STABILITY O F LAMINAR WAKES
T h e s i s by H a r r i s Gold
I n P a r t i a l Fulfillment of the Requirements F o r the Degree of
Doctor of Philosophy
California Institute of Technology Pasadena, California
1963
ACKNOWLEDGMENTS
The author wishes to e x p r e s s his sincere thanks and apprecia- tion to Professor L e s t e r Lees for h i s guidance throughout the course of this investigation. The author would a l s o like to acknowledge many helpful discussions with Psofe s s o r Toshi Kubota and Dr. Anthony
Demetriades. He a l s o wishes to thank Kiku Matsumoto for programming a l l the IBM computer computations and Mrs. Geraldine Van Gieson for her outstanding typing of the manuscript.
The author acknowledges the receipt of a Del Mar Science Foundation Fellowship for the 1959- 60 academic year, a Boeing Fellowship for 1960-61, and a RAND Corporation Fellowship i n
Aeronautics for the y e a r s 1961-62 and 1962-63. This project was made possible through the sponsorship and with the financial support of the U. S. Army Research Office and the Advanced Research P r o j e c t s Agency.
The author would like to e x p r e s s h i s gratitude to his wife, Sandra, for her encouragement and per serverance throughout his studies.
ABSTRACT
This investigation d e a l s with the effects of compr essibility on the hydrodynamic stability of wake flows. It is found that the effect of t e m p e r a t u r e i s two-fold: ( I ) , a s the wake c o r e t e m p e r a t u r e i n c r e a s e s , the range of Mach n u m b e r s over which n e u t r a l and self- excited subsonic disturbances c a n exist a l s o i n c r e a s e s ; (2) a s long as the r e l a t i v e Mach number i s below the c r i t i c a l M a c h number the n e u t r a l inviscid wave number will d e c r e a s e with increasing c o r e temperature, implying that a hot wake will be m o r e stable than a cool one.
The a n a l y s i s of Batchelor and Gill for the inviscid stability of axi- s y m m e t r i c incompressible j e t s h a s been extended t o the m o r e g e n e r a l problem of c o m p r e s s i b l e wakes and jets. It i s shown that the r e s u l t s a r e d i r e c t l y analogous to those obtained for the two-dimensional problem. The sinuous ( n = 1 ) mode is the m o s t unstable allowable mode. This unstable mode is observed i n a hypersonic wake.
TABLE O F CONTENTS PART
Acknowledgments A b s t r a c t
Table of Contents L i s t of F i g u r e s L i s t of Symbols I. Introduction
11. Formulation of the Problem
11. 1. Outline of the Stability Problem 11.2. Two-Dimensional F r e e Shear Flows.
Small Disturbance Equations and Boundary Conditions
11. 2, a. Viscous, Incompressible P r o b l e m 11. 2 . b. Inviscid, Compressible Problem 11. 3. Axi-Symmetric F r e e Shear Flows.
Equations and Boundary Conditions
11. 3, a. Viscous, Incompressible Problem 11. 3, b. Inviscid, Compressible Problem
II.
4. Mean Flow Model111. Stability of Two-Dimensional Incompressible Wake Flows 111. 1. Solutions f o r Small a R
111. 2. Idealized P r o f i l e s a t Long Wavelengths 111. 3. Inviscid Limit
111. 4. Large, Finite Reynolds Nurnber Stability
PAGE ii iii i v vii viii 1 3 3
IV. Stability of Two- Dimensional Compressible Wake Flows 44 IV. 1. Inviscid Disturbance Equation
a n d Outer Boundary Condition IV. 2. Solution of the Inviscid Equation IV. 3. Numerical Results
IV. 4. The Hyper sonic Wake P r o b l e m 60 V. Stability of Axi- Symmetric Compressible Wake Flows 64
V. 1. Similarity Between the Small Disturbance Equations and Boundary Conditions f o r
Axi-Symmetric and Two-Dimensional Flows 66 V. 2. Inviscid Disturbance Equation and Outer
Boundary Condition 69
V. 3. Singularities of the Inviscid Disturbance
Equation 76
V. 4. N e c e s s a r y and Sufficient Conditions f o r the
Existence of Inviscid Subsonic Disturbances 83 V. 5. Energy a n d Vorticity Relations 88 V. 6. Inviscid Stability of Axi-Symmetric Wake
Flows 97
VI. Suggestions f o r F u r t h e r Study 104
VII. Concluding R e m a r k s 106
R e f e r e n c e s 108
Appendix A
--
Boundary Conditions for the Axi-Symmetric P r o b l e m 113
Appendix B
- -
Two-Dimensional Wake Model 115 Appendix C--
Method of Solution of Tatsurni a n d3Xakutani31 f o r Small a R 120
Appendix D
- -
Solution of the Inviscid Equations f o rAmplified Disturbances 126
Appendix E
- -
Expansion about C r i t i c a l Point-Axi-Symmetric C a s e 130
Appendix F
- -
E x t r e m u m of Density-VorticityProduct 1 3 3
Appendix G
- -
Viscous Corrections i n the C r i t i c a lLa ye r 1 3 5
Table I
--
Neutral Inviscid Stability Character-i s t i c s 1 4 2
Table 11
--
Amplified, Invi scid StabilityC h a r a c t e r i s t i c s 143
F i g u r e s 1 4 5
LIST O F FIGURES
NUMBER PAGE
1 Modes of Instability 145
2 Typical Neutral Stability Curve f o r Wake-Type
P r o f i l e s 146
3 Wave Number vs. Complete Wave Velocity 147
4 F r e q u e n c y vs. Reynolds No. and Spatial
Amplification Rate 148
5 Direction of Propagation of the Wave F r o n t s 149 6 Relative Wave Velocity and C r i t i c a l Point vs.
T e m p e r a t u r e Difference 150
7 Relative Wave Velocity vs. Relative Mach Number 151 8 Inviscid ,Wave Nurnber vs. Square of Relative
Mach Number 152
9 Dimensionless Frequency vs. Spatial Amplification
Rate 153
10 Hyper sonic Axi s y m m e t r i c Wake
( a ) Mm
=
13, B a s e Diameter=
0.400",P
=
100 mrn. H g . , Air, Cone Angle = 16O 154(b) n = 1 Mode of Instability 155
vii
LIST O F SYMBOLS
The symbols used i n the p r e s e n t r e p o r t a r e i n g e n e r a l those commonly used i n the l i t e r a t u r e on hydrodynamic stability. In some r e g r e t a b l e instances a symbol will r e p r e s e n t m o r e than one item. To minimize the confusion, the different definitions of the s a m e symbol will have listed the s e c t i o n of the r e p o r t i n which they appear.
Dimensional Dimensionless Reference
Quantities Quantities Quantitie s
Two- Dimensional Flows :
Positional Coordinates:
longitudinal n o r m a l Velocity
Components : C A (x- c t )
longitudinal
-
\n/. =
u* - 5;
L A t w- c t )
-
n o r m a l ys = V' +
v * ~
"(7) 4-9 ( 4 e v "
Density
-
c d ( w - c k )P r e s s u r e pX = px + p XI P(Y) c ~ (
e
~ 1P,"
L A t w - ct)
T e m p e r a t u r e 7% =
TX +
T * ' T ~ Y )+
QCV)e 7."
