hr. 3. Heal Mass Transfer.

Vol. 14, pp. 144-1464.

Pergamon Press 1971.

Printed in Great Britain

AN EXPERIMENTAL
COAXIAL

INVESTIGATION

TURBULENT

OF

JETS

F. H. CHAMPAGNE and I. J. WYGNANSKI

Boeing Scientific Research Laboratories, Seattle, Washington, U.S.A.
(Received 30 Muy 1970)
Abstract--The flow field generated by two coaxial jets was investigated experimentally with hot-wire
anemometers. The area ratio between the external and internal nozzle was varied as well as the velocity
issuing from each of the nozzles. The distribution of the mean velocities, turbulence intensities, and shear
stresses were determined for the various cases. The development of the flow field and its approach to a
self-preserving state is discussed. The Reynolds numbers based on the nozzle diameters varied from 0 to
10 and the velocities were low enough that the flow can be considered incompressible.

total ;
center-line.

NOMENCLATURE

A,
Q
J,

u, v,
u

Inax,

7;;
1,
u, v, w

x, y,
Y

Ill,

Ym/2,

nozzle area ;
nozzle diameter ;
jet momentum flux ;
axial and radial mean velocity components ;
maximum velocity in any transverse
plane ;
reference velocity defined in text ;
function defined in text ;
length scale ;
axial, radial, and azimuthal components of intensity velocity fluctuation ;
axial and radial coordinates ;
distance from the axis to the point
where U = U,,, ;
the larger distance from the axis to
the point where U = +U,,,.

Greek symbols
fluid density ;
P?
similarity variable.
%
Subscripts
hypothetical origin ;
h,
refers to inner nozzle ;
i,
refers
to outer nozzle ;
0,

INTRODUCIION

THE ESSENTIAL features of a turbulent jet issuing

into a still ambient fluid, or a secondary stream,
have been established from the results of many
investigationsand are discussed quite thoroughly
in the book by Hinze [l]. Although the general
features of these relatively simple jet flows are
well known, very little is known about the
complicated flow field that arises from the
interaction of two or more turbulent jets. The
present study is an experimental investigation of
coaxial turbulent jets issuing from two round
coaxial nozzles as shown in Fig. 1. Such a flow
field provides one of the simplest cases of a wide
range of engineering and geophysical problems
that involve the interaction or mixing between
turbulent shear flows.
Some experimental work has been done on
coaxial jets by investigators interested in flame
reactors. Combustion reactions are commonly
diffusion-controlled, that is the kinetics of the
reaction are so rapid that the rate of reaction
is completely determined by the turbulent
mixing of the reactants. Stark [2] experimentally
investigated the flow field in a flame formed by a

1445

1446

F.

H.

CHAMPAGNE

and

1. J.

WYGNANSKI

Coxlal nozzles y
Outer
potential
core
inner

Typical axiol mean

velocity profile in
developing region

Devel~ment

FIG. I. Sketch

regton -

Velocity proWe in
s$miicrity region

--S~miiortty region --

of flow field. Velocity components

along coordinate
and t;i, u and R u respectively.

coaxial jet for a range of o.d., D,.to id., Di7

from 1.9 to 3.8, while varying the initial velocity
of the outer jet, U,, to the initial velocity of the
inner jet% ti,, by a factor of 3 to 1. Total head
measurements were made and the mean velocity
profiles werepresented for various axial distances
up to eighteen jet diameters downstream. The
results, however, are complicated by the presence
of temperature and density variation caused by
the flame and rather thick boundary layers in
the nozzles. Stark pointed out that the initial
velocity and concentration
distributions at
the nozzles had a substantial effect on the mean
flow field and the flame.
Arutyunov [3] carried out an experiment
similar to Starks, only he used air in both
nozzles. The air to the nozzles was heated
differentially and the resulting temperature and
velocity fields were measured using a total head
tube and thermocouples. The velocity ratio
U, !Lr, was varied from 044 to 2.0 while the
diameter ratio D,/Di was varied from 1.5 to 3.0.
Arutyunov converted his temperature profiles
into concentration profiles assuming that the
turbulent transfer of heat and mass are identical.
The results were compticated by the presence
of temperature variations in the jets and the
unknown initial turbulence intensity at the
nozzle exit plane. Some qualitative conclusions

