Studies of the Developing Flow Between

Concentric Cylinders With the Inner

k. n. a s t i l l

Tutrs unSrMidjorjM^. Cylinder Rotating

Associate Professor of

The present paper is part of the third phase of an investigation of the phenomena and
variables which control the rate of heat transfer in the air gap of a rotating electrical
machine. Experimental studies were made in the developing adiabatic flow of air in an
annulus having a rotating inner cylinder and stationary outer cylinder. Experiments
were made to determine when Taylor vortices, and/or turbulence occurred as a function
of Reynolds number of the flow; rotational speed of the inner cylinder; and distance from
the inlet. Lines of demarcation of the several modes of flow were established for posi-
tions along the axis of the annulus.
Two length effects were present: the distance to the point where vortices occurred and
the distance required for the vortices to develop. A criterion is presented for determining
the point in a developing flow where Taylor vortices originate. The value is given in
terms of the dimensionless parameter,

v i n

A discussion of the occurrence of vortices in well-developed flows is presented.

Introduction T h e s e regions are related t o axial flow and speed of rotation in

Fig. 1.
o N S I D E R A B L E interest has been shown in recent
CI O T h i s paper presents quantitative measurements to determine
years in the p r o b l e m of viscous flow between concentric cylinders lines of demarcation of these m o d e s of flow in terms of R e y n o l d s
with one or b o t h of the cylinders rotating. Such systems are of numbers, T a y l o r numbers, and distance f r o m the entrance. Cri-
interest in the design of cooling systems f o r rotating electrical teria f o r predictng the onset of the instability are presented.
machinery; chemical mixing or drying machinery; and t h e y T h e data are compared t o results f r o m other sources for fully de-
suggest possible application t o combustion systems or condensers. veloped flow.
T h e o b j e c t of the w o r k reported here was to study the adiabatic Historical Review. In his famous paper of 1924, T a y l o r [2] used
flow of a viscous fluid in the entrance region of a smooth-walled the m e t h o d of small disturbances to predict the conditions under
annulus having a rotating inner cylinder and a stationary outer which a flow between concentric rotating cylinders would b e c o m e
cylinder. Specificially, an experimental p r o g r a m was c o n d u c t e d unstable and the nature of the vortices which developed. He
t o determine the conditions under which T a y l o r vortices or ran- demonstrated the validity of his results b y simple experiment.
d o m turbulence occurred in a gas flow as a function of the f o l l o w - B o t h the analytical prediction and the experimental results have
ing variables: flow R e y n o l d s n u m b e r ; speed of rotation of the been verified a number of times since. F o r the case of a rotating
inner cylinder; and distance f r o m the inlet. Essentially this adds inner cylinder and a stationary outer cylinder, these vortices
the variable of axial length to the work of K a y e and Elgar [ l ] 1 occur in pairs having opposite rotation and are in the f o r m of
who showed that with an axial flow between concentric rotating
cylinders four modes of flow exist. T h e s e are:

1 Laminar flow
2 Laminar flow with T a y l o r vortices
3 Turbulent flow
TURBULENT FLOW
4 Turbulent flow with T a y l o r vortices PLUS VORTICES
1 Numbers in brackets designate References at end of paper. 13
UJ UJ
Contributed by the Heat Transfer Division of THE AMERICAN 5 >
SOCIETY OF M E C H A N I C A L E N G I N E E R S a n d p r e s e n t e d a t t h e ASME-
AIChE Heat Transfer Conference and Exhibit, Boston, Mass., August
0 SPEED OF ROTATION OF INNER CYLINDER, N
11-14, 1963. Manuscript received at ASME Headquarters, April 23,
1963. Paper No. 63—HT-15. Fig. I Schematic representation of different regions of flow in annulus

•Nomenclature-
characteristic parameter of equation ( 1 1 ) p = pressure
transition defined in equa- Ft = geometrical correction factor Re = R e y n o l d s number
tion ( 8 ) defined in equation ( 3 ) (Re), = length R e y n o l d s number
(Re),i = characteristic parameter de-
A* = critical value of A * — axial distance f r o m entrance to
fined in equation ( 1 2 )
a = w a v e number transition point
r = radius
b = annulus width p — dimensionless parameter de- n = radius of inner cylinder
D = integral defined in fined in equation ( 4 ) (Continued on next page)

Journal of Heat Transfer august 1 96 4 / 383

Copyright © 1964 by ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 vortex rings with their axis of rotation in a plane normal to the
axis of the cylinder, Fig. 2.
While T a y l o r ' s w o r k was in an annulus with n o axial flow (closed
ends) the problem of cooling electrical machinery required an
understanding of the occurrence of T a y l o r vortices in the presence
of axial flow. S o m e experimental work had been done b y F a g e
[3] and Cornish [4] and an incorrect analysis performed b y G o l d -
stein [5] when K a y e and Elgar [1] began their study. M o r e re-
cently analytical solutions h a v e been published b y D i Prima [12),
Chandrasekhar [11] and [13], and experimental results presented
b y D o n n e l l y and Fultz [6] and Snyder [14]. T h e s e last five
papers treated cases with v e r y low R e y n o l d s numbers. Assuming STATIONARY
OUTER CYLINDER
that the flow was fully developed, K a y e and Elgar [1 ] e m p l o y e d
hot-wire anemometers to show the existence of the four m o d e s of Fig. 2 Pairs of counterrotating Taylor vortices
flow previously noted, and to indicate lines of demarcation a m o n g

1
AIR INLET
them. T h e y reported results for air with t w o geometries, pre- AIR EXIT ANNULUS
senting the results in terms of R e y n o l d s number and T a y l o r
SCREENS
number. In the work of K a y e and Elgar, as in the present work,
* U - L 1U-
the R e y n o l d s number is

w2b |Tr -SUCTION

Q v
Re : (1) PIPE
v

and the T a y l o r number modified for cases where the ratio of the
gap t o radius of the annulus is n o t negligible is ^BOUNDARY LAYER MOTOR
BLOWER SUCTION FLOWMETER

(Ta), (2)
-MANOMETERS

BLOWER
where the geometrical factorr , L.