Axi- Symmetric Flows:
Positional Coordinates:
axial r a d i a l angular
Velocity Components :
-
~ ~ ( s - r t ) tin+axial
9: -
w'+
w(*) q x i r )e
- - -
w * : q,X
-
0;-
- ,Acy.- ct) + c n 4)radial
7,
=9:
43:' qvcy~
4-9vCq) e -
angular
7
=91
+9;'
Angular WaveNumber:
Density f * = f w 4
f r ' -
P r e s s u r e
px .
p" + p''Temperature
Time
t"
tWave Length A* A
Wave Number A* = 2 q / h Y = I*/\
Di s tur banc e
-
C Propagation
c' - v e x C
Velocity
a speed of sound
b unknown constant (Section IV. 2 )
C~ drag coefficient
group velocity of disturbance
c
*/u,*
dimensionless group velocity glocal relative propagation velocity ( Eq. (5. 20) ) Section
III.
1, Appendix C )c h a r a c t e r i s t i c body dimension exp (ia (x-ct)
+
in#)pq.
(5.49)]G n'/a2 T' n
[
Eq. (4. 18)]G
defined by Eq. (5. 52)H
1
G (Appendix D)H* stagnation enthalpy
HZ=)
Hankel functions of o r d e r 1/3, f i r s t and second kind h dimensionless static enthalpy (Appendix B)~ ~ ( a f i r ) modified B e s s e l function of the second kind of o r d e r one
K ~ ( ~ E
r ) modified B e s s e l function of the second kind of o r d e r one k gradient of density-vorticity product (Section V) L* c h a r a c t e r i s t i c lengthL(z) Lommel function (Appendix G)
M
v*/ae*
relative Mach numberMe local Mach number outside the m e a n wake M~ local Mach number of disturbance
[Eq.
(5. 21)]Moo f r e e s t r e a m Mach number
m total wave number
N~ Eq. (5. 68)
P Eq. (5. 10)
a*(>,
t*) quantity of total flow6*(;*)
m e a n o r steady component of flow quantity Q * ~ (;*, t*) fluctuating component of flow quantity<*I(;*) fluctuation amplitude
q(y)
I
fluctuation amplitude for nearly parallel flowv*L*/~
e* wake Reynolds number g a s constantReynolds number based on displacement thickness local external Reynolds number based on d*
local external Reynolds number based on x*
local external Reynolds number based on 0*
L*
, position vectorr
r (Appendix F)
AT temperature e x c e s s a t centerline of wake
U transformed mean velocity component (Appendix B)
U* mean velocity component in x* direction A U velocity defect a t center of wake
V transformed mean velocity component (Appendix B)
V* c h a r a c t e r i s t i c velocity
W
1-
U (Appendix B)W
b a g (3 ? Y"
) (Section V)X transformed x coordinate (Appendix B) xi
X (1/ ?$ r T) (Appendix E)
Y transformed y cowdinate (Appendix B) Y ( 1 / ~ )( sw'
'
(Appendix E)r
disturbance vorticity- -
I '
mean vorticity%
r a t i o of specific heatsA dimensionless heat t r a n s f e r coefficient (Appendix B) 6 tan' l (ar/n) (Section V)
6* net heat t r a n s f e r (Appendix B)
c
l/(aR) 1/3 (Appendix G)Dorodnitsyn-Howarth variable Eqs. (2. 47) and (2. 56)
r
-
rCic (Appendix G)C
dimensionless momentum thickne s a (Appendix B) momentum thickness
viscasity
kinematic viscosity
S
Eq. (5. 10)G
p - a
= a G (Section 111 and Appendix C) T - - J q X ' q r i Reynolds shear s t r e s sf
r - r cSubscripts
c c r i t i c a l point
e local condition outside mean wake I imaginary p a r t of quantity
i point a t which disturbances begin to amplify o initial values (Appendix B)
o values a t a x i s (Appendix D) R r e a l p a r t of quantity
s neutral, inyiscid values
00 f r e e s t r e a m conditions 1, 2.. , f i r s t , second solution, etc.
Super s c r i p t s
A
complex conjugate
0 , 1 ) , 2 ) zero, first, second o r d e r quantities, etc.
A bar over a quantity indicates mean value.
P r i m e s generally denote differentiation with r e s p e c t to y o r r.
The few instances where p r i m e s denote a fluctuating quantity should not cause any confusion.
xiii
I. INTRODUCTION
Experimental 2-4 studies have shown that transition i n incom- p r e s s i b l e f r e e boundary l a y e r s is preceded by a linear and non- linear wave-type instability. The linear instability can be described
5-31
by the s m a l l disturbance theory of hydrodynamic stability
.
Thenon-linear instability is a v e r y complex phenomena and is not yet understood. Recently, the problem of laminar: turbulent transition in hypersonic wakes has a l s o been of considerable interest. The purpose of t h i s p r e s e n t investigation i s to study the effects of com- p r e s sibility on the hydrodynamic stability of wake- type flows.
The stability c h a r a c t e r i s t i c s of wake flows a r e relatively
insensitive t o Reynolds number, for sufficiently high Reynolds numbers, because of the o c c u r r e n c e of a point of inflection i n the density
vorticity product 22' 32m Therefore, interesting and important r e s u l t s can be obtained by considering the "inviscid limit" of the s m a l l dis- turbance equations, i n which the viscosity and conductivity of the fluid can be neglected to a c e r t a i n o r d e r .
This study will be r e s t r i c t e d to subsonic disturbances, i* e*
,
disturbances whose propagation velocity i s subsonic with r e s p e c t to the f r e e s t r e a m velocity. These disturbances have amplitudes that die out exponentially far f r o m the wake axis. If such disturbances exist then the m e a n flow is unstable to s m a l l disturbances. The question of the stability of a flow to supersonic disturbances h a s not been resolved 32-35
The inviscid stability c h a r a c t e r i s t i c s of two-dimensional in- compressible and c o m p r e s s i b l e wake flows a r e studied using Gaussian
distributions f o r velocity and t e m p e r a t u r e i n the Dorodnitsyn-Howarth variable. The incompressible wake was a l s o studied a t low Reynolds n u m b e r s and wave numbers. The r e s u l t s of Batchelor and Gill 36
and ~ i 1 1 ~ ~ f o r a x i - s y m m e t r i c incompressible wake-type flows have been extended t o include the effects of compressibility.
11. FORMULATION O F THE
PRQBLEM
An infinitesimally s m a l l disturbance i s imposed upon a m e a n o r a steady flow and the behavior of the amplitude of the disturbance i s examined a s t i m e p r o g r e s s e s . If, for l a r g e values of time, the disturbance is damped out, the motion is said to be stable; i f not, the motion is said to be unstable with r e s p e c t to infinitesimally s m a l l
disturbances. It is m u c h e a s i e r to prove that a motion i s unstable than stable. If the flow i s unstable to disturbances of a n y kind, even
P
-
the simplest kind, i t i s always unstable, but the flow m a y be stable with r e s p e c t to one type of disturbance and not another. In hydro- dynamic stability theory, the disturbance i s a s s u m e d to have a wave- like nature. The problem is to find c e r t a i n combinations of the wave number and wave speed of the disturbance and the Reynolds number of the m e a n flow f o r which the fluid motion i s unstable, o r neutrally stable.
