3 e 3
Hence the proper ansatz, including constants A and B for later nor-
malization, is again equation (10.33),
" 1/4 #
ρ 1/4 F
A ψ=f B y = f (η) . (10.56)
F νx ρν 3 x3
Substitution of this ansatz in the momentum equation (10.48) yields

4ABf 000 + f f 00 + 2f 0 f 0 = 0 , (10.57)
with boundary conditions
f (0) = f 0 (0) = 0, f 0 (∞) = 0 , (10.58)
corresponding to ψ = u = 0 at y = 0 and u = 0 at y = ∞. If
AB = 1, equation (10.57) is identical with my version of Glauert’s
result, equation (10.17). Substitution of equation (10.56) in equation
(10.52) for F gives
Z∞
A3
f f 0 f 0 dη = . (10.59)
B
0
The singular behavior of the flow at the origin, mentioned earlier, is
demonstrated by the relations
Z∞ 1/4 Z∞
F 3ρ

B
J =ρ uu dy = 2 f 0 f 0 dη (10.60)
A νx
0 0
and 1/4
B2 F3

τw
= f 00 (0) . (10.61)
ρ A ρ3 νx5
10.1.3 The boundary-layer solution
Glauert’s second major contribution was to obtain the eigenfunc-

tion f (η) in closed form. First, multiply equation (10.57) by f and
integrate to obtain
f 0f 0

00
4AB f f − + fff0 = 0 , (10.62)
2
where the constant of integration vanishes by virtue of the first two

boundary conditions (10.58). Multiply this result by f −3/2 and inte-
grate again, to obtain
f0 2 3/2 3/2

4AB + f − C =0 , (10.63)
f 1/2 3
where C > 0 is a constant of integration. The boundary condition
(10.58) at infinity requires
C = f (∞) . (10.64)
Finally, integrate equation (10.63) with the aid of the change of vari-
able
f = Ch2 = CH (10.65)
and the method of partial fractions. An intermediate result is
C dh 2dh hdh
dη = + + . (10.66)
4AB (1 − h) (1 + h + h2 ) (1 + h + h2 )
The final result in terms of h, after use of the boundary condition

f (0) = h(0) = 0 to evaluate the constant of integration, can be
written
√ !
C (1 − h3 ) √ −1 3h
η = ln + 2 3 tan . (10.67)
2AB (1 − h)3 2+h
Equations (10.65) and (10.67) are a parametric system for f (η), with
h as parametric variable. Note that h depends not directly on η but
on Cη/2AB.
Pause here to look at experimental data for the laminar
profile; see
BAJURA and SZEWCZYK (1970)

BAJURA and CATALANO (1975)
TSUJI et al. (1977)
TSUJI and MORIKAWA (1977)
HORNE and KARAMCHETI (1979)
SCIBILIA and DUROX (1980)
PAIGE (1988)
ZHOU et al. (1992)
It remains to consider the streamlines of the boundary-layer
flow in compact outer variables (x, y) having equal scales. A unique
representation of the flow can again be found, without regard for the
values of the three constants A, B, and C. Rewrite equation (10.56)
in terms of H as
" 1/4 #
A ρ 1/4 f (η) C C F
ψ= =H η =H y .
C F νx C 2AB 2A ρν 3 x3
(10.68)
In the combinations containing ψ and y, use the second of equations
(10.30) to eliminate the quantity F in favor of L = ρν 3 /F . Thus
write " 4 1/4 #
A L 1/4 ψ

C y
=H . (10.69)
C x ν 2A Lx3
Compact outer variables define themselves immediately as
4 4
ψ C x C y
Ψ= , X= , Y = , (10.70)
2ν 2A L 2A L
and equation (10.68) takes the form

1/4 Y
Ψ=X H . (10.71)
X 3/4
Note that H = h2 = f (η)/C, but that the argument of H is the

quantity Cη/2AB on the left in equation (10.67). The example of
the free jet suggests that a useful relation involving the constants A
and C should emerge when the integral invariant (10.59) is evalu-
ated for Glauert’s closed-form solution. Use equation (10.65) and its
derivative, together with equation (10.63), to replace the variable f
by h. The result is
Z∞ Z1
0 0 C4 C4 A3
f f f dη = h4 (1 − h3 )dh = = , (10.72)
3AB 40AB B
0 0
from which
C 4 5