7Ti ( b \"»
(3)
FLOWMETER MUFFLER
Fig. 3 Schematic arrangement of apparatus
and
1697 \ ~ 2r^,/ *
T h e inner cylinder was m a d e of paper-laminated phenolic
fastened to a 0.875 in. shaft. T h i s shaft, supported in ball bear-
ings, was driven b y an electric m o t o r with V-belts. T h e inner
P = 0.0571 j^l - 0.652 + 0.00050 - 0.652 ' (4) cylinder was 47 in. long turned and sanded to give an outside
The Experimental Program diameter of 2.751 ± 0.001 in. T h e lucite outer cylinder was
Description of Apparatus. T h e general arrangement of the 4 8 ' A in. long with an inside dia of 3.785 ± 0.02 in. T h i s was
apparatus is shown in Fig. 3. Air was drawn into a stilling supported in phenolic end plates fastened t o a frame of welded
chamber equipped w i t h baffles and screens. T h e air passed into angle iron 1 1 / 2 X IV2 X '/«. A d j u s t i n g screws allowed the outer
the annulus through an entrance nozzle having an elliptical profile cylinder t o be located concentric t o the inner cylinder to within
on b o t h inner and outer walls, and a contraction ratio of 5 ' / 2 : 1 - ± 0 . 0 0 2 . T h e s e dimensions provided an annulus of 0.517 ± 0.002
T h e outer wall of the nozzle was made of phenolic c o m p o u n d , the in. W h e n assembled, the inner cylinder wall had a run-out of
inner wall was polished brass. T h e inner nozzle wall was held 0.00125 at the entrance end, and 0.0010 at the exit end.
stationary. Boundary-layer suction was p r o v i d e d through the T h e main air flow was supplied b y means of a 1 hp blower
nozzle cone with the boundary-layer drawn off at the edge of the regulated b y means of a variac. Air left the annulus section,
rotating cylinder, then through the space between the nozzle and entered a small baffled mixing chamber, and then entered the
the shaft. T h e diameter of the inner cylinder was m a d e Vi6 in. piping t o the blower. In this section of two-in. piping a sharp-
larger than that of the nozzle cone to p r o v i d e a space f o r b o u n d a r y - edged orifice was installed to meter the flow. E a c h orifice was
layer suction. T h e suction section was sealed at the upstream calibrated with a gasometer. I t was necessary to a d d a plenum,
end b y means of graphite packing in a gland. Boundary-layer and a set of screens between the blower and orifice d u c t in order
suction flow was metered with an orifice meter, and maintained at to minimize turbulence in the test section.
a b o u t 5 percent of the total flow. W i t h this a m o u n t of suction I t was estimated that the flow rates in the test section might be
the boundary-layer thickness was equivalent t o t h a t which in error b y as m u c h as 3 percent. Errors in metering the b o u n d -
would develop 0.12 in. f r o m the leading edge of a flat plate in ary-layer suction were estimated as ± 1 0 percent because of dif-
laminar flow f o r R e y n o l d s numbers up to 2200. T h e profile was ficulties in calibrating the very small diameter orifices used, and
n o t improved sufficiently for higher suction rates to j u s t i f y their because of possible leakage p a s t the end seal. Speed of rotation
of the inner cylinder was measured b y counting at low speeds and

Nomenclature
)'2 = radius of outer cylinder v0 = velocity of inner cylnder wall Se* = tangential displacement thick-
rm — mean radius of annulus w = axial c o m p o n e n t of velocity ness
Ta = T a y l o r number uVnax = m a x i m u m axial v e l o c i t y X = wavelength
(Ta)m = modified T a y l o r number w = mean axial v e l o c i t y v = kinematic viscosity
U = free-stream velocity z = axial length p = density of fluid
v = tangential c o m p o n e n t of ve- So = tangential boundary-layer 03 = angular velocity of inner cylin-
locity thickness der

384 / august 1 964 Transactions of the ASSV1E

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 b y a strobatac at the higher speeds. Errors in counting were d e -
termined to be less than one part in sixty, while errors in speed
measurements with the strobotac were estimated less than 2
percent.
Flow Visualization Studies. T o b a c c o smoke, f r o m a simple s m o k e
generator, introduced into the inlet mixing chamber, provided ex-
cellent means f o r flow visualization at low R e y n o l d s numbers.
A flow rate and speed of rotation of the inner cylinder was set
and smoke introduced. T h i s was illuminated in a radial plane
along the direction of flow b y a slit of light supplied at the side.
Photographs were then taken of the flow using a 9 X 12 c m
Voigtlander camera. Illumination was supplied b y means of an
E. G. and G . Strobolume flash unit with a 40 microsec flash. T h e
arrangement is pictured in Fig. 4 while several photographs of the
smoke patterns can be seen in Figs. 5 and 6. In the series of
sample photographs, Figs. 5 and 6, each mark along the right-
hand edge of the p h o t o g r a p h represents V2 in. of axial length.
R<•(54 R.' 170 • 170
I t can be seen that the instability begins as a series of oscillating C M ; 21800 (T.lj 62500 (TO; 196,000