axes x and y

concerning the design of coaxial jet burners

were presented.
Chigier and Beer [4] also used a total head
probe to determine mean velocity profiles in a
coaxial jet of a specialized design, in which the
outer nozzle was separated from the inner
nozzle by a solid annular ring. No contractio1~
was used in generating the inner jet, and presumably the initial condition of the inner jet
was that of a fully developed turbulent pipe flow.
The initial annular separation of the jets led to
a separation bubble which made the flow field
quite complex and further, makes the results
valid only for their particular nozzle design.
A characteristic common to all of the above
investigations was that they were concerned only
with the velocity field as measured with Pitot
tubes rather than the structure of the turbul~t
flow field.
A theoretical model of coaxial turbulent jets
was proposed by Morton [5]. The model was
based on assuming the double jet consisted of
a core (inner) jet and an annular (outer) jet.
each having uniform but different velocities,
with the mixing between the jets and between
the annular jet and the surrounding ambient
fluid represented by an entrainment constant.
Equations describing the velocities of the jets
and the growth of the jet boundaries were

COAXIAL

TURBULENT

JETS

1447

FIG. 2. Schematic sketch of coaxial jet facility. The inner

diffuser and settling chamber are not shown in dotted lines
although they are enclosed by outer counterparts.

derived and numerically integrated for some

cases. Also, equations governing the mixing of
two fluids emitted from the inner and outer
jets, respectively, were treated.
The present study was undertaken to provide
mean velocity, turbulent intensity and shear
stress data across the entire mixing region for
well defined initial conditions and to investigate
the tendency of the mean velocity, the turbulent
intensity, and shear stress profiles to approach
self-preserving forms.
EXPERIMENTAL

~RANGEMENT

A centrifugal blower supplied the airflow for

both nozzles as shown schematically in Fig. 2.
The contraction ratios were 144 to 1 for the
inner nozzle and about 100 to 1 for the outer
nozzle. The diameter of the inner nozzle was
1 in. for all experiments. The flow emerged
normal to the plane of the nozzles, which were
set nearly flush in the center of a plane vertical
wall that extended about 4 ft in any radial
direction The resulting jet issued freely into the
room except for being confined by the floor
on the bottom and by 18 mesh fiberglass screens
on the other three sides, which were all about
4 fi from the geometric centerline of the jets.
The fiberglass screens helped damp out extraneous roof drafts. All sensing probes were
mounted on a carriage assembly, which rode on
rails aligned parallel with the jet axis and thus
could be moved and locked at any position. The

main part of the carriage assembly was near

the floor and well below the jet.
The temperature of the jets was maintained to
within -+ 1F relative to the room temperature,
and the latter did not vary more than 2F over
an entire day. The temperature of the jet and
the room air were measured with thermistor
probes with an accuracy of _tOlF. The pressures in the plenum chambers were read on an
inclined alcohol monometer and a standard
water monometer. All measurements were made
at subsonic speeds with the larger initial
velocities being approximately 60 m/s.
The area ratio between the external and
internal nozzles, A,/Ai, was changed once by
reducing the diameter of the external nozzle.
All observations were made at the two area
ratios &/A, = 294 and 1.28 with the corresponding Reynolds numbers ranging from 0 to
about 10 for both nozzles. The initial velocity
ratio between the external and internal nozzle
was varied 0 < U,jUj < 10. The internal jet,
however, was never completely stopped (i.e.
Ui # 0) to prevent the creation of an annular
jet with its low-pressure recirculating bubble
[4]. Indeed, for U,,iUi = 10, various flow
visualization techniques were used to ensure that
no recirculation existed near the interior jet exit.
Mean velocity and turbulence measurements
were made using two linearized, constanttemperature hot-wire anemometers. The constant-temperature
anemometers
were Disa

144x

F. H. CHAMPAGNE

55AOl units with Disa 55DlO lihearizers. Hotwires were made from 090020 in. tungsten wire
with length-to-diameter
ratios of about 200.
They were operated at an overheat ratio of 0.8
to minimize sensitivity to temperature. The
wires were calibrated in the potential core of the
inner nozzle where the turbulence intensity was
of the order of 0.1 per cent. Drift of wire calibrations was virtually eliminated as the air was
cleaned using an electrostatic precipitator which
removed particles and hydrocarbons in the air
down to 0.04 p dia. and by controlling the jet
and room temperature.
The processing circuits consisted of a highpass filter, a bank of vacuum thermocouples for
squaring operations, an analogue integrator
for time averaging, and a Disa 551406correlator
for sum and difference equations. The high-pass
filter and vacuum thermocouples were tested
for a frequency range from 0.01 cps to 20 kc and
the response was entirely flat from 0.05 cps to
20 kc. The vacuum thermocouples were a.c.
and d.c. calibrated and were operated only
in the range where their output voltage was
related linearly to the square of the input
current. The analogue integrating circuit consisted of a modified d.c. amplifier with a polystyrene capacitor in the feedback loop and a
Cramer automatic reset interval timer and relay.
The integrating circuit was tested by integrating
a known d.c. voltage from a stabilized power
supply and simultaneously testing the timer with
a Hewlett-Packard
electronic counter. The
accuracy of all time integrations was determined
to be +0.5 per cent. The integration time required to give invariant averages was determined experimentally and a typical integration
time was 100 s. The Disa correlator had a
frequency response which was flat from about
6 cps to 20 kc and was down 3dB at about 3 cps.
Since free jets and other free shear flows are
characterized by the presence of large eddies
of low frequency, the rather high cutoff at 3 cps
could introduce some error in turbulence
measurements. For x/d, less than about 20,
however, measurements of Ju, the axial tur-