= . (10.73)
2A 2
The variables in equation (10.71) can therefore be written
ψ 5x 5y
Ψ= , X= , Y = . (10.74)
2ν 2L 2L
Streamlines for the boundary-layer approximation (10.71) are
shown in FIGURE 10.21 for the case of a laminar wall jet flowing
from the origin along a plane wall that extends to infinity in the
positive x-direction. Rather than calculate Ψ on a large rectangular
array (X, Y ) and find level curves on which Ψ is constant, it is
simpler here to define each streamline separately. The algorithm is:
fix Ψ, vary X. Calculate H = Ψ/X 1/4 = h2 . Calculate h. Calculate
Cη/2 AB = Y /X 3/4 from equation (10.67). Calculate Y .
A local Reynolds number can be expressed in compact outer
variables by beginning with dimensionless versions of equations (10.47);
U (x) = UL1/2 x−1/2 , L(x) = L1/4 x3/4 . (10.75)
Use of equations (10.30) and (10.31) leads to
1/4 3 3 1/4
F2

ρν x
U= 2 2 2
, L= , (10.76)
ρ ν x F
1
A longer handwritten version of the caption for this figure in the 1996 ms.
reads ”Streamlines Ψ = ψ/2ν = constant of the boundary-layer approximation
for the laminar plane wall jet according to equation (10.71). The range of Ψ is
0(1)10 (check).
Figure 10.2: Streamlines of the boundary-layer model

for the laminar plane wall jet according to equation
(10.71). The range of........
and thus to
1/4 1/4
UL Fx x 1/4 2
Re(x) = = = = X . (10.77)
ν ρν 3 L 5
10.1.4 Normalization
The three constants A, B, and C for the plane wall jet can be as-
signed sensible values by operations that run in parallel with similar
operations for the plane free jet in SECTION 9.1.4. The condition
4AB = 1 (10.78)
establishes the standard operator f 000 + f f 00 in equation (10.57). A

second and mandatory condition, just derived, is
C
= (40)1/4 . (10.79)
A
(See end of Part A of this chapter. This is too messy. No tidy

normalization seems to be in view. Sort through this material to find
something simple, elegant, and redundant. Consider equation (10.67)

for h near 1 and for h large and negative. Note that for h = 1 the
angle is 30 degrees. Should square root be ±? Need to match to plane
jet at infinity. Try osculating parabola.)
The third condition determining the constants A, B, C requires
definition of a length or velocity scale. The simplest choice, suggested
by the example of the free jet (see SECTION 9.1.4), is to set the
argument of h; i.e., the left-hand side of equation (10.67), equal to η
itself, so that C/2AB = 1 or C = 1/2.
The maximum velocity U = um (m for maximum) is easily
worked out;
Z∞ 1/4
UL 1 ψ(x, ∞) 1 Fx
Re = = udy = = f (∞) = 2X 1/4 .
ν ν ν A ρν 3
0
(10.80)
This Reynolds number is small compared with the corresponding
value Re = 12 X 1/3 for the plane free jet. In FIGURE 10.2, which
extends (ten) times farther than FIGURE 9.4, the Reynolds number
at the right boundary is (six) times smaller. Comparison of equa-
tion (10.80) with equation (9.57) for the free jet suggests that the
two measures just cited are associated with the exponent and the
coefficient, respectively.
The maximum streamwise velocity ηm occurs when ∂u/∂y ∼
f 00 = 0. With this condition, equations (10.62) and (10.63) can be
restated in terms of h and h0 and solved algebraically to produce
(see Tetervin)
2/3
1
f (ηm ) = C ; (10.81)
4
1/3 2
0 1 C
f (ηm ) = . (10.82)
4 8AB
Thus if f 0 (ηm ) = 1, then C 2 /4AB or C = (32)1/6 . Conversely, if
0 1/3

C = 1/2, then f (ηm ) = (2) 16.
Several other choices suggest themselves, chief among them the
integral scale L for the profile. Define this integral scale in terms of
the maximum velocity um (m for maximum) by

Z∞
um L = udy = ψ(x, ∞) . (10.83)
0
After use of the ansatz (10.56) and the second of conditions (10.82),
this turns into
1/4 Z∞
f 0 (η)

F 8AB
ηe = B L= dη = (4)1/3 . (10.84)
ρν 3 x3 0
f (ηm ) C
0
Hence if ηe = 1, C/4AB = 2(4)1/3 .