waves occurring first near the wall of the rotating inner cylinder. Fig. 5 Smoke patterns in flow—full entrance effect of rotation on
A s the fluid m o v e s downstream, the w a v e motion grows outward transition
finally curling over to f o r m the pairs of vortices. A regular flow
pattern developed as the flow traveled farther along the d u c t , with
flow cells which were trapezoidal.
W h e n the R e y n o l d s number was held constant and the speed of
the rotating cylinder was increased, the point of transition m o v e d
u p the annulus, toward the entrance. If the speed of rotation was
decreased, the transition point m o v e d d o w n the tube. This can
be observed b y comparing the photographs of Fig. 5.
Setting the speed of rotation and increasing the R e y n o l d s n u m -
ber caused the point of transition t o m o v e downstream. R e d u c -
ing the axial v e l o c i t y caused the point of transition t o m o v e
upstream again. T h i s can be seen in the series of photographs of
Fig. 6. T h e speed with which the vortices traveled downstream
increased when the axial v e l o c i t y was increased.
T h e distance f r o m the entrance to the onset of vortices was d e -
termined b y direct measurement f r o m the photographs, f r o m the
entrance mark t o the p o i n t where the first discernible wiggle
occurred in the flow. T h e results are presented graphically in
Fig. 9. Points are shown for several R e y n o l d s numbers for each
of the flow cases. T h e graphs show the modified T a y l o r number
at transition versus length R e y n o l d s number. Length R e y n o l d s
wL
number is defined as ( R e ) s = —.
v Re-170 Re • 4 3 3 Re " 5 7 2
Measurements of the wavelength of the disturbance were made <T„)U " 1 9 5 , 0 0 0 <TJm' 168,000 (T,i)u " 2 2 0 0 0 0

f r o m the photographs. T h e average wavelength was measured

Fig. 6 Smoke patterns of the flow at about the same Taylor number to
to be 2.056, which is equivalent t o a wave number,
show the effect of varying the axial flow
a = 2TT6/X = 3.08.

While these measurements are n o more accurate than ± V « in., [12] c o m p u t e d w a v e numbers of 3.12 at a R e y n o l d s number of 2
they can be compared t o the wave number at a R e y n o l d s n u m b e r and 3.15 at a R e y n o l d s number of 120 for a full}' developed flow
of zero of 3.12, measured b y D o n n e l l y and Fultz [6], D i P r i m a with a parabolic distribution of axial velocity.
Hot-Wire Studies of Transition. Visual observations of the flow
were limited t o R e y n o l d s numbers less than GOO. I t was i m -
TRANSPARENT
OUTER CYLINDER possible t o determine the onset of turbulence b y using smoke
injection. In order t o improve the accuracy of the measure-
OPAQUE INNER ments and to extend the range of the variables, a F l o w C o r p o r a -
CYLINDER REFLECTOR tion H W B hot-wire anemometer system was used, displaying the
o u t p u t f r o m the amplifier on an oscilloscope. T h e p r o b e was of
Tungsten wire, 0.00035 in dia mounted on long curved needles so
LAMP that the wire was located 0.5 in. upstream of the tip of the stem,
ENCLOSURE and displaced radially inward b y the same amount. A p r o b e
support with a traversing mechanism was mounted at any of
several stations along the axis on the outer cylinder. A t each
station a l / s in. dia access hole was made. E a c h hole was fitted
with a plug, numbered, and marked for proper orientation to insure
t h a t it would be located properly in the opening after being re-
BLACK BOX
m o v e d . These plugs were fitted b y hand, until no irregularity of
inner surface was detectable b y touch. T h e distance f r o m the
CAMERA leading edge of the inner cylinder to the center of the openings
was:
Fig. 4 Arrangement for photographing flow patterns

Journal of Heat Transfer august 1 96 4 / 385

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 o. L A M I N A R b. L A M I N A R + VORTICES
(Ta) m = 0 (Ta) m = 1 9 , 3 0 0

M
c. L A M I N A R -1- V O R T I C E S
(Ta) m = 3 0 , 0 0 0
d. T U R B U L E N T + VORTICES
(Ta)m- = 4 6 , 2 0 0

anemometer at midstream, station 14

Fig. 7 Oscilloscope traces from hol-wi

Station N o . 1 2 3 4 5 6 7 .STATION 3
Distance, in. 0.5 1.0 1.5 2.5 4.5 6.5 8.5 . L> I inch L / 2 b = 0.97

8 9 10 11 12 13 14 STATION 8 .STATION 10
10.5 12.5 16.5 20.5 24.5 30.5 36.5 L = 10 inches . L * 16 Inches
L / 2 b 1 9.6 ,L/2b»l5.4