11. 1 . Outline of the Stability Problem
-
The total flow c o n s i s t s of a time-independent o r m e a n component,
-
Q*, and a n infinitesimally s m a l l component, Q*', which i s both spaceand t i m e dependent:
Q"
(F*,
t") =-
Q* (?*) S Q*' ($*, t*) where( ~ * i /lad
< < 1The total flow, Q*, s a t i s f i e s the conservation equations of m a s s ,
momentum and energy and a n equation of state. The m e a n flow,
s*,
s a t i s f i e s the steady flow equations o r some approximation to them,
for example, the boundary l a y e r equations. The conservation equations f o r the disturbance a r e obtained by substituting e x p r e s s i o n s of the form, Eq. ( 2 . l), for the flow v a r i a b l e s into the total flow equatisns and
subtracting out the m e a n flow equations. The hydrodynamic stability equations a r e obtained f r o m t h i s s e t by neglecting quadratic and higher o r d e r t e r m s of the disturbance quantities.
The coefficients of the resulting linear p a r t i a l differential equations depend upon t h e m e a n flow quantities. Time a p p e a r s only a s the derivative (a/&*) and hence solutions containing a n exponential t i m e factor
hl
-
ia*c*t*Q*T
(y*,
t*) = q*' (;*) em a y be assumed. The resulting differential equations will contain the space coordinates a s the only independent variables.
This study will be limited to parallel o r quasi-parallel flows, i. e., motion i n which the m e a n n o r m a l velocity is z e r o o r v e r y s m a l l compared to the m a i n velocity component. The following o r d e r of magnitude relations f r o m boundary layer theory apply to the m e a n flow quantities:
where
longitudinal coordinate i n two-dimensional flow o r axial coordinate i n axi- s y m m e t r i c flow
nor e i n two-dimensional flow o r radial coordinate, r*, i n axi- s y m m e t r i c flow
6*
m e a n velocity component in x* direction*
m e a n velocity component i n y* directionR6* Reynolds number based upon displacement thickness.
The m e a n flow quantity,
G*
is a function of the position coordinates, x* and y*. Expand Q)k about the point x* = x*-
P '
The region adjacent to the point under consideration is taken t o be of the o r d e r of a few wavelengths of the disturbance i n the x* direction
Then f r o m Eqs. (2. 3b) and (2. 4b)
and f o r l a r g e values of aR6*
,
Eq. (2.4a) becomes- -
Q* (x*, y*) = Q*(x P
**
Y*)[
1+
0 (l/aRsL )] (2.4d)Therefore a l l m e a n quantities can be considered to be independent of the n o r m a l ( o r r a d i a l ) space variable to o r d e r (l/aR6+).
By considering disturbances that a r e spatially periodic, both i n the direction of flow and i n the direction perpendic'ular to the plane
e t r y of the m e a n motion, Squire 38 h a s shown for incompressible that two dimensional disturbances a r e l e s s stable than t h r e e -
dimensional disturbances. F o r c o m p r e s s i b l e flow, Dunn and Lin 39 have shown, by neglecting dissipation t e r m s and some t e r m s involving the fluctuating viscosity and t h e r m a l conductivity (which a r e valid a t m o d e r a t e Mach numbers), that a n equivalent two-dimensional disturbance i s not possible, but that the t r a n s f o r m e d three-dimensional disturbance equations a r e of the s a m e f o r m a s those for two-dimensional disturbances.
In particular, by neglecting viscosity and t h e r m a l conductivity, the three-dimensional disturbance equations a r e exactly of the s a m e f o r m as the two-dimensional ones. Therefore, important f e a t u r e s of the stability problem can be obtained by considering two-dimensional disturbances alone. F o r parallel and q u a s i - p a r a l l e l flow, the coefficients of the equations a r e independent of x* and consequently solutions of the f o r m
rV q*' (x*, y*)
=
might be expected. The exponent is purely imaginary since the dis- t u r bance m u s t be bounded f o r x* a t both 4- co and
-
go. F o r two- dimensional flows a disturbance of the f o r mQ*' (x*, y*, t*) = q*I(y*) e ia*(x*- c*t*)
will reduce the s e t of l i n e a r p a r t i a l differential equations to a s e t of o r d i n a r y differential equations i n y*.
T h e r e i s no d i r e c t analogue of S q u i r e ' s r e s u l t for axi- s y m m e t r i c parallel and quasi-parallel flows. However, Lessen, et a12' and P a i 2 7 indicate that for rotationally s y m m e t r i c disturbances, the incompressible disturbance equations, except for the obvious coordinate scale factors, a r e s i m i l a r to those f o r two dimensional disturbances. Batchelor and
Gill 36 show, by a suitable velocity transformation, that for d i s t u r - bances with a n angular dependence, the incompressible equations a r e exactly analogous to the two-dimensional ones, again, except for the obvious coordinate s c a l e factors. In Section V. 1 t h i s l a t t e r r e s u l t i s
,
extended to the c o m p r e s s i b l e case. The m e a n flow quantities for p a r a l l e l and quasi-parallel flows do not depend on the angular coordin- ate,$b,
and depend only on the coordinate n o r m a l t o the direction of the m e a n motion, r*, t o o r d e r ( l / a ~.
Since the amplitude of the6*
disturbance m u s t be single-valued with r e s p e c t to the angular coordin- ate, a disturbance of the f o r m
Q* (r*,
0,
x*, t*)=
q*l(r*) e ia*(x*- c*t)+inq)where n i s a n integer, m a y be assumed. The resultant s e t of ordinary differential equations will have r* a s the only independent variable.
I n Eq. (2.6) o r Eq.
[
- ( 2 . 7)] the disturbance amplitude qw(y*) [ o r q*'(r*) and the wave velocity c* a r e taken to be complex.I
The m a i n flow i s stable, neutrally stable, or unstable to t h e s e waves according to whether the imaginary p a r t of c* is negative, zero, o r positive, respectively. The quantity a* i s the wave number of the disturbance and is taken to be r e a l and positive. The r e a l p a r t of c*is the phase o r propagation velocity of the wavy disturbance.
-ia*c*t*
The assumption that the disturbance h a s the f o r m e
pq.
( 2 . 24 i s known a s the n o r m a l mode approach to hydrodynamic stability. If t h e r e a r e some values of a*c*, with cI*>
0,
s u c h that a non- t r ivial solution satisfying the disturbance equations and boundary conditions exists, then the flow is said to be unstable; i f not, i t isstable. Recently, the initial value problem h a s been emphasized by
Case 45
40-44 and Lin An a r b i t r a r y s m a l l perturbation is introduced into the flow at time, t*
=
0, and i t s subsequent motion i s followed by m e a n s of a n o r m a l mode expansion, If t h e r e is a single mode withc
* >
0 the perturbation grows exponentially with t i m e and is said toI
be unstable. The n o r m a l mode approach should be equivalent to the initial value method although t h e r e s e e m s to be some inconsistencies between the two r e s u l t s a s the Reynolds nurnber becomes infinite 40-42
4 5
L i n h a s pointed out that t h e s e c a n be resolved by considering the limit of a n o r m a l mode i n the viscous theory. This limit i s
-
not the n o r m a l mode i n the inviscid theory, and vice v e r s a , The n o r m a l mode approach i s applicable t o differential o p e r a t o r s having d i s c r e t e eigen-I
values while the initial value method should be used with singular
o p e r a t o r s and/or continuous eigenvalue s. However, i t appear s that the m o d e s leading to instability a r e associated with the d i s c r e t e eigenvalue s.
It is f o r t h i s r e a s o n that the n o r m a l mode approach will be used i n t h i s text.