and I. J. WYGNANSKI

bulent intensity, using a single wire and the

0.05 cps filter agreed with those made with an
x-wire and the Disa correlator. Hence, all
measurements of ?,;;2 and G (for xid, < 20)
were made using the correlator. For x/d, > 20,
measurements were made using the 0.05 cps
high pass filter circuit to determine the tendency
of the jet towards self-preservation.
Proper interpretation
of measurements of
transverse velocity fluctuations and shear stress
using x-wires requires that the x-wire is aligned
symmetrically about the mean flow direction
and that the velocity component sensitivities
are directly calibrated (as normal component or
cosine law cooling is not valid [6, 71). In the
present investigation, the mean flow direction
at any point in the flow field is dependent upon
A,/Ai and U,/Ui. Thus, as it is difficult and
time consuming to determine the mean flow
direction in a high-intensity turbulent field
[8], the data was taken with an x-wire probe
parallel to the centerline of the jet. The results
were corrected for tangential cooling [7] assuming the probe was aligned with the mean flow
direction. Corrections for angular deviation from
the mean flow direction were considered using
the equations in [7]. The mean flow directions
were computed from continuity considerations
based on similar velocity profiles for the fully
developed region. A mean flow direction of
- 10 relative to the axis of the jet corresponds
to a y/y,,,
= 2.06 in the region. For the value of
y/y,,, the computed corrections for angular
deviation are 5 per cent for 2,4 per cent for z,
and 2 per cent for 2. These corrections are
rather small and thus were neglected, especially
in view of the uncertainty in the interpretation
of the mean flow direction when the flow is
intermittent. Also because the entrainment is
less in the developing region, the angle of
inclination of the mean flow to the jet centerline
should be smaller than in the fully developed
jet. Thus, the above estimation of errors resulting from misalignment should apply at least
to y/y,,, = 2.06. For y/y,,, > 1, the local
turbulent
large and
..~ ~~~~.intensitv, becomes auite
I

COAXIAL

TURBULENT

approaches unity near the outer edge of the jet.

In this region, even the response of a linearized
constant-temperature
anemometer is nonlinear
in the velocity component fluctuation because
of the effects of large fluctuations in flow direction [9]. No correction for these effects was
attempted.
Mean velocity traverses were made to ensure
that the jets were symmetrical about their
common geometrical axis. The flow is considered
incompressible and the overall Reynolds number
was large enough so that its effects could be
ignored.

DISCUSSION

1449

JETS

be Uref D, (x - x,,)- 1 and (x - x,,), where Uret

is determined from the total jet momentum flux,
since far downstream this is the only parameter
of importance. In defining Urefr it must be kept
in mind that for U,/Ui > 1, the static pressure
in the inner potential core may be somewhat
lower than the ambient pressure, thus creating
a static pressure gradient. For the flows to be
considered here, this effect was kept small by
limiting the maximum value of U,/Ui to 10 as
will be discussed subsequently. The static
pressure gradient will therefore be neglected in
determining the total jet momentum.
Uref is defined by,

AND RESULTS

The

initial flow close to the nozzle exit

consists of two potential cores separated by an
annular mixing region and another mixing
region between the outer jet and the ambient
air. The width of each core decreases approximately linearly with downstream distance and
the cores terminate when the annular mixing
regions join as shown in Fig. 1. The flow is then
entirely turbulent and developing in the downstream direction until the jet becomes identical
to the simple axisymmetrical free jet.
Tendency

toward self-preservation

Far downstream the turbulent flow becomes

fully developed and self-preserving and the
velocity at any point depends only on the
position of the point and the total jet momentum.
In this region, the distributions of mean
velocity and turbulent intensity can be expressed
[lo] in terms of functions of a single variable
? = Y/l,,
U = Cf(rl)
I2 = u,2g,(q) etc.