Openings were arranged in a spiral around the tube. In the dis-

cussion which follows, p r o b e location will be given in terms of
station number. W i t h the p r o b e having curved needles at station
_i - STATION 12 1
5, the p r o b e tip was 4 in. f r o m the entrance of the test section. L= 2 4 inches L= 36 inches -
z 3
>- - L / 2 b = 23.2 L/2b = 34.8 -
Transition measurements were made with the p r o b e 0.05 in. LU 5.
CC ' STATION 14
f r o m the inner wall and at midstream, 0.26 in. f r o m the inner
wall. T h e air flow was set to give a prescribed R e y n o l d s number,
and the boundary-layer suction was adjusted t o r e m o v e 4 - 5 per- 10' 2 3 5 10' 2 3 5 10' 2 3 10' 2 3 5 10* 2 3 5 10* 2 3
cent of the air entering the annulus. TAYLOR NO. (^^rP^fe)
Transition was determined b y the nature of the trace on the
Fig. 8 Demarcation lines for flow regions in annulus. Probe 0.05 in.
oscilloscope. Laminar flow produced a s m o o t h line, Fig. 7(a).
from inner wall.
W h e n laminar flow with vortices occurred, a s m o o t h sinusoidal-
like trace appeared, Fig. 7(b). In this case, the change in the These vortices increased in intensity with a further increase in
trace occurred so abruptly at transition that the readings of speed the rotating speed. T h e frequency with which vortices passed the
could be repeated to within 5 rpm. If the rotating speed was in- wire increased with increasing R e y n o l d s number. T h e speed at
creased further, the signal amplitude grew Fig. 7(c) ( f r e q u e n c y re- which transition f r o m turbulent flow t o turbulent flow plus
mained about constant) until a speed was reached where the vortices occurred was reproduceable t o within ± 5 percent.
peaks of the sinusoidal trace were noticeably irregular. This was T h e s e observations are consistent with the descriptions given
designated turbulent plus vortices, Fig. 7(d). It was difficult to b y I i a y e and Elgar [1 ] f o r their criteria f o r determining m o d e s of
determine the transition f r o m laminar plus vortices to turbulent flow.
plus vortices with great precision. T h e range of reproducibility F r o m these results, the modes of flow could b e m a p p e d for each
in the measured critical speed was ± 1 5 percent for this type of axial station in terms of the modified T a y l o r number and the flow
transition. R e y n o l d s number. M a p s were made f o r t w o radial positions of
If laminar flow was present at a large R e y n o l d s number and the probe, 0.05 in. f r o m the inner wall and 0.26 in. f r o m the inner
the T a y l o r number again increased, the s m o o t h trace of laminar wall (midstream). M a p s made at stations 3, 5, 8, 10, 12, and 14
flow became the jagged trace associated with turbulent flow. I t at 0.05 in. f r o m the inner wall are shown in Fig. 8. Open points
was difficult t o ascertain transition f r o m laminar t o turbulent denote origination of vortex flow; solid points represent onset of
flow better than ± 8 percent f r o m the mean. Initially turbulence turbulence.
with zero rotation occurred at length R e y n o l d s number of a b o u t T h e conclusions of the visual studies were reaffirmed. As the
70,000. Screens were added upstream of the annulus; a plenum p r o b e was m o v e d toward the inlet, the speed of rotation had to be
was placed downstream between the blower and metering orifice; increased to induce transition f o r any given R e y n o l d s number.
and a nest of paper straws 6 in. long was p u t in the piping just F o r a given T a y l o r number, the transition occurred at lower
ahead of the plenum. Following these changes, laminar flow was R e y n o l d s number when the p r o b e was located near the entrance.
indicated at station N o . 14 with m a x i m u m blower speed at zero Also, transition occurred at the inner (0.05 in.) radial position
rotation. This corresponded to a length R e y n o l d s number of at a smaller T a y l o r number than for midstream.
a b o u t 140,000. It was rather difficult to establish the onset of turbulence, as
T h e transition f r o m turbulent flow to turbulent flow plus can be seen f r o m the spread of these data. F o r p r o b e positions
vortices was easily detected b y increasing the speed of rotation at near the entrance, the blower was inadequate to obtain velocities
high R e y n o l d s numbers. Sharp peaks appearing on the trace large enough for transition t o turbulent flow.
marked the first vortices being shed and passing b y the h o t wire. Cross plots of the transition data are given in Figs. 9, 10, and 11
I t was the point where the first peak was detected, that was d e - with the modified T a y l o r number at transition being plotted
fined as the transition point to turbulent plus vortices flow. against the length R e y n o l d s number for a series of diameter

386 / august 1 964 Transactions of the ASSV1E

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 lo'r

Re=3000
Re=l500
Re=500

10* 10' ^ 10'

WL
LENGTH REYNOLDS NO.

Fig. 9 Taylor number al onset of vortices as a function of length Reynolds Re = 1000

number at 0.05 in. from inner wall. Points are from smoke studies.

2 3 5 10'

C R I T I C A L TAYLOR NO.

Fig. 12 Distance a disturbance travels along the annulus in growing

radially from 0.05 in. to 0.26 in. from inner wall

..- 4000

\Re=3000
=2000
Re =

10* 10' 10* 5,. 10* 10*

L E N G T H R E Y N O L D S NO. ( R e ) , * ^

Fig. 10 Taylor number at onset of vortices as a function of length

Reynolds number at midstream
TAYLOR NO. ^ ' ^ m b ' ^ J ^

Fig. 13 Comparison of critical Taylor numbers at midstream and at wail.

Stations 10 and 5.