The amplification r a t e of the disturbance is defined a s follows:
and
Q*' = Qi*' exp
i"
ti* a* cI* dt* *In the laboratory, the experimenter h a s a quasi- stationary problem. As the disturbance propagates downstream of i t s origin, i t s amplitude
changes i n both space and time. One m e a s u r e s the spatial
9
amplification r a t e , o r the r a t e a t which a disturbance will amplify with distance i n the m e a n flow direction. The wave speed i s a
function of both the wave number of the disturbance and the Reynolds number of the m e a n flow. If the parallel and quasi-parallel assumptions a r e made, then the group velocity of the disturbance, i. e., the velocity a t which the disturbance energy m u s t propagate
46 ,
isThe spatial amplification r a t e i s
and
The spatial amplification r a t e i s constant for parallel flows. F o r quasi-parallel flows, the spatial amplification r a t e is computed a t each s t r e a m w i s e station by using a m e a n velocity profile that i s
a s s u m e d to be independent of x a t that station. The total amplification ( o r decay) of the disturbance a s i t m o v e s downstream i s found by the piecewise integration of the local amplification ( o r decay) r a t e s .
F o r wake-type flows, i t is expected that the spatial amplification r a t e s depend upon the decay of the m e a n velocity profile. Therefore, i t i s convenient to t r a n s f o r m f r o m a coordinate s y s t e m i n which the o b s e r v e r i s fixed i n the body to one i n which the o b s e r v e r is fixed in a fluid a t r e s t . In this l a t t e r system, the o b s e r v e r s e e s the tfvelocity
defect" of the wake. [ S e e Sketch 2. 1 J
Body- Centered Coordinates Coordinates Fixed i n Fluid a t Rest
Sketch 2. 1
In t h i s coordinate s y s t e m the wave h a s a propagation velocity (c*
-
Ue*). Let the m e a n flow be dimensionally r e p r e s e n t e d by a c h a r a c t e r i s t i c length,L*,
and a c h a r a c t e r i s t i c velocity, V*, and the t e m p e r a t u r e , density, p r e s s u r e and viscosity by their external values,so that [see L i s t of Symbols. :
V* and
L*
will be taken to be the velocity defect a t the centerline[w(o)
=
I ] and the half-width of the wake, respectively [section 11. 41 .
The disturbance equations and boundary conditions will be derived for two-dimensional [section 11. 2'1 and axi-symmetric [section 11.31 flows.
The outer boundary conditions for the compressible problem will be
discussed in detail in Section IV. 1 (two-dimensional c a s e ) and
Section V. 2 (axi-symmetric case). The quantities a r e defined in the L i s t of S p b o l s .
11.2. Two-Dimensional F r e e Shear Flows. Small Disturbance Equations and Boundary Conditions
11. 2. a. Viscous, Incompressible Problem
The incompressible, s m a l l disturbance equations a r e a limiting c a s e of the complete c o m p r e s s i b l e equations. This c a s e i s obtained by neglecting the viscous dissipation and heat conduction t e r m s i n the conservation equations and assuming that the m e a n temperature, p r e s s u r e , density, viscosity and t h e r m a l conductivity a r e constants.
In addition, i f the gradient of the t e m p e r a t u r e fluctuation vanishes at the a x i s and the outer edge of the m e a n flow, the t e m p e r a t u r e fluctuation and hence the density fluctuation can be s e t equal t o zero32. The
dimensionless s m a l l disturbance equations a r e 3 2*
Continuity
x- Momentum
y- Momentum
*
P r i m e s indicate differentiation with r e s p e c t to y for the two- dimensional c a s e and differentiation with r e s p e c t t o r for the axi-s y m m e t r i c case.
This s y s t e m c o n s i s t s of t h r e e linear disturbance equations i n the t h r e e dependent perturbation amplitudes fa
6
and n, where the m e a n velocity w is determined f r o m the m e a n o r steady- state equations. Thes y s t e m i s of the fourth o r d e r i n the dependent variables.
If the m e a n velocity profile i s s y m m e t r i c a l [wl(0) = 01
,
then the disturbance amplitudes can be decomposed into even and odd p a r t s , each p a r t satisfying Eqs. (2. 12)-
( 2 . 14). The even p a r t of thelongitudinal velocity disturbance, f a c o r r e s p o n d s t o anti- s y m m e t r i c a l ( o r sinuous) oscillations and the odd p a r t to s y m m e t r i c a l ( o r v a r i c o s e )
oscillations. I
The anti- s y m m e t r i c a l oscillations a r e analogous to two p a r a l l e l rows of equally spaced v o r t i c e s i n a l t e r n a t e positions (Sketch 2. 2. a)
( d r m i n vortex s t r e e t ) and the s y m m e t r i c a l oscillations to s y m m e t r i c a l l y placed vortices (Sketch 2. 2. b). ~ a r r n ~ n ~ ~ I h a s shown that the s y m m e t r i c a l
(a) Flow i n Alternate Vortex Street ( b ) Flow i n Symmetrical Vortex Street
Sketch 2. 2
vortex s t r e e t i s unstable f o r a l l values of the spacing r a t i o ( r a t i o of t r a n s v e r s e t o longitudinal dimension) a n d will tend to r e a r r a n g e itself
/
into the a l t e r n a t e vortex s t r e e t . The a l t e r n a t e position i s stable for only one spacing r a t i o and unstable f o r all others. Physically, the a n t i -
symmetrical disturbances a r e observed m o r e often than the symmetrical ones; for example, i n the wake behind a c i r c u l a r cylinder a t very low Reynolds numbers. This fact suggests that the anti- symmetrical
disturbance i s m o r e unstable than the symmetrical one and the minimum c r i t i c a l Reynolds nurnber, below which a l l disturbances a r e damped, will be lower for the f o r m e r .
Using Eqs. (2. 12) and (2. 13) the boundary conditions a t the a x i s a r e :
Anti- symmetrical oscillations (Sketch 2. 3. a)
Symmetrical oscillations (Sketch 2. 3, b)
--
r9
"t,c' L*
(a) Anti- symmetric oscillations (b) Symmetrical oscillations Sketch 2 . 3
The quantities f and n can be eliminated f r o m Eqs. (2. 12)
-
(2.14) and a fourth o r d e r equation in
b
can be foundThe boundary conditions for l a r g e values of y a r e obtained f r o m Eq. (2. 17). As y
-
m,
w ---+ 0 exponentially and Eq. (2. 17) becomeswhere
P2 =
a 2-
iaRc.
The solutions of Eq. (2. 18) a r e :
Now $D m u s t be bounded for l a r g e values of y (y
>
0). If the r e a l p a r t of f3 i s positive then the solutions with the positive exponent m u s t berejected and the outer boundary conditions f o r Eq. (2. 17) a r e
and f r o m Eqs. (2. 12) and (2. 13):
11. 2. b. Inviscid, Compressible P r o b l e m
If the solution of the disturbance equations is a s s u m e d to be of the f o r m
and the limit
a R
---p oo i s taken, the resulting equations for the zeroth approximation, q(O! a r e called the inviscid s m a l l disturbance equations. They a r e identical with the equations obtained by ignoring viscosity and t h e r m a l conductivity. The dimensionless inviscid3 2 . equations a r e
.