(1)

(2)

where u, 01(x- x,)-i, I,,+ - x,,), and x,, is

the distance of virtual origin from the nozzle
exit plane. The scales of velocity and length will

where Jr is the total jet momentum flux from

the nozzle. Then equations (1) and (2) become
U
-=
U ref

&

f(Y/X - x,)

(4)

For the degenerate cases Ui = U, and U, = 0,

equations (4) and (5) reduce directly to the
results of Townsend [lo] for the single free jet
and thus provide a direct comparison with the
classical experiments.
The velocity profiles become similar when the
velocity scale is normalized by the maximum
velocity, U,, (which occurs on the axis far
downstream) and y/(x - x,,) as shown in Fig. 3.
The shape of this profile agrees well with the
measurements of Hinze [l]. The decay of Ue
with x and the growth of the characteristic
width, Y,,,,~,defined as the outermost position
where U = $J,,,
are shown in Figs. 4 and 5.
Both quantities are insensitive to changes of
velocity and area ratios and agree well with
the similarity arguments.
It is interesting to note that the measurements

1450

F.

l-f. CHAMPAGNE

and

I.

J.

WYGNANSKI

0 23.4
0 15.67

L!.

4 o.50

Single

jet

----_

z yiz
1
Hinze OI 01.

(The shaded area represents

scatter in this measurement)

0.25

FIG. 3. Normalized

velocity profiles in similarity

region.

0.507
0.203
OJOI
0.245
0.492
0492
0.492
0.492
Decoy of the

FIG. 4. Variation

of centerline

velocity

velocity

with x

along

the 5

COAXIAL

TURBULENT

1451

JETS

2-

Symbols

the
the

the same

0
f
tie
Jw

ep+

0
0

IO
ix+,4

FIG. 5. Growth

20

30
MD 0

of y,,+ the characteristic

jet width, with x.

FIG. 6. uz profiles plotted in terms of universal

asfor

decay of velocity
axis

function

gl(q).

along

1452

F. H. CHAMPAGNE

of the characteristic width collapse on our curve

at
X - X,
-_>8
D,
while the centerline velocity does so at
x - XI,
~
> 11.
D*
The variation of the distance between the plane
of the nozzle and the hypothetical origin, x,,, was
almost randomly scattered with
-.+% < 4.

QJ

The tendency of self-preservation was examined

by measuring the axial turbulent intensity and
the results are plotted in terms of the universal
function, gl, in Fig. 6. The flow bezomes selfpre~rving su~~iently far downstream
% 40 corresponds to 2
1

however, the universal function itself is significantly different from that given by Corrsin [ 1l]
and Laurence [12]. This discrepancy is dis-

and I. J. WYGNANSKI

cussed in detail elsewhere

Fiedler [ 131).

(Wygnanski

Close to the nozzle exit the flow consists of

two potential cores and mixing regions as
shown in Fig. 1. The width of each core
decreases with downstream distance. The length
of the external core appears to be independent
of the initial velocity ratio U,/Ui and is equal to
approximately 8 times the width of the annular
nozzle. The length of the inner core, however
(Fig. 7), strongly depends on U,/Ui, as well as on
the area ratio. The effect is particularly sign%cant when the outer velocity is larger than the
inner velocity. This may be explained by the
relatively low pressure created in the inner core
(under these flow conditions) which bends the
outer jet inwards. When the outer jet is thinner.
i.e. AO/Ai = 1.28, the effect of the pressure
differential across it is more pronounced, bending the jet faster inwards and shortening the
inner core. When 0 < U,/Ui < 1, the length of
the inner core is somewhat longer (Fig. 7) than
for the single jet, and this may result primarily
from the decreased shear between the inner jet
and its surroundings. It may be mentioned that

10

9-

87-

Length of the tnner pctantiol axe

6-

o A@, = 2.94

5-

o AdA/a, =I.28

4-

3f
2l-

Q--I k
0

/
3

i
5

,
6

, 0
8

$
FIG.

7. Length

and

of the inner core as a function of the velocity

ratio for two area ratios.

a
/
9

(II

COAXIAL

TURBULENT

when
V,/Vi = 1, the remnants
of the up
stream boundary layers quickly disappear and
the two cores merge into one.
When U,jUi > 1, the disappearance of the
external core is characterized by a rapid reduction of U,,, near the top of the outer core as
shown in Fig. 8. The same effect is noticeable
for both area ratios, and it is followed by a
short length at which U,,, is virtuahy constant.
When 0 < U,/Ui < 1. no such effect is ob-

The

decay

JETS

served and the termination of the interior core

results in a continuous reduction of U,,,
Figure 9 gives an indication of how the flow
developsfor thecaseof U,jUi = 5, A,/Ai = 1.28.
The dotted line gives the position of the inner
potential core as determined from the mean
velocity and turbulent intensity measurements.
Mean velocity profiles were obtained in the
core regions and beyond for 0 < UJUi d 10
and the two area ratios 1.28,294. Some typical

of the targert axial meon velocity

FIG. 8. The decay of the largest axial mean velocity.