A similar calculation, with h replacing f , leads from the defi-
nition (10.13) to
Z∞
1 C3
g(0) = f 0 f 0 dη = . (10.85)
18 AB
0
Finally, integration of the primary differential equation (10.57) be-

tween the limits zero and infinity leads to
g(0) 1 C3
f 00 (0) = = . (10.86)
4AB 72 A2 B 2
Either of these relationships, as well as
Z∞
f (∞) = f 0 dη = C , (10.87)
0
could provide a third condition if its right-hand side is arbitrarily

set equal to unity, say. The results, respectively, are C 3 = 9/2 if
g(0) = 1, C 3 = 9/2 if f 00 (0) = 1, and C = 1 if f (∞) = 1, where 4AB
is read as unity.
The inflection point in the profile at η = ηi , say, is found by
putting f 000 = 0. This point also marks the maximum velocity along
a streamline, since Du/Dt = ν∂ 2 u/∂y 2 = 0 for laminar flow. Then
equation (10.57) becomes f f 00 + 2f 0 f 0 = 0, and other derivatives

at the inflection point can be calculated from this truncated form
together with equations (10.62) and (10.63). The results are
2/3
5
f (ηi ) = C ; (10.88)
8
1 5 1/3 C 2

0
f (ηi ) = ; (10.89)
16 8 AB
1 C3
f 00 (ηi ) = − . (10.90)
128 A2 B 2
Hence if f 0 (ηi ) = 1, then C 2 /4AB = 8/51/3 .

The vorticity thickness ηζ is defined graphically in FIGURE X
and is defined algebraically by
1/3
f 0 (ηm ) 1 AB
ηζ = − 00 = 16 , (10.91)
f (ηi ) 4 C
where f 0 (ηm ) is given by equation (10.82). Hence if ηζ = 1, C/4AB =