Continuity 1
9' +
i f = ( T t / T )Q) -
i (w-c) ( s / p ) x- Momentump [ i (w-c) f t w'
$1 = -
( ~ B / % M 2 )2
MomentumEnergy
State
-
This is a system of five equations in the five disturbance variables f,
$,
a, s, and 8, where the mean flow quantities p, T, and w a r e determined f r o m the mean equations of motion. Upon eliminating four out of the five dependent variables, the system is seen to be of the second order.Again, as i n the incompressible case, if the mean velocity and
t e m p e r a t u r e profiles a r e symmetrical, then the disturbance amplitudes can be decomposed into even and odd parts, each p a r t satisfying
Eqs. ( 2 . 2 3 )
-
( 2 . 27). The boundary conditions a t the a x i s a r e : Anti- s y m m e t r i c a l oscillationsSvmmetrical oscillations
The
outer boundary conditions will be derived i n Section IV. 1 but will be included h e r e f o r completeness. It i swhere
11.
-
3. Axi-Symmetric F r e e Shear Flows. Small Disturbance Equations and Boundary ConditionsThe s a m e assumptions regarding the derivation of the axi-
s y m m e t r i c s m a l l disturbance equations apply a s f o r the two-dimensional c a s e and will not be repeated here.
*
In this equation, n equals 3. 141.. . . .
11.3. a. Viscous, Incompressible Problem
/
3 6 . The dimensionless s m a l l disturbance equations a r e
.
Continuity
x- Momentum
r - Momentum
6-
MomentumThis i s a s y s t e m of four l i n e a r equations i n four dependent variables qr
,
q g,
qX and a. This s y s t e m is of the sixth o r d e r i n the dependent variables since q @ and q b ' c a n be eliminated algebraically f r o mEqs. (2. 321, (2. 3 4 ) and (2. 35).
F o r the a x i - s y m m e t r i c wake, the boundary conditions on the axis a r e kinematic in n a t u r e (do not depend upon viscosity) and can be derived f r o m the inviscid equations (Appendix A). All the disturbance amplitudes and the vorticity disturbance m u s t be finite on the axis.
The boundary conditions a r e :
n = O q r = o q @ = o q n a r e a r b i t r a r y
x
'
n f
(I q x " o 3 w = O=
l q@ =-
aqr q s i s a r b i t r a r y n > l q q J = o,
q r = O.
The n = 0 and n = 1 m o d e s a r e shown i n F i g u r e 1.
F a r f r o m the axis, the boundary conditions should be the s a m e a s for the two-dimensional c a s e (Section V. 2). These conditions can a l s o be derived by taking the l i m i t of Eqs. (2. 32)
-
(2. 35) a s r---.
co.
The outer boundary conditions a r e :
where the r e a l p a r t of
p
is positive and1x1. 3 . b. Inviscid, Compressible P r o b l e m
The dimensionless inviscid s m a l l disturbance equations a r e : Continuitv
x- Momentum
r - Momentum
6-
Momentumi p (w-c) q+ =
-
(ins/ar % M 2 )Energy
State
This i s a system of six linear equations i n six dependent variables qr s q X 9 q @ , s, 8
,
and s.
As i n the two-dimensional inviscid, compressible case, the system i s of the second order.The boundary conditions on the a x i s do not depend on the vis- cosity o r compressibility of the fluid and a r e Appendix A
I
:n = 0
,
s , 8 , s a r e a r b i t r a r yn # o (2.44)
n = 1 9$
' -
a 9, q is a r b i t r a r y rThe outer boundary condition i s the s a m e a s in the two- dimensional inviscid, compressible c a s e and will be derived i n
Section V. 2 . It is:
where
2 2
Q = ~ - M c - n < a r g
fi
< nII. 4. Mean Flow Model
The wake i n back of a blunt o r slender body can be divided into two regions of i n t e r e s t ; the "near" wake and the l f f a r " wake. The "near"
wake r e f e r s to the region n e a r the body and the ''far" wake to the region f a r downstream of the body. At low Mach n u m b e r s (Ma
< <
l), the n e a r wake i s c h a r a c t e r i z e d by the formation of v o r t i c e s and unsteady phenomena over a wide range of Reynolds numbers4' [sketch 2.4. a].
( a ) M m
<
C 1 (b) M> >
1I a
Sketch 2. 4
F o r Mm
<
i 1 the boundary l a y e r assumptions do not give a n a c c u r a t e description a t the n e a r wake because Re is low and the gradient in theX*
s t r e a m w i s e direction (a/ax) is not s m a l l compared to the n o r m a l gradient
(a/ay).
The flow is unstable a t low Reynolds nurnbers and becomes turbulent. Thus, t h e r e i s no region of laminar flow when the Reynolds number becomes large. F o r flat plates a t low Mach numbers, the v o r t i c e s and unsteadiness "disappear" and the wake becomeslaminar a t low Reynolds numbers49. The boundary layer assumptions apply i n t h i s c a s e and a n analytical solution can be found for the far wake 50 *
F o r M
>
1, o r m o r e specifically, M> >
1,
the near wake i s00 00
c h a r a c t e r i z e d by two f r e e s h e a r l a y e r s ( o r a n annulus) shed f r o m the body surface, that converge into a "necktt5'. F o r blunt bodies the Mach number external t o the s h e a r layer is "frozen" a t about three, while for slender bodies i t is of the o r d e r of the f r e e s t r e a m Mach number.
2 3
Theoretical and experimental 52' 53 studies show that a laminar shear l a y e r is r e m a r k a b l y stable for super sonic external Mach numbers.
T h i s s a m e r e s u l t applies i n the neck region. Therefore, all o r p a r t of the "inner" wake will be laminar over a wide range of Reynolds
n u m b e r s 1 , 5 4
.
The boundary layer approximations a r e valid i n the inner wake region, except v e r y n e a r the neck where the m e a n profileschange v e r y rapidly. In the far field, Oseen-type approximations can be used to l i n e a r i z e the equations and relatively simple analytical expressions can be obtained. Since the m a j o r t r e n d s of the stability problem a r e of i n t e r e s t here, these analytical expressions will be used i n the stability analysis.
5 5
Kubota' s solution for the two-dimensional compressible wake with z e r o p r e s s u r e gradient will be used (Appendix B):
where
In Eq. (2. 11) the c h a r a c t e r i s t i c length
is related to the half-width of the wake, and the characteristic velocity
i s the maximum velocity defect i n the wake. The quantity d* i s a c h a r a c t e r i s t i c body dimension. The Reynolds number of the wake
i s constant. The y coordinate in the stability equations m u s t be
"stretched" by the temperature (Eq. ( 2 . 47) ). In the derivation of
Eq. (2.46), i t i s assumed that
A T
and M a r e v e r y small ( A T , M< <
1).However, since the relative effects (and not the absolute effects) of temperature and Mach number a r e desired, values of AT and M g r e a t e r than unity will be used in the numerical calculations.
F o r the two-dimensional incompressible wake
A transformation analogous to Kubotals d o e s not exist f o r the axi- s y m m e t r i c c o m p r e s s i b l e wake56. The viscous s t r e s s t e r m i n the momentum equation does not t r a n s f o r m to the equivalent incompressible f o r m and t h e r e f o r e the momentum and energy equations m u s t be
integrated simultaneously. F o r the axi- s y m m e t r i c incompressible wake 50.
.
The momentum thickness
1
or d r a g coefficient,
-
DragC~ - ]
i s given byf
P e * ~ e * 2 ~ d *CO
@a2
= CD ( ~ r d * ~ / 2 ) = 2 r1 [
1-
(U*/ue*)] r* dr*0
and
/T
The Reynalds number for axi- s y m m e t r i c flow i s
-.
and v a r i e s a s the r e c i p r o c a l of the square root of the distance downstream of the origin.