and f o r example, a T a y l o r n u m b e r of 10 s , the transition occurs at

length R e j ' n o l d s numbers of 9.2 X 10 4 at midstream and 5.4 X 10 4
at 0.05 in. f r o m the inner wall. F o r this geometry, these corre-
spond to lengths of 23.5 in. and 13.9 in. f r o m the entrance, respec-
tively. I t required nearly 10 in. of length for the disturbance
t o g r o w f r o m the wall t o the midstream. F o r the same T a y l o r
number, b u t at a R e y n o l d s number of 250, the disturbance grew
f r o m the wall t o midstream while the fluid traveled 1 in. d o w n
the annulus. T h i s calculation has been extended t o several
values of R e y n o l d s numbers using the curves of Figs. 9 and 10,
and the results h a v e been p l o t t e d in Fig. 12. T h i s set of curves
10* 10* 10* shows the length along the annulus t h a t the disturbance travels
LENGTH REYNOLDS NO. (Re), - W L in growing f r o m a radial p o i n t 0.05 in. f r o m the inner wall t o the
Fig. 11 Taylor number for transition to turbulence as a function of length midstream.
Reynolds number
T h u s it has been shown that there are t w o length effects in the
flow as t h e transition t o T a y l o r vortices occurs in the annulus.
R e y n o l d s numbers. Curves in Figs. 9 and 10 are for transitions T h e first of these is the distance along the annulus t o the p o i n t
resulting in the T a y l o r vortices at the radial stations 0.05 in. where the instability begins. T h e second is that distance re-
f r o m the wall a n d a t midstream, respectively. There has been quired f o r the disturbance t o g r o w in the radial direction. As a
p l o t t e d in Fig. 9 d a t a f r o m the photographs of smoke studies. A s result, t w o observers would report different critical T a y l o r n u m -
the d a t a are f o r slightly different values of R e y n o l d s n u m b e r i t is bers in a given flow at a given axial position if t h e y were measur-
difficult t o m a k e a quantitative comparison, b u t the d a t a f r o m ing the transition at different radial stations, Fig. 13. I n order t o
the p h o t o g r a p h s p l o t consistently with the h o t wire data. emphasize these effects further, t w o sets of curves are shown f o r
C o m p a r i n g the values of transition at 0.05 in. f r o m t h e wall t o transition at t w o axial positions, and t w o radial positions. W i t h
transition at midstream leads t o s o m e interesting conclusions. this curve, i t is possible t o c o m p a r e the critical T a y l o r numbers
T h e general shape and character of the curves are the same at at midstream w i t h t h e critical T a y l o r number near the inner wall
b o t h locations. I n f a c t , the curves f o r a R e y n o l d s n u m b e r of 100 for a given flow at a station. Curves are given f o r stations 5 and
are nearly identical. H o w e v e r , at a R e y n o l d s number of 4000, 10.

Journal of Heat Transfer august 1 96 4 / 387

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 In discussions that follow, the onset of T a y l o r vortices in a flow
will always refer t o the earliest measurable disturbance, that is
the transition of the flow at 0.05 in. f r o m the inner wall.
Transition to Turbulence. Using the curves of Fig. 8 showing the
transition f r o m laminar flow t o turbulent flow, or laminar flow
plus vortices t o turbulent flow plus vortices, a cross p l o t was made
of the critical T a y l o r number versus the length R e y n o l d s number,
Fig. 11. T h e difference between points at midstream and those
measured 0.05 in. f r o m the inner wall was smaller than the ex-
perimental error, hence, in Fig. 11 the cross p l o t of these results
is given only for the inner radial station. T h e s e curves are limited
b o t h b y the difficulty in obtaining accurate measurements and b y
the limited number of points. I t is clear f r o m Fig. 11 that the ef-
fect of diameter R e y n o l d s number is small, rather it is the length
Reynolds number that dominates. T h e a s y m p t o t i c values of the
curves would give the critical T a y l o r numbers for transition in a
fully developed flow ( R e ) , = co. I t is evident that as the Fig. 14 Explanatory sketch showing boundary-layer development and
principal dimensions of annulus
Reynolds number decreases, the critical T a y l o r n u m b e r increases.
A t zero R e y n o l d s number a value of T a y l o r number in excess of
106 would be expected. F o r zero T a y l o r number, transition t o cor,8g* IS/f
turbulence would be expected only for diameter R e y n o l d s n u m - (8)
V * ''l
bers greater than 1800 and at length R e y n o l d s numbers in excess
of 106. SI as
where the displacement thickness is defined
A Characteristic Parameter for Transition. F o r small annuli, the
T a y l o r number can be expressed as «»[(>', + 5e*Y- ~ ' i 2 ] = 2 I vrdr (9)
J ri
m "o bib
Ta = — \ — (5) Solving for 8g* yields
v 1 r,
= -/•, ± Vn 2 + 2D (10)
T h e m o v e m e n t of the point at which T a y l o r vortices occurred
suggests that the critical T a y l o r number in a developing flow with only the positive root accepted. Here,
might be written in terms of the thickness of the tangential
boundary layer, rather than the w i d t h of the annulus. In other D • (ID
words, that the circulation is confined t o an annulus with a thick-
ness equal t o that of the tangential b o u n d a r y layer, 8g. F o r
definitions of these dimensions, Fig. 14 has been provided. A Evaluation of A*. Measurements of v e l o c i t y profiles were made
new T a y l o r number was expressed as to determine the tangential displacement thickness as a function
of R e y n o l d s number, T a y l o r number, and distance f r o m the en-
trance. W i t h these results, the value of A * could be determined
unSe J 5 e
Ta = (6) at the p o i n t of transition to v o r t e x flow.
V e l o c i t y profiles were obtained b y employing a hot-wire
anemometer to determine the direction and magnitude of the
It was subsequently discovered that Brewster and Nissan [7] had
v e l o c i t y at intervals across the annular gap. A F l o w Corporation
used the distance f r o m the inner wall to the n o d e of v e l o c i t y as a
probe was used having 0.00035 in. dia wire with straight needles,
characteristic length t o describe the occurrence of the instability
V2 in. long. T h e p r o b e was calibrated in an air flow f r o m a
between concentric-counterrotating cylinders, or between cylinders
standard A S M E long radius nozzle, 0.512 in. dia.
partially filled with liquid. Gortler [8] in his analysis of the
instability in the b o u n d a r y layer near a concave wall expressed F l o w direction was determined b y orienting the h o t wire so
transition in terms of the parameter that its axis was a few degrees a w a y f r o m the probable flow
direction. A Wheatstone bridge circuit with the wire as one leg
was b r o u g h t into balance. T h e wire was then turned, upsetting
— J - the bridge balance. Turning was continued until the balance was
v i n restored. Bisecting the two angles at which the bridge was
balanced gave the direction of the mean v e l o c i t y at that point.
Liepmann [9 and 10] subsequently made measurements t o con-
Readings could b e made, and reproduced to an accuracy of '/<
firm this, expressing the p o i n t of transition in terms of this
degree with this technique.
parameter, with m o m e n t u m thickness as the charateristic length.
A v e l o c i t y profile was obtained b y determining the v e l o c i t y
Returning t o the dimensionless group of equation ( 6 ) , it can b e
direction at a point, then determining the magnitude of the mean
regarded as the p r o d u c t of a R e y n o l d s number, and the ratio of a
v e l o c i t y with the wire oriented so t h a t its axis was at 90 deg t o
radial pressure force to a tangential m o m e n t u m force, as it can be
the flow direction. Readings were m a d e f r o m a distance of 0.03
shown that
in. f r o m the inner wall t o 0.02 in. f r o m the outer wall. Profiles
8p _ 8^ were obtained f o r R e y n o l d s numbers of 200, 7 0 0 , 1 2 0 0 , and 1700;
(7) T a y l o r numbers of 0, 10,000, 40,000, and 90,000; at stations 2, 6,
pv2 r
10, 12, and 14. T h i s yielded 80 profiles of the velocity. A s a
T h e value of 8g is a function of the axial v e l o c i t y and distance f r o m check on the accuracy of the measurements the flow f o u n d b y
the entrance, and so can b e changed b y changing the R e y n o l d s integrating the measured axial c o m p o n e n t of v e l o c i t y across the
number of the axial flow. N o t i c e also t h a t the value of the annulus was compared t o the flow measured with the flow meter.
boundary-layer T a y l o r n u m b e r can b e changed at a given location F o r each profile, the difference between the flows was expressed
b y changing the angular speed of the inner cylinder. as a percent of the metered flow. A standard deviation for the 80
A s the tangential boundary-layer thickness 8g is n o t easily profiles was found t o b e 0.13.
measured, it has been replaced b y displacement thickness, 8g*, to Curves were plotted in terms of the dimensionless velocities
formulate the parameter w/wmnx and v/v0. Sample profiles are presented in Fig. 15. T h e