(4)2/3 .
The normalizations used by Glauert and Tetervin can be in-
ferred by using my notation in the ansatz (10.56) and the relations
that follow. Glauert put C = 1 and also 4AB = 1, according to his
equation (4.1). It follows that A = (1/40)1/4 , B = (5/32)1/4 ; and
these are the numbers that appear in Glauert’s equations (4.9) for
the plane case. Tetervin’s final normalization can be deduced from
his equation (20), in which his G(ξ) is the same as my f (η). He
put 4AB = 4/3 and f 00 (0) = 1/4, and these together with equation
(10.86) above imply C = f (∞) = (2)1/3 = 1.259921. His numerical
result for large η, namely f (14.95) = 1.259916, is evidence that his
integration was carried out with remarkable accuracy.
(Give normalized f , f 0 , etc.)
77T777777'77T,'777?'7777777T,TT ~
.
~~~~ge~~e , //! (
··-'·· --~--
. .-._Physic~ne
FIGURE 19.7. Mapping of the outer entrained flow for the laminar plane
Figure 10.3: Mapping of the outer entrained flow for
wall jet.
the laminar plane wall jet.
10.1.5 Entrainment and composite flow
The outer or entrained flow associated with the boundary-layer solu-

tion in FIGURE 10.2 can again be obtained by the method of confor-
mal mapping, as indicated in FIGURE 10.3. The complex potential
for uniform flow in the ζ-plane is
F (ζ) = φ + iψ = U0 ζ . (10.92)
The mapping can be assumed to be of the form

3/4
ζ = L0 z 1/4 eiα , (10.93)
where ζ = Reiω and z = reiθ , with U0 , L0 , and α to be determined.

Angles are related by
θ
ω = +α . (10.94)
4
Consequently, if θ = π when ω = π on OA,
3π
α= . (10.95)
4
Since now θ = 0 when ω = 3π/4 on OB, streamlines of the outer
flow will intersect the x-axis at an angle of 45 degrees. The complex
potential in the z-plane becomes

θ+3π

F (z) = φo + iψo = U0 ζ(z) = U0 L0 r1/4 ei
3/4
4 , (10.96)
and the outer stream function is

3/4 θ + 3π
ψo (r, θ) = U0 L0 r1/4 sin . (10.97)
4
At the wall, where θ = 0 and r = x,

1 3/4
ψo (x, 0) = √ U0 L0 x1/4 . (10.98)
2
This outer flow on the positive x-axis in the physical plane is to
be matched to the inner stream function at infinity, from equation
(10.56) with f (∞) = C;
C F νx 1/4

ψi (x, ∞) = . (10.99)
A ρ
Matching therefore requires

1/4
1 C Fν C
√ U0 L3/4
0 = = UL3/4 , (10.100)
2 A ρ A
where the last equality makes use of equations (10.30). Finally, there-
fore,
1/4
√ C

F νr θ + 3π
ψo (r, θ) = 2 sin . (10.101)
A ρ 4
(Work out pressure here.)

The composite stream function ψc is the sum of the inner com-
ponent (10.56) and the outer component (10.101) with the common
part (10.99) subtracted,
C F νr 1/4 x 1/4 f (η) √

θ + 3π
ψc = − 1 + 2 sin .
A ρ r C 4
(10.102)
Figure 10.4: Streamlines Ψ = constant of the composite

model for the laminar plane wall jet into a stagnant
fluid according to equation (10.103).
In terms of the reduced similarity variables X and Y defined by

1/2
equations (10.70), with R = (X 2 + Y 2 ) and tan Θ = Y /X, this is
√

1/4 Y Θ + 3π
Ψc = R I(Θ) H − 1 + 2 sin ,
X 3/4 4
(10.103)
where
I(Θ) = (cos Θ)1/4 for x >0 ,

(10.104)
=0 for x <0 .
Streamlines for the composite flow are shown in FIGURE 10.4. The
calculation here requires an iteration for h(η) and a contour sub-
routine. The figure can be viewed as a conceptual model for flow
near the nozzle of a plane wall ejector with small induced flow (see
SECTION X).
(Want S, U , V , T ; see free jet. Plot corrected profile, etc.)
10.1.6 The laminar radial wall jet
10.1.7 Stability and transition
10.2 Laminar plane wall jet into moving fluid
10.2.1 Similarity
10.3 Turbulent plane wall jet into fluid at rest
10.3.1 Similarity
A preliminary step is to determine if the turbulent plane wall jet is

also an eigenvalue problem. Physically, it can never be established
whether or not there is a finite initial momentum flux J, because tur-
bulent flow cannot be observed at sufficiently low Reynolds numbers.
The argument below refers only to the boundary-layer approxima-
tion, with the momentum equation written as
∂u ∂u 1 ∂τ
u +v = . (10.105)
∂x ∂y ρ ∂y
The boundary conditions are u = 0 and ψ = 0 at y = 0 and u = 0

and τ = 0 at y = ∞. (Use the best and highest version of the
invariant.)
With the laminar problem of SECTION 10.1.1 in mind as both