The m e a n profiles f o r the axi- s y m m e t r i c compressible c a s e a r e a s s u m e d to be Gaussian i n the Dorodnitsyn-Howarth variable,
T ,
T = 1 + A T e
3'
where
7.
These m e a n profiles, although not s t r i c t l y valid (as mentioned p r e - viously),will be used to i l l u s t r a t e the stability c h a r a c t e r i s t i c s of
slowly varying axi- s y m m e t r i c velocity and t e m p e r a t u r e m e a n profiles.
111. STABILITY O F TWO-DIMENSIONAL INCOMPRESSIBLE WAKE FLOWS
Wakes belong to the c l a s s of two-dimensional q u a s i - p a r a l l e l flows known a s I1free boundary layers"', i. e . , flow f i e l d s i n which solid boundaries a r e not p r e s e n t . Usually flows which belong to t h i s c l a s s of q u a s i - p a r a l l e l flows have one o r m o r e points of inflection i n t h e velocity profile. The p r e s e n c e of a n inflection point indicates that the flow is dynamically unstable in the limiting c a s e of vanishing viscosity,
and that i t would become unstable a t r e l a t i v e l y low Reynolds n u m b e r s . ,
Hence the c l a s s i c a l m e t h o d s of solution f o r l a r g e Reynolds number, o r m o r e p r e c i s e l y l a r g e aR, cannot be used to d e t e r m i n e the minimum c r i t i c a l Reynolds number. The quantity, I / ( ~ R )
,
i s a m e a s u r e of the diffusion d i s t a n c e f o r vorticity during one period. New m e t h o d s of solution f o r the O r r - Sommerfeld equation have to be found f o r s m a l l values of aR. In addition, the asymptotic methods developed by~ o l l r n i e n ~ ~ ' , ~ e i s e n b e r ~ ' ~ and U n Z 2 have to be modified f o r l a r g e values of aR.
Another problem a r i s e s i n that the q u a s i - p a r a l l e l flow assurnp- tions leading to the Or r
-
Sommerf eld equation a r e not valid throughout the e n t i r e flow field, since the t r a n s v e r s e m e a n velocity component i s of the s a m e o r d e r a s the longitudinal m e a n velocity i n c e r t a i n r e g i o n s of the field f o r the s m a l l v a l u e s of Reynolds number of i n t e r e s t i n t h i s problem. Near the t r a i l i n g edge of a flat plate, f o r instance, the t r a n s v e r s e m e a n velocity m u s t be taken into account if a "precise"prediction of the flow stability i s d e s i r e d . However, g e n e r a l quantitative r e s u l t s a r e of i n t e r e s t h e r e , and the q u a s i - p a r a l l e l a s s u m p t i o n s will be
r e t a i n e d f o r all values of
aR.
The s u b t l e t i e s of i n c o m p r e s s i b l e wake-type flows will be d i s c u s s e d i n t h i s section. Section 111. 1 will d e a l with the solutions of the O r r - S o m m e r f e l d equation f o r s m a l l values of
aR.
A m i n i m u m c r i t i c a l Reynolds number of 39, based upon the length of a f l a t plate, i s found f o r a n t i - s y m m e t r i c a l disturbances. The stability c h a r a c t e r - i s t i c s of a smoothly varying profile a t long wave lengths c a n be found by using discontinuous velocity p r o f i l e s [section 111. 23 In Section 111. 3 the inviscid stability of a n i n c o m p r e s s i b l e G a u s s i a n f l a t plate wake i s d e t e r m i n e d by n u m e r i c a l methods. T h e s e t h e o r e t i c a l r e s u l t s a g r e e v e r y well with the e x p e r i m e n t a l r e s u l t s of Sato and Kuriki 2.
The effect of viscosity on the eigen-value equation f o r l a r g e , but finite,a R
flows c o m e s i n through the "inviscid solutions'' a n d d o e s not depend on the"viscous solutionstf. However, the calculation of t h e d i s t u r b a n c e a m - plitudes m u s t include the "viscous solutions", since the "inviscid solutions" a r e singular a t w
=
c. [section 111.41.
The O r r - S o m m e r f e l d equation c a n be d e r i v e d by eliminating the p r e s s u r e a n d longitudinal velocity perturbation a m p l i t u d e s f r o m Eqs.
( 2 . 12)
-
(2. 14) ( o r by considering the d i s t u r b a n c e vorticity equation),The boundary conditions a t the axis a r e Eq. (2. 29) and Eq. (2. 3 0 )
I
:= 0 anti- s y m m e t r i c oscillations
(3.3) 0 s y m m e t r i c oscillations
.
The boundary condition f o r l a r g e values of y is obtained f r o m Eq. (3. 2).
A s y
-
ca,
w -40 exponentially, and Eq. (3. 2) becomes The solutions of Eq. (3. 4) a r eNow $?I m u s t be bounded f o r l a r g e values of y ( y
>
0 ) ; if f o r definiteness, we takesolutions with the positive exponent m u s t be rejected, and
Eq. (3. 1) together with the boundary conditions Eq. (3. 3) and Eq. (3. 7) constitute a c h a r a c t e r i s t i c - v a l u e problem. The c h a r a c t e r i s t i c - values ( o r eigen-values) a r e d e t e r m i n e d by the usual s e c u l a r determinant, leading to a r e l a t i o n of the f o r m
E ( a , c , R )
=
0 ( 3 . 8 )w h e r e E i s s o m e g e n e r a l function of the argurnents.
If
the i m a g i n a r y p a r t of c is positive, the d i s t u r b a n c e s willamplify with t i m e and the motion is s a i d t o be unstable. If i t i s negative, the d i s t u r b a n c e s will eventually be damped out. If c is z e r o , the d i s -
I
t u r b a n c e s a r e c o n s i d e r e d to be neutral. At e a c h Reynolds number, the s p a t i a l amplification r a t e s c a n be computed. A typical n e u t r a l stability c u r v e f o r wake type flows i s sketched i n F i g u r e 2. The h i s t o r y of a d i s t u r b a n c e a s i t p r o g r e s s e s d o w n s t r e a m is indicated i n t h i s figure.
F o r G a u s s i a n wake profiles, the Reynolds number is constant
[Eq. (2. 501) a n d the wave number is proportional t o [Eq. (2.50)]
.
The d i s t u r b a n c e will be amplified within the n e u t r a l stability c u r v e and damped outside of it.
111. 1. Solutions f o r S m a l l a R
Wake-type flows a r e v e r y unstable because of the o c c u r r e n c e of a point of inflection i n the m e a n velocity profile, i. e . , the m i n i m u m c r i t i c a l Reynolds number, below which all d i s t u r b a n c e s a r e stable, i s r e l a t i v e l y low. The effect of v i s c o s i t y is not confined to a thin l a y e r , a s in the boundary l a y e r c a s e , but i s felt throughout t h e e n t i r e flow field.
M o r e precisely, the r e l a t i v e d i s t a n c e that t h e d i s t u r b a n c e i s diffused i n one period is proportional t o the r e c i p r o c a l of s o m e power of a R, and f o r s m a l l v a l u e s of a R, t h i s distance is of the o r d e r of one, o r the full extent of the wake.
An e n e r g y balance shows that the r a t e of i n c r e a s e of the kinetic energy of t h e d i s t u r b a n c e is equal t o the conversion of e n e r g y f r o m the basic flow into the d i s t u r b a n c e by the Reynolds s h e a r s t r e s s , m i n u s the viscous dissipation 22
.