388 / august 1 964 Transactions of the ASSV1E

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TANGENTIAL VELOCITY momentum-integral analysis for laminar flow. T h e details of
this are omitted here [15 ]. It was assumed that the axial velocity
profile was fully developed but the tangential b o u n d a r y layer
grew. This reduced the four equations (three m o m e n t u m and
c o n t i n u i t y ) to two. A parabolic v e l o c i t y distribution was as-
sumed in the tangential direction, while the axial velocity was
assumed parabolic in one case, and uniform in the second. T h e
tangential b o u n d a r y layer was found to depend on the parameter

2( « )
(Re),, = (12)
Re(6)

being independent of the speed of rotation as indicated b y the

experimental results. A s the differences between computed
I T values of boundary-layer thickness was less than 15 percent (fig <
A D
0.9) for the t w o axial profiles assumed, a single curve was used.
DISTANCE FROM ENTRANCE. INCHES
T h u s , for the solid curve of Fig. 16, the assumption is a uniform
Fig. 15 Velocity distribution in the inlet of the annulus with air
axial v e l o c i t y at the entrance, b e c o m i n g the parabolic profile at
a point a w a y f r o m the entrance, where ( R e ) 2 i 0.033. The
numerical result showed g o o d agreement with the data in the
KEY range of small ( R e ) , i . Agreement at large ( R e ) , i was improved if
10000 a linear profile was assumed for the tangential velocity. T h i s is
40000 also a better approximation to the measured profiles in that range.
;
90000 E m p l o y i n g the transition test data, Fig. 9, and the experi-
mental curves of So*, the parameter A * was evaluated for each
set of R e y n o l d s number and T a y l o r number. These results are
presented in T a b l e 1. A constancy was noted f o r the critical
value, A * . A mean value of A * was computed to be 24.3 with
a standard deviation among the data of 0.08 based on the mean.