A./A,

=I.28

u,/u,

=5-o

=a+ in.W

6fin. B

FIG. 9.

t453

Flow field development for U,/LI, = 5 and A,/Ai =

1.28. Abscissa indicates magnitude of mean velocity.

1454

F.

H.

CHAMPAGNE

and

I.

J.

WYGNANSKI

__
u&i

2.1

= IO-O

A,/A, = 2.94

2.0

2.0

R,, = 0.1 XlP

I.9

I.9

I?,, = 0.96X105

x/D*

1.7

0 Xz1.25in.
.a X = 2.25ii-t
0 X =4.251n.
@ X=6.25
l X =8,25m
0 X=12.25
v X=16.25
v X= 20.25in.
* X=24.25
v X=36.25
* X=48.25

1.6
I.5
I.4
1.3
Y
A!
*

I.2

0.606
1.16
2.14
309
407
6.05
co2
IO.0
11.98
17.9
23.85

I.3
I.2
c
2

1.1
I.0

0.9
O-8

0.8

0.7

0.7

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

01
0.0
0.4

0.5

06

@7

0.8

0.9

O-606
I.16
v 2.14
d 3-09
v 4.07
+ 6.05
A 8.02
l
IO.0
q

0.9

0.3

1.5
1.4

1.0

0.2

Rho = 0.96~105

I.7

0.1

=2:94

1.6

1.1

=5

R*i =0,2x105

I.8

1.8

q/q
AJAi

1.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

FO

UG,,

FIG. 10. Normalized

mean velocity profile for U,/Ui = 10,

A,/A, = 2.94.

FIG. 11. Normalized

mean velocity profile

A,IAi = 2.94.

for U,/CJ, = 5.

COAXIAL

TURBULENT

I455

JETS

2.6

29

2.5
2.4

2.7
U,/Uj

2.6
2.5

=0,5

R,,

Single

2.3

u,

2.2

= 105

Reo = 0.48 xl@

2.4

2.1
2.3

o X=125
A X=2.25in
0 X=4.25
l X=825
@ X=12.25

2.1
10
l-9 -

in.

0606
1.16
2.14
4.07
605

in
In
In

V X=24.251n

=o

D,;D,

=2.02
x/o,

0
A
0
@
.
0
A
+

1.7

11.98

l-6
1.5

1.6 -

1.4

1.5 -

*
I.4 -

&

0606
x = 1.25 in.
1.16
x = 2.25 in
X =425
tn. 2.14
x =a.25 in. 407
802
X =16,25in.
X ~2425
in 11.98
X =36,25in.
17.9
X =4825in.
23.85

1.3
1.2

l-3 -

1.1

1.2 -

1.0

I.1 1.0 -

0.9

0.9 08

=o

1.8

l-7 -

= 2.94

Uo/Ui
R

1.9

I.8 -

=Q

AdA,

R*i = 105

2.0

x/Do

2.2

jet

0.8

0.7
0.7 -

0.6

0.6 -

05

0.4 -

0.4

0.3 -

03

0.2 -

0.2

05

0.1 -

01

0.2

0.3

@4

0.1

05

06

07

0.8

0.9

O_
u

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

09

1.0

ffmw,

FIG. 12. Normalized mean velocity profiles for U,/U, = 0%

AJA,

= 2.94.

~~%ax

FIG. 3. Normalized mean velocity profiles for U,/ui

AJA, = 2.94.

= 0,

1456

F.

H.

CHAMPAGNE

and

I.

J.