Viscous dissipation is a l w a y s a stabilizingf a c t o r (always l e a d s t o a d e c r e a s e i n energy) and f o r s m a l l Reynolds n u m b e r s i s v e r y l a r g e . D i s t u r b a n c e s will be damped out v e r y rapidly i n t h i s region.
Tatsurni and Kakutani 31
,
anticipating a s m a l l value of t h e m i n i m u m c r i t i c a l Reynolds number, have expanded the solution i n p o w e r s of a R a s follows:L
YI=o
w h e r e
Substituting Eq. (3. 9) into the O r r - S o m m e r f e l d equation, Eq. (3. 2), and matching p o w e r s of i a R, the following equations relating the $(n)'s a r e obtained:
w h e r e
The solutions of Eq. (3. 10a) a r e :
The solutions of Eq. (3. l o b ) c a n be found by the method of variation of
C.
I *
The g e n e r a l solution of Eq. (3. 2) iswhere C
,
C2,
C3,
and C a r e a r b i t r a r y constants. 4F r o m the outer boundary conditions,Eq. (3. 7),C2 = C 4 = 0 , and f o r a non-trivial solution,
6
and@
m u s t satisfy the following1 3 '
c h a r a c t e r i s t i c equations [ E ~ . (3. 3)
1
:Anti- s y m m e t r i c Oscillations
Symmetric Oscillations
Eq. (3. 9 ) i s substituted into Eqs. (3. 13) and (3. 14) and a corn- plex eigenvalue equation is obtained a s a power s e r i e s i n (i a R).
**
*
The solution Eq. (3. 9) converges uniformly f o r the GaussianL 1
wake profile, w =
-
e - Y,
w h e n a<
1 a n d R I P C 1 Reference 31, page 270, whereI <
constant1 u f ( ~ ) I <
constant**
T a t s u m i and 3Sakutani5' considered only anti- s y m m e t r i c a l- -
oscillations. Their method is e a s i l y extended t o the c a s e of s y m m e t r i c a l oscillations (Appendix C).
T e r m s of the f o u r t h o r d e r and higher w e r e neglected, because i t w a s found that t h e s e t e r m s did not affect the f i r s t t e r m of the asymptotic behavior of the lower b r a n c h of the n e u t r a l stability curve. Appendix C,
[
Eq. (C. 18) a n d Eq. (C. 19)]
.
The eigenvalue equation w a s f u r t h e rJ
simplified by neglecting t e r m s of o r d e r
c3
and higher w h e r eand
Since the a n t i - s y m m e t r i c oscillations a r e m o r e unstable than the s y m m e t r i c ones, the m i n i m u m c r i t i c a l Reynolds number w a s
r 3
d e t e r m i n e d only f o r t h e f o r m e r c a s e . Eq. (C. 18) Appendix C w a s solved graphically f o r the n e u t r a l c u r v e
I
The r e s u l t s a r eJ
indicated i n the following table f o r a Gaussian profile:
Table 3. 1 - TI --.--
+O. 40 SO. 45 4-0.50
The m i n i m u m c r i t i c a l Reynolds number is R = 4. 7 a t 'a = 0. 17. At t h i s point, a R = 0. 8 which is exactly the s a m e r e s u l t found by T a t s u m i and KakutaniS1 f o r a jet.
R
. 0 7 7 - 0 9 5
.
125a .-
.
120.
16. 1 9
R
4. 8 4. 7 4. 7
c R
-.
022-.
034-.
046F o r a n i n c o m p r e s s i b l e f l a t plate (length d ) G a u s s i a n wake
T h e r e f o r e the m i n i m u m c r i t i c a l Reynolds number based upon the length of the plate is Red+ = 39. T h i s value i s considerably below the experi- m e n t a l value of about 600 m e a s u r e d by Hollingdale 5
,
and about 700 m e a s u r e d by ~ a n e d a ~ ~ , f o r which oscillations w e r e observed. At low Reynolds n u m b e r s the amplification r a t e i s a strong function of Reynolds number, i. e . , v i s c o u s dissipation t e n d s t o d a m p out the d i s t u r -bances. T h e r e f o r e , oscillations will begin to occur f a r d o w n s t r e a m of the plate and the Reynolds number a t which they a r e f i r s t o b s e r v e d will be considerably higher than the m i n i m u m c r i t i c a l Reynolds number d e t e r m i n e d by stability.
The a s y m p t o t i c behavior of the lower branch w a s found by taking the l i m i t of Eq. (C. 18) a s
$ ---
0,
and Eq. (C. 19) a srI
---+-
rn.
Any other l i m i t did not produce a n y meaningful r e s u l t s . The f i r s t approximation consisted of solving the eigenvalue equations using only t e r m s up to o r d e r a R; the second approximation, using only t e r m s up to o r d e r ( a R) 2 ; etc. The r e s u l t s a r e tabulated below:
Anti- s y m m e t r i c Disturbance
-
Approximation
/
1 s t 2nd
I
3rd 1a R 2 cR/a 2
3. 00 1. 48
-
1. 51
-1.50
, -1.64 -1.57 36.40
22.96
1. 29 1. 2 2
Svmmetric Disturbances
Table 3, 2
F o r the anti- s y m m e t r i c disturbances, the fourth approximation in-
volves t e r m s of o r d e r ( a ~ ) 4
,
which do not modify the t h i r d o r d e r approxi- mation. F o r the s y m m e t r i c disturbances, the r e a l part of the eigen-2 2
value equation h a s t e r m s of o r d e r a R to the approximations m a d e c R
- 1 . 0 5 - 1 . 2 6 Approximation
1 s t
- 2nd
(even though the imaginary p a r t h a s t e r m s of o r d e r
(aR)
51,
and i t i s--
a R 0. 67 0. 47
not obvious that the 3 r d approximation will not effect the 2nd approximation.
In the l i m i t a
-
0, R -
co,
for anti- s y m m e t r i c disturbances, the productaR
2 approaches a constant along two branches. The lowerSketch 3. 1
of the two branches, in a r e f e r e n c e system fixed i n a fluid a t r e s t ,
corresponds to a wave travelling in the s a m e sense a s the centerline velocity while the upper branch corresponds to a wave travelling i n the sense opposite to this velocity. The flow i s unstable if
2 2
-
1. 57 a<
c<
1. 22 a and stable outside t h i s region.R
Along t h e s e two branches a ~ - F a s a -4 0
,
R-
m,
thus confirming the validity of the expansion [ E ~ . ( 3 . 9 j
.
At thec r i t i c a l point a R = 0. 8; t h i s value i s higher so that m o r e t e r m s i n the eigenvalue equation should probably be retained i n o r d e r to find a m o r e p r e c i s e value of the minimum c r i t i c a l Reynolds number. However, the purpose of the computation w a s to find a n approximate value of the min- i m u m c r i t i c a l Reynolds number, and i t was felt that any additional cal- culations w e r e not commensurate with the a i m s of the investigation.
Moreover, a t these low Reynolds numbers, the boundary l a y e r equations themselves a r e not accurate.
In reviewing the paper of Tatsumi and ~ a k u t a n i ~ l, t h i s author found that t e r m s of o r d e r ( a R) 3
,
which w e r e neglected i n Eq. (6.4) of their paper, should have been retained. In the limiting c a s e a ---+ 0, R ---s. co,
they a r e of the s a m e o r d e r a s the t e r m s that w e r e retained.In addition, the second branch was not recognized by these authors.
Drazin 14 reported these additional r e s u l t s i n 1961.