Table 1 Values of Ac" based on experimental results

Transition
Re Ta (Re), L/D bg* A*
200 10,000 4800 24.00 0.340 24.7
40,000 900 4.47 0.230 27.5
90,000 440 2.19 0.180 28.5
700 10,000 22400 31.70 0.340 24.7
40,000 5600 7.94 0.210 24.0
90,000 3000 5.25 0.155 23.4
1200 10,000 43000 35.50 0.340 24.7
40,000 11600 9.60 0.200 22.5
90,000 7000 5.78 0.165 24.7
1700 10,000 67200 39.40 0.300 20.5
40,000 22100 12.48 0.200 22.5
90,000 13000 7.59 0.160 24.0
10 20 30 40
DISTANCE FROM ENTRANCE, L , INCHES Values of A * based on the displacement thickness obtained
Fig. 16 Growth of the displacement thickness along the annulus. The numerically also showed constant (Ac* « 22) until the value of
dashed line is a faired curve through the experimental data points. The ( R e ) 2 i ~ 0.05, whereupon A c * begins to increase. A value of
solid line is from the analysis. ( R e ) z i = 0.05 corresponds to a value of Sg = 0.7. This implies a
well-developed flow corresponding t o small R e y n o l d s numbers at
ordinate is the distance f r o m the inner wall and the abscissa large distances f r o m the inlet.
represents the dimensionless velocity. E a c h profile represents Transition in Well-Developed Flows. While the experimental re-
the results at a given distance from the entrance. T h e distance sults indicate t h a t t h e criterion for the occurrence of vortices in
f r o m the //-axis t o the p o i n t of zero v e l o c i t y on the profile repre- a developing flow is A * > 24, it does n o t obtain in flows with a
sents that distance. Points of transition in the m o d e of flow are well-developed tangential velocity. If it did, transition in a fully
marked on the curves, A designating laminar to v o r t e x ; D developed flow would be independent of R e y n o l d s number, and
designating laminar-vortex to turbulent-vortex. N o t i c e h o w the would occur at a T a y l o r number of 9170. T h i s c o m e s f r o m the
profile of tangential v e l o c i t y changes as angular m o m e n t u m is requirement that Ac* = 24 with fig* for fully developed flow which
transferred radially outward when the vortices occur. A radial is 0.343 for a parabolic profile. In their analyses of the stability
c o m p o n e n t of v e l o c i t y which cannot be neglected is introduced b y of fully developed flow between concentric rotating cylinders with
the vortex. T h i s must be considered in noting the magnitude of axial flow, Chandresekhar [11 and 13], and D i P r i m a [12] show
the velocity after transition has occurred. that the critical T a y l o r number is a function of R e y n o l d s number
Determination of A* at Transition. T o evaluate A * at transition, in the range of small R e y n o l d s numbers. Furthermore their
the displacement thickness was computed for each v e l o c i t y p r o - results are supported b y experiments of Donnelly and Fultz [6],
file with equation (10). T h e integral was evaluated b y Simpson's Snyder [14], K a y e and Elgar [1 ], and the author.
rule. I t was noted t h a t the displacement thickness was a strong In Fig. 17 is shown a line representing critical T a y l o r number
function of R e y n o l d s number and was relatively unaffected b y versus R e y n o l d s number f r o m D i P r i m a ' s analysis [12] and a line
speed of rotation. Graphs were m a d e of displacement thickness representing the criterion Ac* = 24 for the case of fully developed
versus length f r o m inlet f o r each of the four R e y n o l d s numbers tangential velocity, i.e., ( T a ) m = 9170. Also plotted are the
for which data were obtained. A single dashed curve was faired experimental points for station 14, chosen because it would repre-
through the data, Fig. 16. T h e solid curves in Fig. 16 s h o w the sent fully developed flow most closely. D a t a points are all in the
development of the displacement thickness computed f r o m a " u n s t a b l e " region, and follow the general contour of the lines of

Journal of Heat Transfer august 1 96 4 / 389

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TCMT
2000-

II
PROBE 0 . 0 5 INCHES
3 -
FROM INNER W A L L
j ! I600|-
2
<1
(\J
tr 2 -
t-
£C
1200- UJ 8 CO
5 ° UNSTABLE Q
STABLE

>-
800- §
o >-
UJ z LU
(r => tt
o 1
CD
400 10 s 10'
FULLY DEVELOPEII <b
FLOW (DIPRIMA)
— IT'S"5" TAYLOR NO. [jJ-]

16 Fig. 18 Demarcation lines for flow regions in annulus; station 10.

L = 16 in. L/2b = 15.4 comparison between theory and experiment.
TAYLOR NO. ) * 10"'