a model and a special case, multiply equation (10.105) by the product
ψu, where ψ is the usual stream function, and add to the result the
two identities
∂u ∂v
ψu +ψu =0 (10.106)
∂x ∂y
and
∂ψ ∂ψ
uu + uv = −uuv + uvu = 0 (10.107)
∂x ∂y
to obtain
∂ψuu ∂ψuv ψ ∂τ
+ = . (10.108)
∂x ∂y ρ ∂y
10.3. TURBULENT PLANE WALL JET 525
Integration, with ψ = 0 at the wall and τ = 0 at infinity, gives

Z∞ Z∞ Z∞
∂ψuu 1 ∂τ 1 ∂ψ
dy = ψ dy = − τ dy . (10.109)
∂x ρ ∂y ρ ∂y
0 0 0
Consequently,
Z∞ Z∞
d 1
ψuu dy = − τ udy . (10.110)
dx ρ
0 0
If the flow is laminar, τ = µ ∂u/∂y, and
Z∞ Z∞
∂u2 /2
τ u dy = µ dy = 0 (10.111)
∂y
0 0
because u2 vanishes at both limits. Thus

Z∞
ψuu dy = F = constant . (10.112)
0
If the flow is turbulent, the integral (10.111) does not necessarily

vanish. The velocity is presumably always positive, and the stress τ
changes sign, being positive near the wall and Rnegative farther out,
so there is the possibility. Calculate τ from Du/Dt dy and see
what happens.
Forthmann did not look at wall law because he did not know
the wall friction. He also introduced the half-velocity scheme for
defining δ.
Schwartz and Cosart separated the profile into two parts at
the point of maximum velocity. The wall jet should be treated by
the same methods that are successful for the boundary layer.
The growth rate for a wall jet is about 3/4 of the growth rate
for a free jet. This is because the vertical fluctuations are inhibited
and the effect is felt in all of the Reynolds stresses and in entrainment.
Comment on cases where τ and du/dy do not go to zero at the same
point. This is a blow to the idea of eddy viscosity.
Need a handout on methods for measuring surface friction.
10.3.2 The law of the jet
Measured mean-velocity profiles in plane wall jets can be found in
FORTHMANN (1934)
BAKKE (1957)
SIGALLA (1958)
SCHWARZ and COSART (1961)
PATEL (1962)
RAJARATNAM (1965)
SRIDHAR and TU (1966)
TAILLAND and MATHIEU (1967)
GUITTON (1968)
KOHAN (1968)
HUBBARTT and NEALE (1972)
SPETTEL et al. (1972)
KIND and SUTHANTHIRAN (1973)
HO and HSIAO (1983)
SCHNEIDER (1987)
ABRAHAMSSON et al. (1991)
KATZ et al. (1992)
WYGNANSKI et al. (1992)
Measured Reynolds-stress profiles can be found in
FORTHMANN (1934)
TAILLAND and MATHIEU (1967)
GUITTON (1968)
SPETTEL et al. (1972)
SCHNEIDER (1987)
ABRAHAMSSON et al. (1991)
WYGNANSKI et al. (1992)
10.4. THE TURBULENT RADIAL WALL JET 527
10.3.3 Entrainment and composite flow
10.3.4 Coanda effects
10.4 The turbulent radial wall jet
10.4.1 Similarity
The mean-velocity profile in turbulent radial wall jets is reported in
POREH (1959)
POREH and CERMAK (1959)
SCHRODER (1961)
TSUEI (1962)
LUDWIEG (1964)
CHAO and SANDBORN (1966)
DONALDSON (1966)
JOHNSON (1967) 19D
POREH et al. (1967)
HRYCAK (1970)
SCHOLTZ and TRASS (1970)
DONALDSON et al. (1971)
GOVINDAN and RAJU (1974)
ERA and SAIMA (1976)
BOLDMAN and BRINICH (1977)
LEISTER (1977)
MITACHI and ISHIGURO (1977)
TANAKA and TANAKA (1977, 1978)
TANI and KOMATSU (1977)
ARAUJO et al. (1981)
DESHPANDE and VAISHNAV (1982)
KATAOKA et al. (1983)
CODAZZI et al. (1983)
Tanaka. Figure 13 is useful.

Tanaka. Combined jet far downstream behaves like a single jet. In-
cludes single jet for reference.

Bradshaw. Clumping.
Knystautas. Study this again.
Hegge Zijnen (two papers). Note no side plates. Figure 6 implies
turbulent Prandtl number.
Foss and Jones. Effect of low aspect ratio with side walls. Mecha-
nism? Mean velocity profile is insensitive.
Curtet. Not ejector; secondary stream is controlled. Figure V.4 has
fitting constant. Figure V.7 shows positive and negative flow ratios.
Figure VII.1 shows separation bubble.
It is probably time to invent the law of the jet. The sketch2
shows the decomposition of the profile. The corresponding formula
is
u yuτ uc y
=f − j (10.113)
uτ ν uτ δ
where a tentative form for j(y/δ) is sin2 (πy/2δ). The defect form is
obtained by subtracting the local friction law

δuτ uc
0=f − (10.114)
ν uτ
to obtain
u 1 y uc π y
= + cos2 . (10.115)
uτ κδ uτ 2δ
In wall-law variables, equation (1)3 has the form shown in the sketch.
If the defect law is equivalent to equilibrium, then uc /uτ must be
constant, and so must δuτ /ν. Each equilibrium flow has an invariant
profile, which changes with (??) The data do not seem to have this