F o r the s y m m e t r i c oscillations, a R and c approach a constant R
a s a
-
0, R -
w.
The value of lcRl,
however, i s l a r g e r than the maximum value of[wl .
This behavior of the lower branch w a s f i r s tsuggested by c u r l e l ' and verified numerically by Clenshaw and Elliot 7 for the Bickley jet, w = s e c h y. L
The r e s u l t s obtained by the method of Tatsumi and Kakutani 31
suggest another type of expansion f o r s m a l l values of a R. Along the lower branch, R v a r i e s f r o m i t s minimum value to infinity while a i s always v e r y s m a l l (a
< .
2). Another type of expansion, then, would be to hold R fixed and expand c and0
a s a power s e r i e s i n a. Howard 18 used this method f o r the Bickley jet and obtained r e s u l t s that a r eidentical to those of Tatsumi and Kakutani
I
if the additional t e r m s a r e added to Eq. (6.4),
Reference 311
This author h a s used t h i s method for the Gaussian flat plate wake and a l s o obtained the s a m e minimum c r i t i c a l Reynolds number, and the s a m e asymptotic behavior of the two lower branches (anti- s y m m e t r i c oscillations) a s by the method of Tatsumi and Kakutani.III. 2. Idealized P r o f i l e s a t Long Wave Lengths
In b o t h t h e m e t h o d s of T a t s u m i a n d ~ a k u t a n i ~ l a n d H o w a r d 18
,
the m e a n velocity profile a p p e a r s only i n integrals, indicating that the p r e c i s e shape of the velocity profile i s unimportant f o r s m a l l wave- numbers o r l a r g e wave-lengths of the disturbances, Drazin and Howard 15 and ~ r a z i n ' ~ have used discontinuous velocity profiles to find the stability c h a r a c t e r i s t i c s of flows a t low wave-numbers. Some of t h e i r i d e a s will be presented h e r e a s applied to the wake-type flow stability problem.
F r o m Eq. (2. l l ) ,
R = ( v * L * / ~ * ) a
=
a* L*,
where the s t a r r e d quantities r e p r e s e n t dimensional quantities, and
w(y) 4 0 a s y
-
k-
00.
The eigenvalue equation, Eq. (3. 8), leads t o a relation between c, a',
and a R o r (R/a), i. e.,c = c (a, R/Q ) (3. 19)
F o r a fixed a*
,
a-
0,
( ~ / a ) = W/(a*3*) = constant a s L* ---o 0.
Therefore, c-
c (0, R/a )=
f(~*/a*a*)
andA s L*
-
0 f o r a fixed dimensionless velocity profile, w(y),>
then w*(y*) ---A 0 since y = ( y * / ~ * ) --p. oo (y*
<
0 ).
T h e r e f o r e the two limits, a*
--+
0,
L* and (R/a)=
(v*/a*a*) fixed, a n d L*-
0,
a* and( v * / ~
*) fixed, give the s a m e r e s u l t a --A 0,
R/U fixed. In other words, f o r w(y) fixed, each l i m i t gives the s a m e limiting f o r m of the eigenvalue relation, Eq. ( 3 . 20), and thestability c h a r a c t e r i s t i c s of the flow a r e the s a m e f o r both the limiting profile w*(y*) a s a*
-
0.
In other words, by using the limiting profiles w*(yJF) a sL* -
0,
which m a y be discontinuous, the stability c h a r a c t e r - i s t i c s of smoothly varying profiles actual w * ( ~ * ) ] a s a*-
0 c a n be determined.[
~ r a z i n l * d e r i v e s jump conditions f o r
[
a, oR bounded1
a t the points where w and/or (dw/dy) a r e discontinuous and applies them t o the c a s e of a broken line jet.He finds that t h r e e n e u t r a l " b r a n ~ h e s ' ~ exist for a n t i - s y m m e t r i c disturbances
a R 4 co
,
a fixed ( a ) a ,,p-tanh " ( 7-
4 5 )2 2
a R -constant a
-
0 (b) a R-
1 . 3 4 c R - - 4 - 1 . 54 a 2(3. 21)*
R-co
( c ) a R 2
-
32.4 cR 4 1.21 a 2and f o r syrnmetric disturbances
2 2
a R fixed, a ---r 0
,
c --r-
[ l+
2n (n / a ~ ) ]The branches (b) a n d ( c ) of Eq. (3. 21) correspond t o those found in Section 111. 1 and have approximately the s a m e limiting values. The wave speeds a g r e e t o within 2 O/o, while the product a R a g r e e s t o 2 within 10 O/o on the lower branch (b) and to within 50 O/o on the upper branch (c). Thus the u s e of discontinuous velocity profiles a t low wave n u m b e r s is justified a t l e a s t f o r qualitative purposes.
The f i r s t branch ( a ) i s not meaningful in that the assumption a R bounded i s not met. T h i s branch was not found i n Section 111. 1.
However, t h i s calculation indicates that the flow is unstable between the branches (b) and ( c ) and above branch (a). In Section III. 3, another
limit i s found, namely the inviscid limit, on which a R
-
oo,
a ---edS .
Below t h i s branch, the flow is unstable, which l e a d s t o the hypothesis that t h e r e might be a n "island of stability" within the unstable region 14 This situation i s i l l u s t r a t e d i n the following sketch. It was suggested
9 In t h i s text, c i s defined a s the relative wave velocity and i s R
the negative of the values found i n Reference 14.
Sketch 3. 2
14 P
by Stuart that branch (a) corresponds to the limit
aR -
constant,F o r symmetric disturbances, a R + constant, c
R -
constanta s R --L
,
o-
014. This behavior is exactly that found in Section 111. 1 f o r the smoothly varying profile. F o r n=
1, cR- -
1. 3 3 andTherefore, for low wave numbers o r long wave lengths, the stability c h a r a c t e r i s t i c s of smoothly varying velocity profiles can be found by using suitable discontinuous velocity profiles.
111. 3. Invi scid Limit
A point of inflection in the mean velocity profile indicates that the flow i s unstable i n the limiting c a s e of vanishing &scosity, and that the main f e a t u r e s of the instability mechanism can be obtained by neglecting the viscous forces2'. The effect of viscosity i s a stabilizing
one and can be taken into account once the inviscid instability mechanism i s understood.
The inviscid l i m i t is formally obtained by expanding
56
in apower s e r i e s i n (l/aR),
substituting this expansion into Eq. (3. I), and taking the limit a R 4 0 0
.
The equation for the zeroeth-order approximation,
,
i s known a s the inviscid Orr-Sommerfeld equation and isF o r neutral disturbances ( c
-
0) the wave speed,I
-
C~'
i s equal to them e a n velocity, w
,
a t the point where w" vanishes. The vanishing of w"is a n e c e s s a r y and sufficient condition for the existence of a n e u t r a l distur banceZ2
,
and the disturbance amplitudes a r e finite everywhere i n the flow region if t h i s condition holds. The inviscid neutral wave speed and wave number a r e commonly denoted by c and a s.
2
F o r a Gaussian flat plate wake, w =
-
e-' ,
the inviscid neutral1
wave speed i s c s =
-
e m Z = -0.606. ~ c ~ o e n ' ~ approximated thevelocity profile in t h r e e different regions and found that a
=
2. 0. In theS
p r e s e n t investigation, a was determined by numerically integrating
S
Eq. (3. 23) on a n IBM 7090 electronic computer. This method i s described i n Section IV. 2; a s was fo