WYGNANSKI

2.5
24

2.3

23

2.2

2.2 -

21

2.1 o
2.0 -

Uo/Ui

=10

AJA,

= 1.281

Rei'

v9-

Uo/Uj =5

2.0

=I.281

A,/A,
I.9

O,lXlOS

Reo=D49x105

RBi =0.2x105
Reo =0~49x105

1.8

\
1,7

1-8
I.7
0
A
0
*

15
1.4
?
$

I.3

1.2

16

x/Q

1.6

Q807
145
2.74
404

t4

6.72
0 10.47
+ 15.65
v 23.40

1.3

$
3
4

I.2

0
A
.

0807
I.45
2m

0
0
0
*
v

2.74
5.33
1047
1565
23.40

1-I
IO

I.1

0.9

I.0
0.9

O+

0.8

07

0.7

06

0.6

05

0.5

a4

0.4

o-3

0.3

o-2
0.1

0.2

.A,*

0.I
0

X/D,

1.5

0
0.1

0.2

@3

0.4

05

0.6

0.7 0.8

0.9

0.4

,I

0.5 O-6

0.7 @8

+$_.._

O-9

1-O

10

FIG. 15. Normalized

FIG, 14. Normalized

0.1 0.2 0.3

mean velocity profiles for U,!U,

AJA, = 1.28.

= 10.

mean velocity profiles

A,iA, = 1.28.

for U,iU,

= 5,

COAXIAL

TURBULENT

2.5

2.7 fl

2.4

2.6

2.3

2.5
2.4

2.2
2.1
2.0

1457

ETS

U,/Ui

=l.O

A./A;

2.3

1,281

/?,,=105

A,/Ai

= 1.281

/?,;=lW

2.1

I?,,
=0.49x105

1.9

U,/Ui = 0.5

2.2

R,, =025x1@
2.0

1.8

X/D,

1.7

0
A
0
@
v
+

1.6
1.5
l-4

1.9

0.807
2.74
5.33
10.47
15.65
23.40

r/o0
o 0807

1.8

A I.45
0 2.74
@ 533
l
791

1.7
1.6

0 13.08
+ 20.8

1.5

c
c 1.3
*

c
2

1.4

1.3

1.2

I On.3
CWB

1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
OL
0

0.1 0.2 0.3 0.4

0.5

06

0.7

0.8

0.9

1.0 1.1

u/(J.n,x

FIG. 16.Normalized mean velocity profiles for UJUi = 1.0,

A,/Ai

= 1.28.

FIG. 17.Normalized

mean velocity profiles for (IJLli = O-5,

A,/Ai

= 1.28.

1458

F.

H.

CHAMPAGNE

velocity profiles are shown in Figs. 10-17. In

these plots the velocity scale was chosen as the
maximum velocity at a given core section (not
necessarily on the C&of the jet) while the length
scale was the distance from the axis at which
the velocity has reduced to 3 its peak value.
When there are two such points, the outer one
is always chosen. Sufficiently far downstream,
these velocity profiles become similar. The effect
of changing the area ratios on the relative size

and

I.

J.

WYGNANSKI

of the external core may be observed by comparing Figs. 12 and 17 for U,/Ui = @5.
Turbulence measurements were taken for
three different cases : for A,!A, = 1.28, measurements for U,/Ui = 05 and 5.0 were taken to
determine the effect of the velocity ratios on the
turbulent field in the development region, and
for U,/Ui = 5.0, measurements were also taken
for 4,iAi = 2.94 to determine the effect of
varying the area ratio.

35

3.0

U*/iJj

=m

A,/Ai

= I:281

Re9ei =I05
f?,i

=O:25x105

X/D,

2-5

Q807
. 1.45
@ 2.74
l

FIG.

18. Distribution
of axial turbulence
Ui = OS. ,4,.,/l, = 1.28.

intensity

for

COAXIAL

TURBULENT

Figures 18-23 show the distribution of the

axial and radial turbulent intensities for U, jUj =
O-5and 5-Oand both area ratios. The distribution
of the turbulent intensities is closely related to
the shape of the mean velocity profile, as might
be expected from consideration of the production terms in the turbulent energy equation.
The dependence of the intensities on the
gradients in the mean velocity is rather striking.
For example, after the disappearance of the
inner core for U,/Ui = 5, the magnitudes of
both intensities on the jet axis dropped and
reached their minimum at approximately the

UJU, = w5
A,,/A, = 1,201

1459

JETS

axial position coinciding with the disappearance

of the outer core as shown in Figs. 24 and 25.
The reduction is especially evident for An/A, =
1.28 where the external core disappeared quicker
than for the case of A,/Ai = 2.94.
The distribution of the turbulent shear stress
E is shown in Figs. 26-28 and the dependence
on the mean velocity gradients and the extent
of the potential cores is evident. When U~/Ui < I
the turbulent shear stress and @U/&J) are only
negative, but for U,jUi > 1, the outer jet
initially accelerates the inner flow, (XJ/dy) is
positive and so is G. The distribution of ?,Z
and uo reach their more familiar shapes (as in a
single free jet) when the mixing takes place across
the entire flow and the mean velocity gradient
(XJ/C?~)is negative across the entire jet.
Note that for AojAi = 1.28, the turbulent

R,, = 102
3C

R,, = 0.25~105

UJe =50

xl4

A,/A,

0 @807
l 1.45
0 2.74

Disappears

at$z

3.0

= 1.281

ff,, = 0.2~105
2.5

X/o,
0 0.807
a 1.45

a 533
= 7.91
* 1308

0
it

c1
F
e
$

1.5

2.10

274
%, 4.04
l
533
l

20

lD47
15.62

almost

classical

mixing

region

I.0

0.5

FIG. 19. Distributions of radial turbulence intensity for

CJ,/lJ, = @5. A,IA, = 1.28.