property, perhaps because none of the flows are fully developed.
The boundary-layer problem has been wrapped up for 30 years.
The wall jet has a number of properties in common, but is still being
treated by 19th-century methods.
2
No such sketch has been found.
3
Unclear reference.
Tanaka and Tanaka also studied the free jet and wall jet. They
divide the profile at the maximum velocity. Note that r is measured
from the outside of the pipe, not from the axis of symmetry. The
estimates of Cf from the momentum equation and the fit to the wall
law do not agree.
The stability paper seems to show a vena contracta, but this
may be the cylindrical geometry. The paper allows the observed vortex
pairs to be treated along with the eigenfunctions. This is a good paper
on stability.
Describe profile formula for wall jet. Invent the law of the jet
to supply rigor and detail. (Need sketch). Since the jet function is
not known, the process is iterative. (Need sketch in wall-law coordi-
nates.) If uc /uτ is constant, so is δuτ /ν, and there is only one profile
for a given flow. Different profiles may apply for different Reynolds
numbers.
The colliding round jets (Witze and Dwyer) form a radial jet
that grows at an abnormally large rate.
Several authors have compared the flow near the wall in a wall
jet to the log law, with considerable variations on both sides of the
Prandtl law.
Irwin’s data have low scatter.
These wall-jet data should reinforce a conviction that the law
of the wall, which was originally recovered from pipe data, is really
universal. The law of the wall is the largest handle striking out of the
problem of shear flow near a wall. It should be central in any global
study.
It remains to consider the usual momentum-integral equation.
The development is at first completely general. Rewrite equation
(xxx), with the aid of the continuity equation ∂u/∂x + ∂v/∂y = 0,
as
∂uu ∂uv du∞ ∂τ
ρ + = ρu∞ + (10.116)
∂x ∂y dx ∂y
and integrate from y = 0 to some fixed value of y, with the boundary
condition v = 0 at y = 0 (either because of symmetry or because of
the presence of a wall). The result can be written

Zy Zy
∂uu du∞
ρ dy + ρuv = ρ u∞ dy + τ − τw (10.117)
∂x dx
0 0
where τw is the value of τ at y = 0; this value may be zero.

The continuity equation also yields an integral,
Zy
∂u
dy + v = 0 (10.118)
∂x
0
so that equation (10.117) can be written

Zy Zy Zy
∂uu ∂u du∞
ρ dy − ρu dy − ρ u∞ dy − τ + τw = 0 . (10.119)
∂x ∂x dx
0 0 0
To eliminate the divergent integrals when the upper limit goes to

infinity, add to this the identity
Zy Zy Zy
∂uu∞ ∂u du∞
−ρ dy + ρu∞ dy + ρ udy = 0 (10.120)
∂x ∂x dx
0 0 0
to obtain
Zy Zy
∂ ∂
ρ (uu − uu∞ )dy + ρ(u∞ − u) udy +
∂x ∂x
0 0
Zy
du∞
+ρ (u − u∞ )dy − τ + τw = 0 . (10.121)
dx
0
Finally, let the upper limit go to infinity and change the signs;
Z∞ Z∞
d du∞
ρ u(u∞ − u)dy + ρ (u∞ − u)dy − τw = 0 . (10.122)
dx dx
0 0
Equation (10.122) is the momentum-integral equation of Karman. It

is customary to define a displacement thickness δ ∗ or δ1 by
Z∞
∗
u∞ δ = (u∞ − u)dy (10.123)
0
and a momentum thickness θ or δ2 by

Z∞
u2∞ θ = u(u∞ − u)dy . (10.124)
0
(Use a sketch to define these graphically, particularly con-

cept of displacement.) Equation (10.122) can then be succinctly
written
ρd 2 du∞
τw = u θ + ρu∞ δ ∗ . (10.125)
dx ∞ dx
If u∞ = constant, this becomes
dθ
τw = ρu2∞ . (10.126)
dx
This expression is often used to determine the surface friction τw (as
is 10.125) in the more general case). If a turbulent boundary layer is
visualized as a wake, with a wall continuously removing momentum,
this equation exposes the process of local momentum removal.
The invariant J for the radial jet should probably include a
factor 2π to represent integration over the azimuthal angle.
Practice source-sink method for plane laminar jet, where outer
flow is known (obtained by conformal mapping). (Will separation of
variables work? Variables are separated in known answer.)
Handout on boundary layer. Ideas are rare, one every few years or
occasionally much longer. Millikan looked at the departure h from the
log law, but had no good data except for pipe and possible channel in
1938. Millikan’s paper was absolutely ignored. Are there any citations