Fig. 17 Lowest limit of stability in a long annulus. Data at L/De = 35. ary layer has developed fully. In the former case the transition
is characterized b y a T a y l o r number based on displacement thick-
ness. Experimental results show that transition occurs when
stability. A t R e = 1400 the critical T a y l o r number starts in-
creasing. TliiB corresponds t o a value of ( R e ) 2 i = 0.10, which is
equivalent t o a value of Sg = 0.9. T h e transition f r o m then on . unSe* Ifo
A* ; 7~ > 24-
occurs in a flow with a developing tangential b o u n d a r y layer. v in
T h e thesis which can be induced f r o m these observations is
that two types of transition are possible. T h e first is where the F o r a fully developed flow, the critical T a y l o r number increases
flow is well developed when critical speed is reached. This with increasing R e y n o l d s number for low R e y n o l d s numbers.
corresponds t o the T a y l o r problem and has been analyzed b y D i Transition will be of the boundary-layer t y p e or fully developed
P r i m a [11]. Here the critical T a y l o r number is dependent u p o n t y p e , whichever is attained first. T h e m i n i m u m T a y l o r number for
the flow R e y n o l d s number. T h e second t y p e occurs in a d e v e l o p - which a flow is stable an infinite distance f r o m the annulus inlet is
ing flow (similar to the Gortler p r o b l e m ) and is characterized b y a function of R e y n o l d s number f o r R e y n o l d s numbers less than
the parameter A* > 24. In a fully developed flow, the transition 200, being equal to 1700 at zero R e y n o l d s number, and is 9200 for
will occur at the point indicated b y the limit shown b y D i P r i m a R e y n o l d s numbers over 200. F o r most physical systems vortices
or at Ae* = 24 whichever occurs first. T h u s , while vortices occur will originate in a developing flow. Exceptions are cases of small
at ( T a ) m = 4200 for R e = 100 they should o c c u r at ( T a ) m = 9170 R e y n o l d s numbers in long annuli.
if the R e y n o l d s number is greater than a b o u t 200. If the flow is I t was shown that laminar, laminar plus vortices, turbulent,
increased to R e = 1500, the transition p o i n t would m o v e d o w n and turbulent plus vortices occurred in the flow, and flow maps
the annulus f r o m station 14 and would require increasing the were m a d e at a number of axial stations. W h i l e detection of
rotational speed t o restore it, at which time A* = 24. transition to turbulent flow was difficult it appeared that this
There is g o o d agreement between the data and D i P r i m a ' s t y p e of transition was primarily a function of the length R e y n o l d s
result (Fig. 17). T h e data appear t o follow D i P r i m a ' s curve number.
b e y o n d T a = 9170 then reverse slightly t o w a r d that limit. T h i s
is also evident in one of the experiments of K a y e and Elgar [1 ].
I t m a y be that with care in adjusting the speed that it is possible Acknowledgments
t o exceed the limit A* > 24 at low R e y n o l d s numbers. This T h e research presented in this paper represents a portion of the
might be similar to the case of being able t o extend laminar flow doctoral thesis of K e n n e t h N . Astill [15]. I t was part of an ex-
in tubes to R e y n o l d s numbers in excess of 2000. tensive s t u d y that was conducted for several years in the R e -
Fig. 18 was prepared to show h o w the t h e o r y might b e used to search L a b o r a t o r y of H e a t Transfer in Electronics at M . I T .
predict transition t o vortex flow. T h e curve was generated for under the late Professor Joseph K a y e . T h e A u t h o r wishes t o
station 10, using the a b o v e criteria, with displacement thickness acknowledge the important role Professor K a y e played in this
based on the m o m e n t u m integral analysis. T h e tangential b o u n d - work.
ary layer is fully developed at R e = 485. T o this point the lines I should also like to thank M r . Richard M a c W h o r t e r and Mrs.
of stability are based on D i Prima's analysis or T a = 9170. I r m a Wallace for their help in the preparation of this paper.
B e y o n d this the curve was generated for the developing flow Ac*
= 24. T h e agreement with the data is excellent.
References
1 J. Kaye and E. O. Elgar, "Modes of Adiabatic and Diabatic
Conclusions Fluid Flow in an Annulus With an Inner Rotating Cylinder," TRANS.
ASME, vol. 80,1958, pp. 753-765.
In a developing axial flow in an annulus with an inner rotating 2 G. I. Taylor, "Stability of a Viscous Fluid Contained Between
cylinder, T a y l o r vortices originate near the inner wall, and g r o w Two Rotating Cylinders," Phil. Trans., A, vol. 223, 1923, pp. 289-
radially outward, the vortices m o v i n g in the direction of flow. 343.
This leads t o t w o length effects: one to the point where the in- 3 A. Fage," The Influence of Wall, Oscillations, Wall Rotation
and Entry Eddies on the Breakdown of Laminar Flow in an
stability occurs and a second t o a point where the vortices are Annular Pipe," Proc. Roy. Soc., A, vol. 165, 1938, pp. 513-517.
fully developed. T h e distance f r o m the entrance t o the point 4 R. J. Cornish, "Flow of Water Through Fine Clearances With
where vortices occur increases with increasing R e y n o l d s number Relative Motion of the Boundaries," Proc. of Roy. Soc., A, vol. 140,
and decreases with increasing T a y l o r number. T h e r e appear t o b e 1933, pp. 227-240.
5 S. Goldstein, " T h e Stability of Viscous Fluid Flow Between
t w o t y p e s of transition t o v o r t e x flow: one when the tangential
Rotating Cylinders," Proc. Camb. Phil. Soc., vol. 33, 1937, pp. 41-
b o u n d a r y layer is developing and one when the tangential b o u n d - 61.

390 / august 1 964 Transactions of the ASSV1E

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 6 R. J. Donnelly and D. Fultz, "Experiments on the Stability of
Spiral Flow Between Rotating Cylinders," Proc. Nat. Acad. Sci., vol.
46, 1960,pp. 1150-1154.
7 D . B. Brewster and A. H. Nissan, "The Hydrodynamics of
Flow Between Concentrie Cylinders-I, Flow Due to Rotation of
Cylinder," Chem. Eng. Sc., vol. 7,1958, p. 215.
8 H. Gortler, "Uber Eine Dreidimensionale Stabilitat Laminarer
Grenzschichten an kon Kaven Wanden," Zamm 21, 1941, pp. 250- 252.
9 H. Liepmann, "Investigations on Laminar Boundary Layer
Stability and Transition on Curved Boundaries," NACA Wartime
Report W-I07, 1943.
10 H. Liepmann, "Investigations of Boundary Layer Transition
on Concave Walls," NACA Wartime Report W-87, 1945.
11 S. Chandrasekhar, "The Hydrodynamic Stability of Viscid
Flow Between Co-Axial Cylinders," Proc. Nat. Acad. of Sc., vol. 46,
1960,pp.137-141.
12 R. C. Di Prima, "The Stability of a Viscous Fluid Between
Rotating Cylinders With an Axial Flow," Jour. Fluid Mech., vol. 9,
1960, pp. 621-631.
13 S. Chandrasekhar, "The Stability of Spiral Flow Between
Rotating Cylinders," Proc. Roy. Soc., A, vol. 205, 1962, pp. 188-197.
14 H. A. Snyder, "Experiments on the Stability of Spiral Flow
at Low Axial Reynolds Numbers," Proc. Roy. Soc. A, vol. 265, pp.
198-214.
15 K. N. Astill, "Modes of Adiabatic Flow in the Entrance Re-
gion of an Annulus with an Inner Rotating Cylinder," PhD thesis
M. I. T., Cambridge, MlI6s., 1961.

Journal of Heat Transfer AUG U S T 1 9 64 / 391

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use