FIG. 20. Distributions of axial turbulence

LlO/U, = 5.0. A,/A, = 1.28.

intensity for

F.

H.

CHAMPAGNE

FIG. 21. Distributions

of radial turbulence
U,/Lf, = 50. AJA, = 125.

intensity

and

I. J.

WYGNANSKI

for

1.5

1.0

0.5

a-----I

Fm

005

22. Distributians
U,jV,

040

of axiai turbufenoe
= 5.0, _&/AI = 2.94.

0.15

intensity

10

far

COAXIAL

TURBULENT

JETS

3.0

20
$\
\

1.0

FIG. 23. Distributions of radial turbulence intensity for

u,/Ui = 5.0, A,//$ = 2.94.
N.

0.11
0
II
ISl
I

j!
0.0,

FIG.

24.

Variation of axial turbulence inknsity along the jet

centerline with x.

1461

1462

F.

H.

CHAMPAGNE

and

1. J.

WYGNANSKI

0+375

0.025

FIG. 25. Variation

Radiol

0
A

of the radial turbulence

the centerline with x.

intensity

velocity fluctuations

%Iu,

WA,

5.0
0.5

I.281
1-281
2.94

R&l05

0,492
0.961
0.245

along

the $

R,, xl05

0.203
1015
0,203

along
u*/lJ = 0.5
AJA,=

I.281

R,, = w
R,, = 0.25 xl05

XlD,
0 0,807
A 1.45
n 2.74
l 533
0 7.91
* 1308

FIG. 26. Turbulent

shear stress distributions

A,/A, = 1.28.

for l/,/U,

= 05.
C

COAXIAL

TURBULENT

1463

JETS

)-

i4&=5~0

u,/q =50

A&J, = 1.281

AJAi=2.94

F&=0.2x105

.i$,=0.2x105

R,, = 049x105

F?,,= 0~96x105

44

O-

0 0807
h. 1.45
0
l

x/Do

2.c

2.10
2.74
4.04

5.33

Q
@

10.47
1562

0.616

A
@

1.16
2.14

3.09
4.07
6.04
8.02
10.0

0
0
v

ld

1.1

FIG. 27. Turbulent

shear stress distributions

A, 8Ai = 1.28.

for U,/ Ui = 5.0,

FIG. 28. Turbulent shear stress distributions for U,/U, = PO,

&;A, = 294.

intensities are smaller when U,/Ui = 0.5 in the

initial region (small x) than for U,/Ui = 50.
Coupled with the result that the inner potential
core is smaller for Ub/Ui = 5-Othan for U,/Ui =
O-5; this indicates that for a fixed A,/A,, UJU,
should be greater than one to enhance rapid
mixing between the two streams.
No attempt to compare with the simplified
model proposed by Morton [5] was undertaken
as there was no meaningful way of dehning
the boundary separating the inner from the
outer jet. Also, the model does not treat the

part of the developing region containing the

potential cores where much of the mixing
between the jets takes place.

REFERENCES
1. J. 0. HINZE, Turbulence. McGraw-Hill,
New York
(1950).
2. S. B. STARK, Mixing of gas streams in a flame, Zh. Tekh.
Fiz. 23, 1802-1819 (1953).
3. V. A. .~uJTYUNO~, Concerning mixing processes in
coaxial turbulent streams, Zzv. Vyssh. Ucheb. Zuved.
No. 11, 207-215 (1963).

1464

F.

H.

CHAMPAGNE

and

4. N. A. CHIGIER and J. M. BEER, The flow region near

the nozzle in double concentric jets, J. Eur. Engng 86.
797804 (1964).
5. B. R. MORTON, Coaxial turbulent jets, Inr. J. Hat Muss
Trunsfer 5,955-965
(1962).
6. F. H. CHAMPAGNE, C. A. SLEICHER and 0. H. WEHRMANN, Turbulence
measurements
with inclined hotwires, Part I: Heat transfer measurements
with inclined
hot-wires, J. Fluid Mech. 28, 153-175 (1967).
7. F. H. CHAMPAGNE and C. A. SLEICHER, Turbulence
measurements
with inclined hot-wires, Part 2: Hot-wire
response equations, J. Fluid Mech. 28, 177-182 (1967).
8. W. G. ROSE, A swirling round turbulence jet, J. Appl.
Mech. 29,615-625 (1962).