before my thesis? Ludwieg and Tillmann were interested in the fail-
ure of the momentum-integral equation to give the correct friction in
flows going to separation. Ludwieg invented the heated element as an
alternative. They stumbled on the result that the log law is indepen-
dent of pressure gradient as well as Reynolds number. (Put ∂w/∂z
in derivation.) Clauser also has a plot of h but did not comment on
the shape or meaning.
10.5 Turbulent plane wall jet into moving fluid
10.5.1 Similarity
10.5.2 Relaxation
10.5.3 Effectiveness
10.5.4 Boundary-layer control
10.6 Three-dimensional wall jets
10.6.1 Single jets
10.6.2 Film cooling through holes

Topics in Shear Flow Chapter 10 - The Wall Jet

Uploaded by

Document Informationclick to expand document informationRead tho

Document Informationclick to expand document information

Copyright:

Available Formats

Topics in Shear Flow Chapter 10 - The Wall Jet

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Topics in Shear Flow Chapter 10 - The Wall Jet

Uploaded by

Copyright:

Available Formats

Topics in Shear Flow

Chapter 10 – The Wall Jet

Assembled and Edited by

Copyright 2017 by the heirs of Donald Coles

THE WALL JET

A plane jet flowing parallel to an adjacent wall—a wall jet—is a

Figure 10.1: Schematic connection between the laminar

The laminar wall jet can be visualized as a laminar plane jet

10.1 Laminar plane wall jet into fluid at rest

10.1.1 The eigenvalue problem

The momentum-integral equation is easily written down by

where τw = µ(∂u/∂y)w and J is the momentum integral previously

momentum flux J measured at the jet exit or elsewhere in a laminar

supplies one relation between the exponents a and b;

together with an ordinary differential equation for f ,

f 000 + (1 − b)f f 00 − (1 − 2b)f 0 f 0 = 0 , (10.8)

whose boundary conditions, from equations (10.2), are

f (0) = f 0 (0) = f 0 (∞) = 0 . (10.9)

Glauert’s first major contribution was to establish that there

f 00 + (1 − b)f f 0 + (2 − 3b) g = 0 , (10.10)

The range of integration is evidently chosen to exploit the fact that

A brief digression disposes of the symmetric problem (the free

0, with f 0 (0) ∼ uc (x) left unspecified. Since g(0) is a positive con-

The integral in equation (10.14) is a positive constant, provided that

Like Glauert, I have some difficulty in assigning a physical meaning to

Denote the integral by W , say;

and observe that good things happen if the equation

For turbulent flow, neither equation (10.26) nor equation (10.27) is

imply, in their alternative role as definitions,

Another brief calculation shows that this expression is equivalent to

Tetervin. Tetervin’s earlier approach to the same problem

where U (x) and L(x) have to be determined. Substitution in the

Substitution in the momentum-integral equation (10.3) gives

Tetervin eventually normalized the integral to unity;

At this point, Tetervin’s argument becomes opaque. The essence

Equation (10.41) becomes

and it follows from this result and equation (10.39) that

U ∼ x−1/2 , L ∼ x3/4 . (10.47)

Tetervin integrated equation (10.45) numerically for the par-

The affine transformation. Discovery of the integral invariant F

Invariance of equation (10.48) thus requires

Substitution of this ansatz in the momentum equation (10.48) yields

10.1.3 The boundary-layer solution

Glauert’s second major contribution was to obtain the eigenfunc-

where the constant of integration vanishes by virtue of the first two

The final result in terms of h, after use of the boundary condition

BAJURA and SZEWCZYK (1970)

and equation (10.68) takes the form

Note that H = h2 = f (η)/C, but that the argument of H is the

Figure 10.2: Streamlines of the boundary-layer model

establishes the standard operator f 000 + f f 00 in equation (10.57). A

(See end of Part A of this chapter. This is too messy. No tidy

something simple, elegant, and redundant. Consider equation (10.67)

the maximum velocity um (m for maximum) by

Hence if ηe = 1, C/4AB = 2(4)1/3 .

Finally, integration of the primary differential equation (10.57) be-

could provide a third condition if its right-hand side is arbitrarily

C F νr 1/4 x 1/4 f (η) √