ETUDE

EXPERIMENTALE

DES

I. J.

WYGNANSKI

9. W. G. ROSE, Corrections
to average measurements
in
unsteady flow, A.S.M.E. Symposium on Measurement
in Unsteady Flow, pp. 85-89 (1962).
10. A. A. TOWNSEND, The Structure of Turbulent Sheur
Flow. Cambridge University Press, London (1956).
11. S. CORRSIN and M. S. UBEROI, Further measurements
on the flow and heat transfer in a heated turbulent air
jet, Tech. Notes Nat. Adv. Comm. Aero.. Wash., No.
1365 (1949).
12. J. C. LAURENCE,Intensity, scale and spectra ofturbulence
in the mixing region of free subsonic jet, Nat. Adv.
Comm. Aero., No. 1292 (1956).
13. I. WYGANSKI and H. E. FIEDLER, Some measurements
in
self preserving jet, J. Fluid Mech. 38, 577-612 (1969).

JETS

TURBULENTS

COAXIAUX

R&mn&Lecoulement
correspondant
a deux jets coaxiaux est Ctudie experimentalement
a laide dun anemometre a fil chaud. On fait varier le rapport des sections de sortie externe et interne aussi been que les
vitesses de sortie de chaque jet. La distribution
des vitesses moyennes, les intensites de turbulence et les
tensions de cisaillement sont dtterminees
dans differents cas. On discute le dtveloppement
du champ des
vitesses et son approche des conditions daffinitt. Les nombres de Reynolds bases sur le diametre a Iejection
varient entre 0 et lo5 et les vitesses sont suffsamment
basses pour que lecoulement puisse &tre considere
comme incompressible.

EINE

EXPERIMENTELLE

UNTERSUCHUNG
VON
DUSENSTROMUNGEN

KOACHSIALEN

TURBULENTEN

Zusammenfassung-Mit
einem Hitzdrahtanemometer
wurde das Geschwindigkeitsfeld,
das zwei koaxiale
Stromungen
erzeugen, experimentell
bestimmt. Das Flachenverhlltnis
der Lusseren und inneren Diise
wurde verandert,
ebenso die Austrittsgeschwindigkeit
aus den Dtisen. Fiir verschiedene
Fiille wurde
die Verteilung der mittleren Geschwindigkeiten,
des Turbulenzgrades
und der Schubspannungen
ermittelt. Die Entwicklung
des Striimungsfeldes
und dessen Annlherung
an einen selbsterhaltenden
Zustand
wird besprochen.
Die auf den Diisendurchmesser
bezogene Reynoldszahl
variiert von 0 bis 105, und die
Geschwindigkeiten
waren geniigend klein, so dass die Striimung als inkompressibel
betrachtet
werden
kann.

3KCIIEPMMEHTAJIbHOE

hnioTaqki~-c

HCCnEAOBAHHE
TYPBYJIEHTHbIX
CTPYtl

KOAKCHAJIbHbIX

IIOMOQbIO
TepMOaHeMOMeTpa
3KCllepPIMeHTaJlbHO
MCCJleROBaJIOCb IIOJle
CKOpOCTI4,06pa30BaHHOeRByMFl
KOaKCBaJIbHbIMH CTpJ'flMM.OTHOLUeHHe
IUIOWaRei BHeUtHerO
IIBHJ'TpeHHerO COIIeJI,KaKEi CKOpOCTb IIOTOKa,BblXO~FlIL(erO113 KaHc~OrOCOIIJIa,H3MeHHJI&iCb
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Typ6J'JleHTHOCTH II CfiBIlrOBbIX HaIIpiSKeHUti. PaCCMaTpHBaeTCH
pa3BHTHe
IIOJIR TeYeHMR M
ero npa6naHteurie
K C~CT~F~HIU~ caMono~~epmm3arrrrr-t.
Qicna
Peirronbxca,
paccqaTbrnaeMbre
HO @laMeTpy
COWIa,
I13MeHRnMCb OT 0 a0 I@, a CKOPOCTH 6urnu AOCTaTO'JHO HH3KIIMK RJlH
Toro,9~06bI MO~HO
6b12-10
cwTaTb
Te4eHxe HecxwfaeMbIM.